ArticlePDF Available

Cannabis Domestication, Breeding History, Present-day Genetic Diversity, and Future Prospects

November 2016
Critical Reviews in Plant Sciences 35(5-6):293-327

DOI:10.1080/07352689.2016.1267498

Authors:

Robert Connell Clarke

International Hemp Association

Mark D Merlin

University of Hawai'i System

Humans and the Cannabis plant share an intimate history spanning millennia. Humans spread Cannabis from its Eurasian homelands throughout much of the world, and, in concert with local climatic and human cultural parameters, created traditional landrace varieties (cultivars resulting from a combination of natural and farmer selection) with few apparent signs of domestication. Cannabis breeders combined populations from widely divergent geographical regions and gene pools to develop economically valuable fiber, seed, and drug cultivars, and several approaches were used with varying results. The widespread use of single plant selections in cultivar breeding, inbreeding, and the adoption of asexual reproduction for commercial drug production, reduced genetic diversity and made many present-day cultivars susceptible to pathogens and pests. The great majority of drug Cannabis cultivars are now completely domesticated, and thus are entirely dependent on humans for their survival. Future ramifications remain to be realized.

Four groups of naturally occurring Cannabis landraces are extant today, along with sinsemilla drug cultivars and industrial hemp cultivars (the latter are not shown). Presented here are proposed taxon names, acronyms representing their biotype groupings, region of origin, routes of diffusion, and primary uses.

…

The various Cannabis taxa (indicated by their acronyms, see Figure 1) have been spread by humans into nearly every part of the world; PA? = putative Cannabis ancestor (from Clarke and Merlin, 2013).

…

Humans have found uses for nearly every part of the Cannabis plant and a number of traditional cultures harvested fibers, seed, and drugs from the same crop. Modern breeders developed special purpose cultivars by selecting certain plant parts (e.g., stalks, seeds, and flowers) for increased yield along with other economically valuable traits (from Clarke and Merlin, 2013).

…

Both qualitative as well as quantitative traits have been strongly selected in BLD X NLD sinsemilla hybrids (from Clarke and Merlin 2013).

…

Cannabis NLD and BLD plants vary in appearance throughout their growth cycles. In addition to their narrower leaflets, juvenile NLD plants are taller and more laxly branched than the shorter, more compact Afghan BLD plants (from Clarke and Merlin, 2013).

…

Figures - uploaded by Robert Connell Clarke

Content may be subject to copyright.

Content uploaded by Robert Connell Clarke

Content may be subject to copyright.

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=bpts20

Download by: [210.48.190.43] Date: 30 March 2017, At: 16:29

Critical Reviews in Plant Sciences

ISSN: 0735-2689 (Print) 1549-7836 (Online) Journal homepage: http://www.tandfonline.com/loi/bpts20

Cannabis Domestication, Breeding History,

Present-day Genetic Diversity, and Future

Prospects

Robert C. Clarke & Mark D. Merlin

To cite this article: Robert C. Clarke & Mark D. Merlin (2016) Cannabis Domestication, Breeding

History, Present-day Genetic Diversity, and Future Prospects, Critical Reviews in Plant Sciences,

35:5-6, 293-327

To link to this article: http://dx.doi.org/10.1080/07352689.2016.1267498

Published online: 02 Mar 2017.

Submit your article to this journal

Article views: 222

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Cannabis Domestication, Breeding History, Present-day Genetic Diversity,

and Future Prospects

Robert C. Clarke

and Mark D. Merlin

BioAgronomics Group Consultants, Los Angeles, California, USA;

Botany Department, University of Hawai’i at Manoa, Honolulu, Hawaii, USA

ABSTRACT

Humans and the Cannabis plant share an intimate history spanning millennia. Humans spread

Cannabis from its Eurasian homelands throughout much of the world, and, in concert with local

climatic and human cultural parameters, created traditional landrace varieties (cultivars resulting

from a combination of natural and farmer selection) with few apparent signs of domestication.

Cannabis breeders combined populations from widely divergent geographical regions and gene

pools to develop economically valuable ﬁber, seed, and drug cultivars, and several approaches were

used with varying results. The widespread use of single plant selections in cultivar breeding,

inbreeding, and the adoption of asexual reproduction for commercial drug production, reduced

genetic diversity and made many present-day cultivars susceptible to pathogens and pests. The

great majority of drug Cannabis cultivars are now completely domesticated, and thus are entirely

dependent on humans for their survival. Future ramiﬁcations remain to be realized.

KEYWORDS

Afghanistan; Africa;

cannabinoids; Europe; ﬁber;

hashish; hemp; India; indica;

marijuana; New World;

ruderalis; sativa; seed oil;

sinsemilla; terpenoids

I. Cannabis botany and ecology

The ecological requirements and genetic inheritance of

plants determine where they grow naturally. Cannabis

plants require well-drained soils, adequate sunlight,

warmth, and moisture, so most naturally growing popu-

lations are found seasonally across accommodating

northern temperate latitudes. Under natural conditions,

Cannabis grows well along exposed riverbanks, lakesides,

the margins of agricultural land, and other areas dis-

turbed by humans (Merlin, 1972; Clarke, 1977,1981;

Clarke and Merlin, 2013; Small 2015). Based on ecologi-

cal constraints, Cannabis evolved somewhere in temper-

ate latitudes of the northern hemisphere, and Eurasia is

favored as its primary region of origin (Clarke and

Merlin, 2013). Seminal uses, early cultivation, worldwide

dissemination, and eventual domestication of Cannabis

all began within this natural biogeographical range.

Cannabis plants are usually dioecious and produce

either male (pollen) ﬂowers or female (seed) ﬂowers. An

individual Cannabis plant’s gender is determined by X

and Y chromosomes, and monoecious plants of this

genus producing ﬂowers of both sexes occur only rarely

in nature. Cannabis also relies on air currents to spread

pollen grains from male plants to female seed plants.

Wind pollination, dioecious sexuality, and X/Y sexual

inheritance are each relatively rare in plants, yet all three

are characteristics of Cannabis.

Cannabis plants grow and develop within their annual

life cycle of sexual reproduction. Moistened seeds germi-

nate as spring weather warms, and juvenile plants grow

rapidly through the summer. Fast-growing juvenile

plants appear much alike, but as autumn day length

decreases, populations begin to ﬂower and plants express

individual phenotypic differences. Male plants within a

single population are often slightly taller than female

plants. Male ﬂowers hang from branches with few leaﬂets

and are exposed to the wind (which facilitates pollen dis-

persal), whereas ﬂowers on female plants are tightly clus-

tered with small leaﬂets to trap male pollen grains that

fertilize the female ovules. Soon after males shed their

pollen they die. Before the arrival of killing frosts, the fer-

tilized female plants ripen viable seeds. These dissemi-

nules fall to the ground via the wind or feeding birds and

other animals that disperse the seeds inadvertently; and

then they overwinter in the soil ready to initiate another

life cycle the following spring. During their fertilized

development, each enlarging seed is surrounded by a

bract covered with thousands of secretory hairs called

glandular trichomes. These hairs produce a resinous

blend of cannabinoids (chemically related compounds

found in Cannabis) and aromatic compounds (mostly

terpene compounds common in many plants); and these

secondary metabolites are believed to protect the devel-

oping seed by repelling pests and pathogens (Clarke,

CONTACT Robert C. Clarke Rob@BioAgronomics.com BioAgronomics Group Consultants, Los Angeles, CA.

CRITICAL REVIEWS IN PLANT SCIENCES

2016, VOL. 35, NOS. 5–6, 293–327

http://dx.doi.org/10.1080/07352689.2016.1267498

1977,1981; McPartland et al., 2000). Psychoactive delta-

9-tetrahydrocannabinol (THC) and nonpsychoactive

cannabidiol (CBD) are the primary cannabinoid constit-

uents in almost all Cannabis. THC and aromatic psycho-

active secretions are evolutionarily signiﬁcant as they

attracted early human attention, and at least in modern

times they remain the primary impetus for the continued

breeding and worldwide dispersal of Cannabis cultivars

(Pollan, 2001; also see Merlin, 1972; Clarke, 1998; Clarke

and Merlin, 2013).

Cannabis favors a mild climate with sufﬁcient water

and sunlight, and early humans spread it into a range of

favorable temperate and sub-tropical niches where it

became naturalized (feral) throughout Eurasia, in parts

of Africa, and more recently in the New World. Canna-

bis thrives on the nutrient-rich dump heaps near human

occupation and has readily apparent agronomic traits,

and it was therefore preadapted to cultivation (Ander-

son, 1967; Merlin, 1972; Clarke and Merlin, 2013).

Annual plants of this genus branch freely when culti-

vated in open areas; when grown in dense stands, they

suppress branching, forming a single central stalk (Iltis,

1983). This naturally adapts Cannabis to high density

sowing for ﬁber production, as well as low density plant-

ings that encourage branching and ﬂower formation for

seed and drug production.

II. Cannabis diversity

Environmental factors, in concert with a plant’s genotype

or set of genes, more or less determine its phenotype, or

the visible expression of its genotype. Genetically and

phenotypically diverse varieties of Cannabis evolved

under the pressures of natural selection within the

diverse environments into which they were introduced,

and were further selected by humans to provide ﬁber,

seed, or drug products. More than two centuries of stud-

ies to characterize Cannabis diversity and bring order to

this multipurpose genus resulted in a series of taxonomic

systems circumscribing one, two, or three species.

Twenty-ﬁrst-century taxonomic research by Karl Hillig

(2004a,b,2005) encompassed a wider geographical range

and diversity of population states than previous efforts,

and set the stage for understanding Cannabis evolution.

Recently proposed taxonomic systems (McPartland

et al., 2000; McPartland and Guy, 2004; Small, 2015)

share many commonalities with Hillig’s conclusions, and

the evolutionary hypotheses of Clarke and Merlin (2013)

are deeply rooted in his work. Cannabis is presently con-

sidered by many, but not all taxonomists, to be a poly-

typic genus consisting of two extant species Cannabis

sativa and Cannabis indica each circumscribing several

biotypes or subspecies (see Small, 2015 for a recent,

detailed discussion of his long-held single species

hypothesis). Taxonomists also recognize three popula-

tion types for Cannabis based on natural origins and

associations with humans; those that are truly wild, those

that are cultivated, and feral escapes from cultivation that

grow spontaneously in areas associated with and often

disturbed by humans (Clarke and Merlin, 2013; Small,

2015). It is important that taxonomic systems reﬂect

both evolutionary history and the relationships between

differing gene pools. Based largely on archaeological and

historical records backed by Hillig’s chemotaxonomic

research, Clarke and Merlin (2013) published names and

acronyms circumscribing geographical and cultural

groupings as an aid to understanding the roles of differ-

ing gene pools in the domestication history of Cannabis

(see Figure 1 for acronym deﬁnitions used throughout

this review.)

Northern temperate origin, annual life cycle, dioe-

cious sexuality, camp-following and weedy tendencies,

and readily observable traits of potential value (large

seeds, strong ﬁbers, and sticky sparkling resin glands)

predisposed Cannabis to utilization, selection, and

breeding by early humans. The following section of this

review focuses on the impact of the human–Cannabis

relationship upon the evolution of Cannabis as a crop

plant (Figures 1 and 2).

III. Early domestication history

The evolutionary history of Cannabis and human inter-

actions span millennia and its ongoing domestication

will continue into the future. Whether early humans ﬁrst

used Cannabis as a source of ﬁber, food, or mind-altering

compounds, they eventually developed differing agricul-

tural techniques to increase the yield and quality of all of

these products, which led to the twentieth-century breed-

ing of Cannabis cultivars speciﬁcally for ﬁber, seed, or

drug production based on local cultural preferences. Phe-

notypic variation between Cannabis populations is well

documented in surveys of the wide range of cultivated

ﬁber, seed, and drug varieties (e.g., Serebriakova, 1940;

Clarke, 1981). Nature and humans have worked together

to control domestication in Cannabis, but at times they

have presented opposing forces that engendered differing

outcomes. For example, isolation, artiﬁcial selection, and

inbreeding imposed by humans during domestication

have limited genetic diversity, whereas natural outcross-

ing and genome mixing have encouraged genetic diver-

sity. The question remains: How did Cannabis evolve

toward domestication during cultivation and breeding,

as opposed to remnant wild populations and feral

escapes that evolved under natural selective pressures

alone? (Figure 3)

294 R. C. CLARKE AND M. D. MERLIN

Wild Cannabis populations readily yielded seed, ﬁber,

and drugs useful to early humans. In addition to being a

multi-use plant, Cannabis is easy to grow, and must have

been included among the ﬁrst plants brought into culti-

vation, almost certainly in the regions where people orig-

inally encountered this versatile resource. As human

populations settled and expanded, they probably

depleted nearby natural stands of Cannabis, and conse-

quently early farmers began to grow plants closer to their

homes. Soon these cultivators began to sow seeds from

plants expressing traits that differed from the norm such

as larger seeds, taller stalks, and/or more resin produc-

tion (e.g., enhanced trichome production, increased tri-

chome size, elevated THC content). As a result of their

artiﬁcial selection, they unwittingly initiated the

processes of domestication. As humans spread Cannabis

beyond its original wild range into recently settled agri-

cultural centers, the selective pressures of new habitats

and human requirements for varying products became

more evolutionarily important. Landraces are not simply

the products of human selections, as natural selective

pressures constantly impose their effects, but cultivation

favors the outcomes of artiﬁcial human selection over

those of natural selection. Farmers and breeders select

plants with useful traits, resulting in different frequencies

of traits (both qualitative and quantitative) between wild

populations and their cultivated and increasingly domes-

ticated descendants. Qualitative morphological changes

(e.g., fruit size, stalk height, ﬂower form and color, pres-

ence or absence of secondary metabolites, etc.) are often

Figure 1. Four groups of naturally occurring Cannabis landraces are extant today, along with sinsemilla drug cultivars and industrial

hemp cultivars (the latter are not shown). Presented here are proposed taxon names, acronyms representing their biotype groupings,

region of origin, routes of diffusion, and primary uses.

CRITICAL REVIEWS IN PLANT SCIENCES 295

Figure 2. The various Cannabis taxa (indicated by their acronyms, see Figure 1) have been spread by humans into nearly every part of

the world; PA? = putative Cannabis ancestor (from Clarke and Merlin, 2013).

Figure 3. Humans have found uses for nearly every part of the Cannabis plant and a number of traditional cultures harvested ﬁbers,

seed, and drugs from the same crop. Modern breeders developed special purpose cultivars by selecting certain plant parts (e.g., stalks,

seeds, and ﬂowers) for increased yield along with other economically valuable traits (from Clarke and Merlin, 2013).

296 R. C. CLARKE AND M. D. MERLIN

controlled by two alleles or gene states operating at a sin-

gle locus or gene position along the chromosome, and

are inherited by simple dominant versus recessive Men-

delian genetic mechanisms. Quantitative physiological

traits (e.g., increased yield, variations in amounts of sec-

ondary metabolites, etc.) are usually controlled by a suite

of alleles at several loci, their inheritance is more com-

plex, and consequently it is more difﬁcult to improve

these traits through phenotypic selection.

Many times throughout history humans most likely

carried only a few seeds to a new settlement, and

often only a limited number of individual plants

(sometimes only one female exhibiting speciﬁcfavor-

able traits) were selected as parents. This produced

genetic bottlenecks that gave rise to populations with

restricted genomes in new areas of Cannabis growth.

When limited numbers of seeds were dispersed into

isolated regions where Cannabis was not already

growing, and the resulting small populations were

brought under new selective pressures, strong founder

effects inﬂuenced the evolution of these new Cannabis

crops and in some cases their feral escapes. Only a

few plants of limited genetic diversity established new

populations which eventually evolved into local

landraces.

There are striking examples of founder effects in drug

Cannabis breeding. Early-maturing Afghan broad-leaﬂet

drug (BLD) landraces were introduced into Europe and

North America in the late 1970s and early 1980s, and a

handful of psychoactively potent and early-maturing

individuals were selected and crossed with the narrow-

leaﬂet drug (NLD) cultivars already growing in the West

to make NLD/BLD hybrids. NLD/BLD hybrid cultivars

were strongly selected for high THC content along with

early maturation and large female inﬂorescences; this

produced potent plants when grown outdoors across

northern temperate latitudes, and thus revolutionized

domestic marijuana production. However, the major

horticultural drawback to NLD/BLD hybrids is their sus-

ceptibility to fungal infections. Their large, dense, tightly

packed female inﬂorescences (commonly called “buds”)

hold moisture and make perfect microclimates for bud

rot caused by the gray mold (Botrytis cinerea Pers.),

known to ﬁne wine makers as the “noble rot”of grapes.

NLD/BLD cultivars have little natural resistance to fun-

gal infection because the BLD founder plants were natu-

rally adapted to the arid conditions of Afghanistan

where pathogenic fungi cause little threat (McPartland

et al., 2000). Many NLD landraces originated from

regions with relatively more humid conditions (e.g.,

Colombia, India, Jamaica, Thailand, etc.) and evolved a

natural resistance to fungal infection. Bud rot was almost

unheard of when only NLD varieties were grown, and

now it accounts for signiﬁcant annual agricultural loses

(Figure 4).

There are no natural sterility barriers between Canna-

bis plants and crosses are generally fully fertile. There-

fore, in order to reinforce and preserve the genetic

integrity of the cultivar’s desirable trait expressions,

plants with variations in favorable economic traits must

ﬁrst be genetically isolated, and then continued selection

and sexual reproduction must be maintained in isolation

to reinforce and preserve those traits. Wind pollination

between separate male and female plants favors hybrid-

ization (recombination of separate gene pools through

sexual reproduction) because male and female gametes

usually must come from different parents, and thus out-

crossing is obligatory. In traditional cropping regimes,

only seeds from favorable female plants were preserved

for sowing, but the seeds on each female plant were fer-

tilized most likely by a wide range of males, thereby

restoring genetic diversity in the offspring population.

Outcrossing favors dominant alleles, leading to rapid

evolution of landraces and cultivars. Natural introgres-

sion (the exchange of genes between populations via a

hybrid intermediate) relies on backcrossing to one (or

both) of the parental populations. Plant breeding

Figure 4. The domestication process starts with individual plants

collected from naturally selected wild populations that are

brought into cultivation, and are then increasingly subjected to

human selection until they reach the cultivar stage. Some culti-

vars may escape human selection pressures, become feral, and

return to natural selection (from Clarke and Merlin, 2013).

CRITICAL REVIEWS IN PLANT SCIENCES 297

transfers genes between populations in the same manner,

and artiﬁcially controlled introgression is the basis for

breeding hybrid cultivars.

When two previously isolated varieties interbreed they

form a hybrid population with more vigorous growth, a

wider variety of genetic combinations than either of the

parents, and unique phenotypes with conferred survival

advantages that drive natural evolution. During the

domestication process, hybrid offspring are artiﬁcially

selected which leads to novel combinations of traits. For

example, broad-leaﬂet hemp (BLH) landraces introduced

from East Asia into Europe and North America were

crossed with European narrow-leaﬂet hemp (NLH) land-

races to create improved industrial hemp cultivars, while

NLD varieties introduced into North America in the

1960s and 1970s were crossed among themselves to cre-

ate improved NLD hybrid drug varieties. However, accel-

erated evolution occurred via artiﬁcial selection in the

1980s when domestic NLD hybrids were crossed with

imported Afghan BLD landraces to create the lineages of

hybrid “sinsemilla”cultivars that are widely grown today

(sin semilla in Spanish literally means “without seed”).

IV. Phenotypic changes during domestication

There are many examples of variability between cultivars

selected (or not) for differing traits that changed signiﬁ-

cantly with domestication. For instance, plants of hemp

ﬁber cultivars (both NLH and BLH) produce longer

internodes and fewer branches, even when grown in

open environments, whereas seed (both NLH and BLH)

and drug varieties (both NLD and BLD) exhibit shorter

internodes and more proliﬁc branching even when

closely sown. Traditional drug landraces produce

5.0–20.0% THC, whereas European hemp cultivars

contain less than 0.3% THC (Small and Marcus, 2003).

Domestication begins with the wild condition of a trait

as the norm. Unidirectional or one-way evolution pro-

ceeds by artiﬁcial selection away from the wild character

state in a continuum of intermediate stages leading to a

domesticated form or function (e.g., from small to large

seeds or from low to high content of stalk ﬁber); how-

ever, we only have sufﬁcient archaeological evidence to

draw ﬁrm conclusions for a few physical stages of

domestication (Fleming and Clarke, 1998). Seeds are the

only propagules of sexually produced landraces. Mor-

phological changes in seed traits accompanying domesti-

cation were largely controlled by unidirectional

evolution of qualitative characteristics. When early

humans collected seeds for food or sowing, they uncon-

sciously selected for nonshattering inﬂorescences. Persis-

tent seeds remain on harvested plants and are taken

home, whereas naturally dispersing seeds fall to the

ground and are lost. Within self-sowing spontaneous

populations there is no selection for nonshattering phe-

notypes. On the other hand, selection for nonshattering

is strong in hemp cultivars because growing ﬁber ﬁelds

requires sowing thousands of seeds, which must be har-

vested the previous year. Traditional Nepalese landraces

are grown to produce drugs, ﬁber, and seed. In the case

of seeds, most are eaten, with some reserved for sowing

(Clarke, 2007), and nonshattering is favored in all Nepal-

ese cultivars.

Landraces, in general, also have larger and lighter col-

ored seeds than wild or feral populations growing nearby;

in addition, the seeds of landraces also generally lack the

horseshoe-shaped base and mosaic-patterned perianth

associated with freely-shattering, camouﬂaged, wild-type

seeds (Vavilov, 1931). The East Asian BLH gene pool

exhibits the greatest diversity in seed coloration and size,

resulting from millennia of natural and human selection

within a large geographically and culturally diverse region.

Chinese wild and feral seeds range in size from 100 to

500/g (approximately 2800 to 14,000/ounce), whereas

large snack food seeds might have as few as 15/g (approx-

imately 420/ounce), representing more than a twenty-fold

range in seed size (Clarke and Merlin, 2013)(Figure 5).

Subtle physiological changes, although less obvious,

are probably more common and possibly also more evo-

lutionarily signiﬁcant than morphological changes. In

fact, increased seed size and loss of seed dehiscence char-

acteristics are among the few morphological clues of sub-

stantial artiﬁcial selection reported; however, changes in

seed physiology also accompany domestication. For

example, mechanisms for perennation (inhibition of

seed germination that allows seeds to survive through

winter before germination occurs in the spring) are often

lost. Cannabis farmers traditionally collect seeds and

store them, ensuring their survival until the following

spring, and, as a result, perennation is no longer of natu-

ral evolutionary advantage. Seeds of naturally selected

feral populations evolved in response to climate ﬂuctua-

tions and germinate slowly and unevenly, whereas culti-

var seeds are adapted to agricultural norms and must

germinate quickly and uniformly. Densely sown hemp

ﬁeld conditions automatically select for uniform and

rapid germination because late-germinating seedlings

cannot compete with already established seedlings and

are crowded out, thus allowing for more uniform treat-

ment of the crop. Cultivars that have lost their dehis-

cence mechanisms, as well as physiological means of

delaying germination, depend upon human intervention

to perpetuate them—a strong sign of domestication.

Cannabis seeds also vary widely in fatty acid content

(M€

olleken and Theimer, 1997) offering physiological

traits for further selective breeding.

298 R. C. CLARKE AND M. D. MERLIN

Plant architecture or gross phenotypic expression has

also been affected by domestication. Both NLH and BLH

are selected for herbaceous unbranched stalks and long

internodes associated with longer and more ﬂexible

ﬁbers, traits expressed by both male and female plants.

Hemp cultivars have also been selected for total ﬁber

yield, and exceptional cultivars (e.g., “Kompolti TC”) can

produce over 40% dry weight of this product (B

ocsa,

1994). Seed-propagated cultivars (NLH, BLH, NLD, and

BLD), which are selected for maximum ﬂower yield, are

more profusely branched with relatively woody stalks

and shorter internodes, and consequently have shorter

and more brittle ﬁbers. Artiﬁcial selection of economic

traits in Cannabis is usually limited to females, which

has had a marked effect on the evolution of the genus

under domestication. However, the selection of female

plants also affects the phenotypes of their male offspring.

Indeed, males are heavily inﬂuenced by the selective

pressures exerted on their female parent, and, during

domestication, changes in male morphology (e.g.,

shorter height, compact inﬂorescences, and more ﬂow-

ers) paralleled changes in female morphology.

The morphology of Cannabis inﬂorescences also

reﬂects extreme selection during domestication. Both

increased seed and resin gland yields rely on increased

female ﬂower production, and both seed and drug culti-

vars have larger inﬂorescences containing more ﬂowers

than ﬁber cultivars, which in turn have larger inﬂores-

cences than the wild and feral types from which they

originated. NLD/BLD hybrid sinsemilla cultivars were

selected for mind-altering potency (e.g., higher THC per-

centage, increased cannabinoid synthesis, more and

larger glandular trichomes) along with immense female

inﬂorescences providing increased surface area for the

Figure 5. Cannabis seeds, stalks, and ﬂowers were selected for a wide number of traits that were unintentionally as well as intentionally

altered during human selection and cultivar development. Quantitative and qualitative changes occur in morphological as well as physi-

ological traits (from Clarke and Merlin, 2013). F DFruit, S DSeed, D DDrug, U DUnintentional, I DIntentional.

CRITICAL REVIEWS IN PLANT SCIENCES 299

elaboration of cannabinoid-producing glandular tri-

chomes. Because sinsemilla cultivars are reproduced

asexually by vegetative cuttings, there is no need to allow

space or divert energy to maturing seeds, and their

inﬂorescences are more densely packed with ﬂowers and

small leaves than NLD and BLD landraces (Clarke and

Merlin, 2013)(Figure 6).

During domestication, traits more frequently diverge

from the wild condition by bidirectional or two-way evo-

lution via disruptive selection—again resulting in a con-

tinuum of phenotypes, but with the median wild

condition lying between two domesticated extremes

(e.g., stalk internode length and branching pattern, varia-

tions in cannabinoid content, and timing of maturation).

Disruptive selection occurs when natural selection of

wild populations favors certain environmentally adaptive

traits, whereas artiﬁcial selection of cultivated crops

favors a certain plant product, and leads to strong evolu-

tionary consequences during the domestication process.

Variation in the timing of ﬂoral maturation offers an

example of natural bi-directional evolution. Long before

the appearance of modern humans, the primordial Eur-

asian Cannabis gene pools were likely divided and

reformed several times during the Pleistocene as ice

sheets advanced and receded across southern Europe

and southeastern Asia. Differing photoperiod responses

naturally evolved through geographical isolation and

selection, and ecotype differences were established.

Extremely early ﬂowering, northern temperate NLH cul-

tivars (e.g., “Finola”) and much later ﬂowering tropical

NLD landraces (e.g., those from Colombia, India, and

Thailand) evolved under natural selection from separate

ancestral genomes that shared ancient roots in central

Eurasia. In other words, these divergent landraces

evolved by following individual “routes”to domestica-

tion at differing latitudes both north and south.

Another example of bidirectional evolution in Canna-

bis results from artiﬁcial selection for THC levels. Early

Eurasians were presented with several non-Cannabis

choices of alternative species producing strong ﬁbers and

nutritious foods, but relatively few other known species

with psychoactive options. Revering our ancestors, creat-

ing deities, and opening cognitive channels of communi-

cation with them predate agriculture. Possibly Cannabis

was ﬁrst appreciated by early humans more for its

entheogenic (e.g., spiritual perception) or other mind-

altering powers, while its uses for ﬁber and food became

more important later as settled populations grew. Only

in modern times has Cannabis been selected for

extremely low THC content. As noted above, the pri-

mary psychoactive constituent of Cannabis is THC, and

its content varies widely between feral as well as domesti-

cated hemp and drug populations. How did such diver-

sity in cannabinoid content evolve? Why are ﬁber

varieties so low in THC, whereas sinsemilla cultivars

produce the most psychoactively potent marijuana the

world has ever known? (Figure 7)

Feral populations, as well as hemp and hashish culti-

vars, that are unselected for cannabinoid content express

a bell curve of CBD:THC ratios ranging from mostly

Figure 6. Both qualitative as well as quantitative traits have been strongly selected in BLD X NLD sinsemilla hybrids (from Clarke and

Merlin 2013).

300 R. C. CLARKE AND M. D. MERLIN

THC with almost no CBD (Type I) through mostly CBD

with almost no THC (Type III). However, the vast

majority of hemp and hashish populations contain

approximately equal amounts of CBD and THC (THC:

CBD 1 or Type II), which, in the absence of human

selection, may be the naturally selected and evolution-

arily neutral character state of major cannabinoid pro-

duction (Clarke and Merlin, 2013). Within intermediate

Type II (average THC:CBD 1) landrace populations,

seeds selected from the lowest THC:CBD ratio plants, as

well as those seeds selected from the highest THC:CBD

ratio plants, when sown in isolated gardens, will produce

offspring populations with median cannabinoid ratios

skewed in favor of either the high-THC (Type I) or high-

CBD (Type III) parents. When these selection processes

are repeated for a few generations, the following two dis-

tinct chemotypes or chemical phenotypes result: one

with a narrower range of cannabinoid ratios and higher

average THC content (Type I); and a second with a nar-

rower range of cannabinoid ratios and higher average

CBD content (Type III). Since psychoactive THC and

terpenes are easily detected by simple human assay, early

Cannabis farmers readily developed more potent landra-

ces by annually selecting higher potency plants and sow-

ing their seeds. Because CBD is not psychoactive, it can

only be detected by laboratory analysis, and its levels

must be increased by controlled selection and breeding.

The recent public availability of analytic laboratories has

revolutionized CBD breeding.

Hemp ﬁber cultivars are bred to produce tall,

unbranched stalks; consequently, ﬂowerandseedproduc-

tion are relatively less important. Hemp landraces evolved

in northern latitudes with a short summer growing season

and long photoperiod that did not naturally favor long

ﬂowering periods and THC production. Natural selection

maintained THC at relatively low levels in both East

Asian BLH and European NLH landraces. Modern Euro-

pean NLH cultivars were artiﬁcially selected for extremely

low levels of total cannabinoids, and especially lower

amounts of THC. Disruptive selection increased THC and

lowered CBD in drug varieties, and lowered THC produc-

tion in ﬁber varieties, resulting in bi-directional evolution

of cannabinoid phenotype over time.

Strong heritability of alleles for THC and CBD pro-

duction (as well as nonproduction) allows for rapid

increases or decreases in potency through selection, and

there is evidence that both marijuana and hemp have

been intensely selected for varying cannabinoid levels

(McPartland and Guy, 2010). Inbred lines produce pre-

dominately either THC (Type I) or CBD (Type III), and

crosses between pure-breeding lines produce offspring of

an intermediate chemotype with approximately equal

amounts of both THC and CBD (Type II); inbreeding of

the F

intermediate generation produces F

offspring in a

Figure 7. Cannabis BLH and BLD landraces are utilized for a wide range of purposes, are generally unselected for cannabinoid ratio, and

synthesize approximately equal amounts (on a population average) of both CBDA and THCA. However, there are usually a few plants

that produce much more THCA than CBDA, and a few others that produce much more CBDA than THCA. If the high-CBDA plants and

high-THCA plants are isolated in separate breeding populations, and selection for either high-CBDA or high-THCA is continued for sev-

eral generations, descendant populations will develop either high-CBDA or high-THCA proﬁles (from Clarke and Merlin, 2013).

CRITICAL REVIEWS IN PLANT SCIENCES 301

simple 1:2:1 ratio including »25% that produce nearly

pure THC proﬁles (Type I), »50% of intermediate can-

nabinoid ratio (Type II), and »25% that produce almost

entirely CBD (Type III). Marijuana-type (drug) plants

inherit a highly expressed THC-acid or THCA synthase

allele; hemp-type (ﬁber) plants inherit a highly expressed

CBD-acid or CBDA synthase allele; and intermediate

types inherit both active synthase genes. Mendelian phe-

notypic segregation in the F

generation indicates a pat-

tern of inheritance where THCA and CBDA synthases

are co-dominant allelic variants (B

and B

) at a single

enzymatic locus (de Meijer et al., 2003). However, subse-

quent genome research suggests that THCA and CBDA

synthases are controlled by a number of homologous

alleles at separate linked loci (Kojoma et al., 2006; van

Bakel et al., 2011).

Recent research by Weiblen et al.(2015) sheds new

light on the inheritance and evolution of cannabinoid

biosynthesis. Cannabis plants synthesize the carboxylic

acid derivatives of cannabinoids, THC-acid or THCA

and CBD-acid or CBDA, from the same precursor mole-

cule, cannabigerolic acid (CBG-acid or CBGA). How-

ever, CBDA synthase has a higher afﬁnity for CBGA

molecules than THCA synthase; therefore, when both

synthases are active, more CBGA substrate is converted

to CBDA than THCA. Low-THC hemp cultivars do

have an active THCA synthase allele, but because they

also have an active CBDA synthase allele they produce

predominately CBD and little THC. On the other hand,

high-THC marijuana drug cultivars have both an active

THCA synthase allele and a nonfunctional CBDA syn-

thase allele (e.g., “Skunk No. 1”), but because the CBDA

synthase gene is nonfunctional there is no competition

between the differing alleles for substrate, and conse-

quently nearly all the CBGA is converted to THCA. This

explains why these cultivars are high in THC with little if

any CBD. Also, according to Weiblen et al.(2015), “the

marijuana-type THCA synthase allele may be dominant

over the hemp-type allele, and the functional CBDA syn-

thase allele may be dominant over the nonfunctional

allele.”Heterozygosity at separate loci, gene duplication,

mutation, and divergence of homologous alleles facili-

tated positive selection of cannabinoid synthase genes,

both functional and nonfunctional (especially the non-

functional CBDA synthase allele). This artiﬁcial selection

precipitated key evolutionary steps leading toward

potent high-THC marijuana. Hybrid vigor, intensive

selection for more and larger glandular trichomes (Small

and Naraine, 2015), allele mutation, gene duplication,

and selection of potent alleles at an unlinked locus con-

trolling cannabinoid quantity likely also played impor-

tant roles in increasing the psychoactive potency of

sinsemilla cultivars.

Traits apparently correlated with potency, such as

later maturation, larger inﬂorescences, and/or increased

resin production, provided clues for early humans, indi-

cating which plants were most psychoactive. It is easy to

observe which inﬂorescences sparkle most in sunlight

and feel the stickiest, and both of these traits indicate

higher resin production and potential mind-altering

potency. Many times in various locations potent wild

and feral plants were selected as a source of seed for sow-

ing; this propelled differing gene pools on separate,

divergent evolutionary paths upon which they remained

more or less genetically isolated from other cultivated

gene pools, as well as from wild and feral populations.

Continued farmer selections for increased potency led to

higher THC content in NLD and BLD landraces. More

recently, hybridization between potent NLD and BLD

landraces resulted in modern sinsemilla cultivars with

increased vigor and potency.

Present-day North American and European drug cul-

tivars are hybrid descendants of two genetically divergent

drug Cannabis gene pools—NLD and BLD. Traditional

Asian NLD landrace farmers from Thailand, India, and

South Africa, as well as New World farmers in Mexico,

Colombia, and Jamaica, maintained varietal purity by

sampling individual plants and selecting seeds from the

most ﬂavorful and potent ones to sow the following year.

Consequently, original NLD landraces imported into

North America had a relatively high THC content with

little if any CBD. On the other hand, imported BLD

landraces from Afghanistan were relatively high in both

THC and CBD. Afghan sieved hashish (a mechanically

concentrated resin gland preparation) was collected

from entire populations, so individual plant selection for

potency was difﬁcult and rarely practiced. As a result,

cannabinoid chemotypes of individual BLD plants varied

widely from producing nearly all THC to nearly all CBD,

with a bell curve distribution in between focused on a

median ratio of 1:1, much like unselected feral East Asian

BLH populations, but with a higher total cannabinoid

content resulting from selection for the readily apparent

traits of larger inﬂorescences and more resin glands,

especially on the small leaves subtending the ﬂowers

(Figure 8).

Dried sinsemilla inﬂorescences can contain more than

20% THC and/or CBD by weight. How did modern drug

cultivars become such proliﬁc cannabinoid producers?

This unprecedented production results from a combina-

tion of simple heritability of cannabinoid proﬁle, hybrid

vigor between evolutionarily distinct and genetically dis-

tant gene pools (NLD and BLD), and subsequent, highly

focused, human artiﬁcial selection for potency. Early

NLD £BLD crosses expressing great vigor and diversity

were used to breed seed cultivars, from which present-

302 R. C. CLARKE AND M. D. MERLIN

day production clones were selected. Sinsemilla cultiva-

tion provided a rigorously selective regimen resulting in

cultivars with extremely high THC levels from their

NLD ancestry, and little if any CBD from their BLD

ancestry. Seed-propagated drug cultivars must be rigor-

ously selected to maintain their potency. Feral drug Can-

nabis populations lose potency quickly because open

pollination does not impose selective pressures favoring

high THC content, low CBD content, or favorable aro-

matic proﬁles. The atavistic lowering of potency in natu-

ralized populations may indicate that THC is not of

great natural adaptive signiﬁcance. Apparently,

enhanced THC synthesis is primarily an artifact of

human selection, and represents another key aspect of

the ancient human–Cannabis relationship (Clarke and

Merlin, 2013).

Natural evolutionary determinants were important

before humans started using and spreading Cannabis,

and natural selection will always play a role in its evolu-

tion. Domestication may have had more profound effects

on the evolution of the functional physiology of Canna-

bis rather than its anatomical physical traits, and perhaps

plants of this genus were so well preadapted for agricul-

ture that few morphological changes were required. Nev-

ertheless, human selection during domestication has had

by far the greatest inﬂuence on changes in Cannabis phe-

notypes, both morphological and physiological. This

process accelerated during the last half of the twentieth

century as industrial hemp and marijuana breeders

developed new cultivars through vigorous artiﬁcial

selection.

V. Twentieth-century cannabis breeding

Cannabis produces copious quantities of both pollen and

seed, yet it is not an easy plant to improve by selective

breeding. Cannabis populations are almost always dioe-

cious, with male and female ﬂowers occurring on sepa-

rate plants which are therefore normally incapable of

selﬁng (self-fertilizing). Selﬁng is the most effective sexu-

ally reproductive means of ﬁxing desirable traits, because

selected genes are more likely to be represented in both

the male pollen and the female ovule if they come from

the same plant. In traditional Cannabis breeding, reces-

sive alleles controlling a selected trait locus (aa) must be

present in two separate individuals, one male or pollen

parent (either Aa, aA, or aa) and one female or seed par-

ent (also either Aa, aA, or aa). As a result, in open-polli-

nated outcrossers such as Cannabis, qualitative traits are

usually controlled at single loci by dominant allelic forms

(AA, Aa, and aA), which are much more common than

recessives (aa) (3:1 ratio) and heritability is high. Quanti-

tative traits are more often controlled by differing allelic

forms at various loci; as a result, heritability is low, which

makes improvement by breeding more difﬁcult. Female

plants supply most of Cannabis’economically valuable

products, including ﬁbers, seeds, or drugs, whereas male

plants merely fertilize the females and are occasionally

harvested for their ﬁber. This makes it difﬁcult for plant

breeders to recognize potentially favorable traits in a

male parent, as these traits must ultimately be expressed

in female offspring. All Cannabis plants are wind-polli-

nated, which allows them to intercross freely; therefore,

in order to avoid random fertilization and seed set,

selected female seed parents must be isolated from males

until they are to be pollinated by a selected male parent.

Successful breeding of open-pollinated cultivars requires

the identiﬁcation of plants with favorable traits, and then

creating breeding lines via recurrent selection, making

hybrid crosses between these lines, and ﬁeld testing their

progeny (Posselt, 2010).

Figure 8. Cannabis NLD and BLD plants vary in appearance

throughout their growth cycles. In addition to their narrower leaf-

lets, juvenile NLD plants are taller and more laxly branched than

the shorter, more compact Afghan BLD plants (from Clarke and

Merlin, 2013).

CRITICAL REVIEWS IN PLANT SCIENCES 303

It is a costly and lengthy process to develop true-

breeding Cannabis seed cultivars. In addition to a high

degree of planning, attention, and patience, a breeder

needs organization, infrastructure, and stability to pursue

crop improvement goals; and then must successfully

maintain commercial cultivars through continued selec-

tion of elite lines. Large population sizes are required in

order to select very few individual breeding plants with

favorable combinations of economically interesting char-

acteristics. Much of recent drug Cannabis domestication

has taken place secretly without legal sanctions, using

small populations with a high percentage of select survi-

vors. Under these circumstances, clandestine breeders still

made lucky crosses; however, progress is much faster with

more plants to choose from, more focused selection crite-

ria, and lower selection percentages. Commercial drug

Cannabis seeds have been widely available for more than

30 years, but continued improvements relied on the many

hobby breeders who made interesting crosses and distrib-

uted the seeds to other breeders. Although breeders oper-

ated on limited scales and in near secrecy, the sum of

their creations overshadows their individual roles; indeed,

their cumulative efforts and early crowd sourcing laid the

foundations for the plethora of asexually reproduced sin-

semilla cultivars available today.

Sexually reproducing and outcrossing Cannabis may

(by the nature of its genome randomizing reproductive

system) be resistant to morphological changes associated

with domestication in other crop plants. Until the recent

popularity of asexual propagation to ensure the perpetu-

ation of consistently high-yielding, high-potency drug

crops, Cannabis cultivars had not become completely

domesticated, but now the asexual sinsemilla varieties

survive entirely at the discretion of humans. The discus-

sion that follows documents the twentieth-century

breeding of ﬁber, seed, and drug cultivars; this is where

we outline the pedigree and lineage of example cultivars,

emphasizing the roles of individual plants from diverse

gene pools in cultivar domestication and evolution via

human selection.

VI. Industrial hemp breeding

European NLH landraces were introduced to the New

World by sixteenth- and seventeenth-century colonists,

and eventually they became naturalized to a broad range

of local climates and soil regimes. Even though ﬁber

yields of the early introductions were relatively low, they

were well-suited for seed production, and were the only

hemp grown in North America until the early 1890s.

During the late nineteenth and early twentieth centuries,

Japanese hemp was introduced into California, and Chi-

nese hemp into Kentucky. Both of these early

introductions from East Asia (likely broad-leaﬂet hemp,

i.e., BLH varieties) were from a gene pool that evolved

independently from European NLH varieties. The Japa-

nese introductions were lost, whereas descendants of

Chinese introductions into Kentucky became known as

“Kentucky”hemp. By the turn of the twentieth century,

the United States Department of Agriculture (USDA)

hemp breeding experiments revealed that the introduced

Chinese varieties would lose their favorable characteris-

tics of high ﬁber quality and yield when randomly repro-

duced by farmers. Soon after seed production was

purposefully restricted to only a few isolated farms under

the attention of plant breeders, the commercial “Ken-

tucky”cultivar became relatively uniform and genetically

stable (Boyce, 1900). Later hybrid crosses involved the

‘Ferrara’cultivar from Italy (possibly an NLH/BLH

hybrid), ‘Kentucky’hemp (NLH?/BLH), and additional

Chinese BLH landraces. Subsequently, further selections

were made from these complex hybrids to produce even

higher ﬁber yield. Sadly, no known North American

hemp cultivars remain alive today, only their feral

offspring.

During the twentieth century, European hemp

breeders improved locally available, native landraces and

also made crosses with alien imports to create higher

yielding ﬁber and seed varieties. It is evident from their

breeding histories and chemotaxonomic proﬁles that

many present-day European cultivars share common

ancestors (de Meijer, 1995) and are less genetically

diverse (Hillig, 2004a) than their related feral popula-

tions and landraces. Crosses between select individuals

from NLH landraces belonging to the Mediterranean

and Central Russian ecotype groups (Serebriakova,

1940) formed the basic breeding lines of European hemp

cultivars. Brief summaries of their individual and shared

breeding histories are provided below: these outline the

breeding strategy employed, with an emphasis on the

interactions of differing gene pools and the effects of

individual plant selections. For a more detailed discus-

sion of European hemp cultivars, see de Meijer (1995)

and Clarke and Merlin (2013).

Open-pollinated, outcrossing plants like Cannabis are

naturally heterozygous having a higher frequency of dif-

fering alleles at the gene loci, especially for many domi-

nant traits (Aa and aA); however, they lose vigor when

they are inbred to achieve homozygosity with a higher

frequency of identical allele pairs at recessive gene loci

(AA and aa). The primary objective of breeders is to

“ﬁx”selected traits through homozygosity while main-

taining overall heterozygosity and hybrid vigor in the

cultivar. There are several techniques used to breed Can-

nabis cultivars. Early hemp and drug landraces were

developed through recurrent mass selection. Farmers

304 R. C. CLARKE AND M. D. MERLIN

favor homozygosity by sowing seed collected each year

from plants with agriculturally valuable traits, whereas

natural selection for health and vigor encourages hetero-

zygosity in the remaining traits. Mass selection is most

effective in improving simple qualitative characteristics

with high heritability, although persistent recurrent

selection can lead to changes in more complex quantita-

tive characters with relatively low heritability. Mass

selection is used effectively in hemp breeding to enhance

highly heritable traits such as ﬁber and cannabinoid con-

tent (Hennink, 1994).

Mass selections or phenotypic selections are usually

made by choosing the best plants in the ﬁeld before har-

vest, and saving their seeds for sowing. Both mass family

and single plant selections are made to produce more

genetically homogenous lines for breeding. This is essen-

tially how local feral populations become landraces. Indi-

vidual female plants are selected, but a single female

plant is usually fertilized by a random assortment of

many unselected male plants, and, as a result, most of

the offspring are half siblings, rather than full siblings. In

an outcrossing plant such as Cannabis, mass selection is

more efﬁciently imposed by identifying superior male

plants in a population prior to pollen dehiscence,

destroying all males that do not ﬁt selection criteria, and

then collecting seed only from select females. This was

how the dioecious European hemp cultivars were

selected and maintained (e.g., “Carmagnola,”“Kom-

polti,”“Lovrin,”and “Novisadska”). Recently, Chinese

hemp cultivars (e.g., “YunMa 1”and “YunMa 5”) were

mass selected from local Yunnan province landraces

(Salentijn et al., 2014).

As scientiﬁc breeding progressed, monoecious hemp

cultivars based on the “Fibrimon”cultivar were developed

by crossbreeding selected individuals from different land-

races and cultivars to increase variability and vigor. Cross-

breeding to combine desirable traits is achieved by

fertilizing a favorable female plant with pollen from a sin-

gle favorable male plant. The seeds are grown and the F

population is open-pollinated so each female is fertilized

by many males. Each selected F

female is evaluated by

growing out her offspring (F

), and favorable lines are

continued (F

,…) by further half-sibling selections. A

select female can also be fertilized by only one select male

from each population and then offspring will be full sib-

lings. This promotes homozygosity while further narrow-

ing genetic variability in the offspring. Single crosses

between a female and a male parent were also used to

establish most sinsemilla cultivar lineages (see below).

Monoecious plants with male and female ﬂowers, and

sub-dioecious female plants with male ﬂowers, allow

self-fertilization and inbreeding. Monoecious individuals

were used to breed monoecious hemp cultivars (e.g.,

“Fibrimon”) and female unisexual cultivars (e.g.,

“Uniko-B”). Female unisex cultivars could serve a similar

purpose as male-sterile lines if they produce no viable

pollen, have very high seed yields, and can be used to

facilitate hybrid seed multiplication. Recessive traits are

ﬁxed by serially selﬁng sub-dioecious females, and vigor

is restored by combining two or more lines in the ﬁnal

hybrid cultivar (de Meijer, 2004). Hybrid single cannabi-

noid medical cultivars have been developed by the fol-

lowing methods: (1) selecting individual female plants

containing the target cannabinoid (or other compound)

from several genetically diverse origins; (2) using hor-

monal stimulants to form pollen-bearing male ﬂowers

on each female plant; (3) selﬁng each female line in isola-

tion for several generations (S

,…) to encourage

homozygosity and “ﬁx”production of the target com-

pound; and ﬁnally, (4) crossing genetically less related

individual inbred lines to restore vigor in the hybrid cul-

tivar (de Meijer, 2014).

The majority of present-day European hemp cultivars

are monoecious. The French cultivar “Fibrimon”was

selected directly from an inbred monoecious line created

from a selfed, individual plant with both male and female

ﬂowers (monoecious or sub-dioecious) that occurred

spontaneously in a NLH population cultivated in Central

Russia. Pseudo-monoecious cultivars were developed by

crossing monoecious “Fibrimon”with dioecious female

high-ﬁber German NLH selections (also originally from

Central Russia), or with late-ﬂowering NLH landraces

from Italy or Turkey, and then backcrossing to “Fibri-

mon.”Genetic homogeneity of monoecious cultivars

must be maintained each year by careful selection of

monoecious parents and the creation of elite monoecious

seed. Second-generation crops sown from elite seed con-

sist almost entirely of monoecious plants, and are multi-

plied by open pollination to produce commercial sowing

seed for ﬁber cropping. Third-generation ﬁber crops

include up to 20% true male and female individuals

resulting from inbreeding and natural genetic drift in the

absence of human selection. Traditional dioecious land-

races were easily maintained by local farmers who pro-

duced their own seed, but today farmers must purchase

sowing seed each year because monoecious cultivars

immediately decline without careful selection and breed-

ing (de Meijer, 1995); this creates a market situation sim-

ilar to seed production of hybrid maize (or corn, Zea

mays).

Although the traditional Italian ﬁber cultivar “Car-

magnola”and its descendants “Fibranova,”“Eletta Cam-

pana,”and “Superﬁbra”are practically unavailable today,

the Italian gene pool formed an important modern foun-

dation for both European and American hemp breeding.

East Asian BLH landrace varieties that were brought to

CRITICAL REVIEWS IN PLANT SCIENCES 305

Italy early on were responsible for much of the heterosis

effect (hybrid vigor) sought by European hemp breeders.

All of the Italian cultivars were dioecious, and most are

now believed extinct. Open-pollinated “Carmagnola”

was of great antiquity, having originated from a Chinese

BLH landrace, and it formed the basis of Italy’s famed

high-quality hemp textile production. “Carmagnola Sele-

zionata”was selected directly from “Carmagnola,”while

‘Fibranova’and ‘Eletta Campana’resulted from crosses

between traditional ‘Carmagnola’and high-ﬁber German

strains obtained from the Central Russian NLH landra-

ces that were also used in breeding ‘Fibrimon.’

Traditional Hungarian landraces and cultivars are mostly

dioecious. For example, “Kompolti”was selected for high

ﬁber content from an Italian variety likely related to “Car-

magnola.”Intentional heterosis breeding resulted in several

hybrid cultivars. “Uniko-B”is the progeny of “Kompolti”

crossed with monoecious “Fibrimon,”and ﬁrst-generation

crops of this cross consist of nearly all female plants, result-

ing in very high seed yields. Unisexual female lines produce

almost no pollen and can be used for seed parents in hybrid

seed production. Second-generation commercial ﬁber and

seed crops produce approximately 30% males and still have

a relatively high seed yield. “Kompolti Hibrid TC”is a three-

way-cross hybrid in which two selections of Chinese origin

(BLH), one dioecious and the other monoecious, were

crossed to produce a unisexual, almost purely female hybrid

known as “Kinai Uniszex.”In the second generation, the

unisex line is used as the female parent in the crossing of

“Kinai Uniszex”£“Kompolti”cultivars. This hybrid

produces commercial sowing seed of the triple-cross (pre-

dominately BLH) ﬁber cultivar “Kompolti Hibrid TC”with

a restored 50% female to 50% male sex ratio, hybrid vigor,

and increased ﬁber yield. “Kompolti S

argasz

ar

u”is a dioe-

cious, chlorophyll-deﬁcient, yellow-stemmed Hungarian

paper pulp cultivar obtained from a cross between “Kom-

polti”and a yellow-stemmed mutant offspring from a cross

between Finnish early-maturing NLH and Italian late-

maturing hemp (possibly an NLH/BLH hybrid); this cross

was then repeatedly backcrossed with “Kompolti”to trans-

fer (introgress) the yellow-stemmed trait to homozygosity

in the “Kompolti”cultivar (Figure 9).

Polish hemp cultivars are monoecious. For example,

‘Bialobrzeskie’resulted from serial crossings of dioecious

NLH cultivars of predominately Russian origin that were

ﬁnally crossed with ‘Fibrimon,’followed by long-term

single line selections for high ﬁber content. Another Pol-

ish hemp cultivar, “Beniko,”was obtained by individual

progeny selection from a “Fibrimon”hybrid.

Romanian “Fibramulta 151”is a dioecious selection

from a single cross between “ICAR 42–118”and German

“Fibridia.”The parent variety of “ICAR 42–118”was the

progeny of an Italian “Carmagnola”£“Bologna”

(another Italian cultivar) hybrid crossed with the Turkish

“Kastamonu”landrace. The dioecious “Lovrin 110”culti-

var was bred by selection among family groups from the

Bulgarian “Silistrenski”NLH landrace. Monoecious

“Secuieni 1”was derived from the crossing of “Dneprov-

skaya 4”£“Fibrimon,”followed by two semi-back-

crosses with “Fibrimon 21”and “Fibrimon 24,”

Figure 9. Four ﬁber hemp cultivars can be visually distinguished in this ﬁeld trial. The Hungarian cultivar “Kompolti S

argasz

ar

u”or “yel-

low-stemmed Kompolti”is a reduced stem chlorophyll variety that was bred as feedstock for the paper industry (from Clarke and Merlin,

2013).

306 R. C. CLARKE AND M. D. MERLIN

respectively. The Russian dioecious parent “Dneprov-

skaya 4”was a descendant from Italian hemp.

Eight cultivars are presently grownincentralandsouth-

ern parts of Ukraine and several regions of Russia, where

the history of traditional and commercial hemp ﬁber and

seed cultivation is very lengthy. Hemp cultivars from the

former USSR are classiﬁed into maturity groups or geo-

graphical types (Serebriakova, 1940;deMeijer,1995). Cur-

rent cultivars belong to either the southern, later-maturing

group or to hybrid progenies from an earlier-maturing cen-

tral group crossed with later-maturing southern hemp;

hybrid cultivars of the latter group are intended for sowing

in higher latitudes of central and northern Europe than

those ecologically preadapted to southern Europe. The dioe-

cious southern Russian cultivar “Kuban”was obtained by

ten cycles of family group selection in the hybrid population

by the crossing of “Szegedi 9”£“Krasnodarskaya 56”culti-

vars. “Szegedi 9”was selected in Hungary from the native

“Tiborsz

all

asi”landrace (NLH) and crossed with “Krasno-

darskaya 56,”which probably came from a cross between a

local Caucasian (NLH) landrace and an Italian cultivar. The

two other cultivars cultivated in areas of Ukraine and Russia

are “Zenica,”adioecioussouthwesternEurasiancultivar,

and the NLH landrace “Ermakovskaya Mestnaya,”which is

cultivated on a signiﬁcant scale in Siberia and belongs to the

Central Russian maturity group (Serebriakova, 1940).

The remaining Russian and Ukrainian monoecious

cultivars exhibit a southern, later-maturing growth pat-

tern, but are also cultivated at higher latitudes to pro-

mote a longer vegetative growth period, resulting in

taller stalks and increased ﬁber yield. These cultivars

include “USO-11”bred from three parental populations,

“Dneprovskaya 4,”“USO-21,”and “Dneprovskaya

Odnodomnaya 6.”“USO-11”is presently grown in

Canada and New Zealand for oil seed production. The

remaining Ukrainian cultivars were produced by cross-

ing various Russian landrace accessions with cultivars

originating in Italy and France (de Meijer, 1995).

“Finola”is an early-maturing Finnish variety used for

seed production in more northern latitudes and was reg-

istered in the European Union (EU) in 1999. It is grown

extensively in Canada for seed and seed oil production

(Small and Marcus, 2002). “Finola”was selected from

the pooled seed of two nearly identical, early-ripening,

far northern Russian NLH landraces originally obtained

from the Vavilov Research Institute (VIR) in St. Peters-

burg, Russia. The Canadian seed variety “Anka”was also

selected from VIR accessions. “Finola”matures early and

provides both high seed yields (up to 1.7 metric tons per

hectare or 1500 pounds per acre) and short straw

(stalks); in addition, it can be harvested by combine har-

vesters for both ﬁber and seed yields from the same crop

(Weightman and Kindred, 2005).

European hemp breeders developed more uniform

monoecious varieties with increased resistance to various

pests (McPartland et al.,2000), while also focusing on

further reducing already negligible levels of THC. The

French cultivar “Santhica”was selected from a single

plant with no THC and high CBG content; and it may

lack the THC synthase gene (B

allele) (de Meijer, 2004),

but it has not been released commercially. Although

research institutes might develop new hemp cultivars for

speciﬁc uses, government organizations and/or seed

companies continue to reproduce and sell sowing seed of

existing varieties and traditional farmers continue to

propagate their landrace seed. As long as hemp remains

an economically viable albeit minor crop, hemp cultivars

will be multiplied; however, as soon as economic interest

wanes, they will most likely become extinct.

Present-day European industrial hemp cultivars share

common ancestors. Most ﬁber cultivars originated from

“Carmagnola,”a traditional hybrid Chinese-Italian BLH/

NLH dioecious cultivar that was combined with various

Central Russian NLH landraces. On the other hand,

while nearly all monoecious varieties are descendants

from French ‘Fibrimon,’a single, inbred, monoecious

cultivar that was also derived from a Central Russian

landrace. Genes of the modern industrial hemp cultivars

came from two taxonomically distinct Cannabis groups,

European NLH and Asian BLH, but via only a handful

of individual parents and hybrid crosses. Hemp cultivars

are maintained as select elite populations, which ensures

inbreeding to preserve favorable traits; some outcrossing

is allowed to encourage health and vigor, but all within a

heavily selected and fundamentally restricted genome

preserved by intensive artiﬁcial selection.

Early in the twentieth century, Chinese BLH landraces

were incorporated into North American hybrid hemp

cultivars, and were later used by European breeders to

make hybrid crosses with NLH landraces of European

origin. The high degree of genetic difference between

East Asian BLH and European NLH gene pools was

illustrated by Hillig (2004a) who placed them in subspe-

cies of different species—C. indica ssp. chinensis and C.

sativa ssp. sativa, respectively. The heterosis effect or

hybrid vigor resulting from crossing plants of diverse

genotypes was important in the development of all mod-

ern ﬁber cultivars. As we will see below, heterosis breed-

ing was also a key factor in the initial development of

present-day drug cultivars.

VII. Sinsemilla cultivars

Early humans repeatedly introduced NLD landraces that

originated in southern Asia into unusual geographical

regions. Many of these introductions evolved into new, local

CRITICAL REVIEWS IN PLANT SCIENCES 307

landraces that eventually (sometimes centuries later)

entered the international market. In the 1960s and early

1970s, all commercial marijuana imported into North

America and Europe owed its heritage to the vast NLD

group, and seeds of many different landraces were available

in relatively large amounts. Seeds from Colombian and Thai

marijuana grown at more nearly equatorial latitudes rarely

produced plants that matured to the ﬂoral stage when culti-

vated outdoors at temperate latitudes before cold autumn

weather killed them, whereas northern Mexican and Jamai-

can varieties often matured earlier, before harsh winter

weather set in. The domesticated NLD hybrid varieties of

the early and mid-1970s originated largely from crosses

between Mexican or Jamaican landraces and more potent,

but later ripening, Colombian, Indian, Panamanian, and/or

Thai landraces (Figure 10).

Traditional Asian and New World drug Cannabis cul-

tures favored and selected potent landrace varieties that in

turn provided North American and European breeders with

the basic genetic building blocks for developing sinsemilla

varieties. In the early 1970s, gardeners began to grow sinse-

milla marijuana by removing all male plants from their

ﬁelds, leaving only the unfertilized (therefore seedless)

female plants awaiting pollination, each producing thou-

sands of ﬂoral bracts covered by a myriad of resin glands.

Sinsemilla cultivation also brought growers to the realization

that a few select male plants couldbeisolatedfromwhich

they could collect pollen to intentionally fertilize ﬂowers on

a few female branches. This in turn produced seeds of

known parentage which essentially gave birth to the inten-

tional breeding of potent hybrid drug Cannabis. Early clan-

destine breeders combined and recombined imported NLD

landraces in multi-hybrid crosses, and by 1980 North Amer-

ican sinsemilla ranked among the world’s most potent drug

cannabis.

Traditionally, both hemp and drug cultivars were

developed by variations on the theme of mass selection,

but modern sinsemilla cultivars evolved in a different

way from hemp cultivars. Sinsemilla cultivars were usu-

ally developed by crossing a single male of one geneti-

cally distinct landrace with a single female of another

landrace (e.g., Mexican NLD £Colombian NLD) to cre-

ate an F

hybrid. In the subsequent F

generation,

selected male or female offspring were bred by following

one of three basic pathways: (1) they were inbred with

one or more siblings (e.g., Mexican/Colombian £Mexi-

can/Colombian) to establish a heterozygous and rela-

tively inconsistent F

population hybrid used for

subsequent mass selection to increase homozygosity and

uniformity; (2) they were backcrossed with an “aunt”

(seed parent) or “uncle”(pollen parent) (e.g., Mexican/

Colombian £Colombian) to increase homozygosity and

enhance speciﬁc traits before establishing a backcross

population hybrid used for subsequent mass selection;

(3) they were outcrossed with an unrelated plant (e.g.,

Mexican/Colombian £Thai) to create a new F

hybrid

serial generation. Strategies one and two were followed

by some early sinsemilla breeders, but pathway three was

by far the most frequently pursued. By introducing a

new exotic landrace or hybrid parent into the breeding

line (e.g., [Mexican/Colombian £Thai] £Jamaican

landrace), heterozygosity and diversity were increased by

creating a new F

hybrid in each serial breeding genera-

tion. Although the offspring looked less and less like any

Figure 10. Narrow-leaﬂet drug (NLD) type Cannabis, closely related to this late-nineteenth-century Indian ganja, was spread from South

Asia through Africa and eventually to the New World (from Clarke and Merlin, 2013).

308 R. C. CLARKE AND M. D. MERLIN

of their increasingly distant ancestors, a level of relative

homogeneity between siblings persisted because in each

new generation half of the gene pool was contributed by

a single parent. Although serial hybridization was not an

intentional breeding strategy, the heterozygosity breeders

unwittingly created a cultivation process that has played

an important role in the continued survival of the inces-

tuously recombined drug Cannabis gene pool

(Figure 11).

In all three paths of drug cultivar breeding described

above, landraces were crossed to make hybrid popula-

tions. However, no true hybrid seed varieties were cre-

ated because no pure-breeding lines were maintained to

use each year as parents of the F

hybrids. Only more

recently have sinsemilla seed companies multiplied true

hybrids by preserving asexually reproduced vegetative

clones of select male and female parent plants. However,

the clonal parental lines are often highly heterozygous,

and their “hybrid”offspring are frequently inconsistent

phenotypically.

Sinsemilla breeders have selected primarily for stron-

ger potency (higher THC content) as well as complex

aromas and ﬂavors, which are all traits related to ter-

peno-phenolic secondary product metabolism in the

glandular trichomes (e.g., cannabinoids and terpenoids).

Initial serial hybridization followed by careful inbreeding

(pathway 1, described above) and/or backcrossing (path-

way 2) of hybrid crosses resulted in some of the early

NLD varieties that are still popular today (e.g., “Original

Haze,”“Big Sur Holy Weed,”etc.). During the late

1970s, many clandestine marijuana breeders successfully

developed early-maturing and high-yielding, potent, aro-

matic and colorful connoisseur NLD varieties, and as a

consequence demand for high-quality sinsemilla grew.

Soon an exotic new series of cultivar introductions would

dramatically change the face of sinsemilla breeding and

the evolution of domesticated drug Cannabis.

VIII. Introduction of the Afghan cultivars

Broad-leaﬂet drug (BLD) seeds were introduced from

Afghanistan into North America and Europe several

times from the late 1970s through the 1980s. BLD plants

were very distinctive in appearance; being well-branched

and standing only one to two meters (about three to six

feet) tall at maturity, they were shorter and bushier than

NLD plants with broader, darker green leaﬂets. After

BLD varieties were introduced in the late 1970s, growers

commonly began to call the original NLD varieties “sati-

vas”because their tall growth and narrower, lighter green

leaﬂets more closely resembled NLH ﬁber varieties than

they did the more recently introduced shorter and darker

green Afghan BLD hashish varieties commonly called

“indicas.”The authors now assume that NLD and BLD

populations belong to different subspecies of C. indica,

subspecies indica (NLD) and subspecies afghanica

(BLD).

Afghan BLD plants mature earlier than most NLD

cultivars at northern temperate latitudes (from late

August through September) providing an abundant

source of highly psychoactive resin traditionally used to

make Afghan hashish (Clarke, 1998). Seedless BLD ﬂow-

ers smelled and tasted much like potent hashish, and

connoisseurs paid premium prices for such an exotic

contraband. Breeders crossed BLD varieties with their

sweet-tasting, but later-maturing, NLD varieties to pro-

duce early-maturing BLD £NLD hybrids, commonly

called “indica/sativa”hybrids (in our view, hybrids of dif-

ferent subspecies of C. indica). Hybrid vigor was strong,

and ﬂower yields increased, yet plants rarely exceeded

3 m (10 feet) in height. Afghan £Thai hybrids proved

particularly pungent and extremely potent. By the mid-

1980s, the vast majority of sinsemilla in North America

owed a portion of its heritage to the BLD gene pool.

Thereafter, it became increasingly difﬁcult to ﬁnd either

pure NLD (pre-BLD) varieties or the pure BLD introduc-

tions popularized only a few years earlier. Australian and

Paciﬁc Island outdoor varieties were often based on

Figure 11. North American and European marijuana growers

crossed NLD varieties from Mexico (above) with NLD varieties

from many other locations (e.g., Colombia, Jamaica, Thailand,

and India), and these became the ﬁrst hybrid “home-grown”sin-

semilla cultivars (from Clarke and Merlin, 2013).

CRITICAL REVIEWS IN PLANT SCIENCES 309

Southeast Asian landraces introduced in the 1970s, and

until more recently were less affected by BLD/NLH

introductions and indoor surreptitious cultivation

(Figure 12).

During the 1970s and 1980s, relatively early in the

evolution of modern-day sinsemilla cultivars, NLD seeds

inadvertently imported in illicit marijuana were far more

commonly sown than artiﬁcially selected hybrids pro-

duced by domestic growers. Even after the introduction

of Afghan cultivars, rare, intentionally bred seeds were

sometimes passed from one breeder to another, but their

distribution was quite limited. During the mid-1980s,

seed companies began to distribute selected varieties, but

accidentally produced seeds from illicit commercial sin-

semilla were still much more commonly grown. Seeds

carrying varying proportions of the NLD and BLD gene

pools were grown and crossed repeatedly without con-

scious selection; this inadvertently produced multi-cross

hybrids in which culturally favorable traits were rarely

selected and thus the average psychoactive and aromatic

qualities of domestic marijuana decreased.

Most of the offspring derived from these crosses were

dissimilar to their parents and their siblings, in both

appearance and potency because their genomes consisted

of small random pieces of genetic information inherited

from assorted predecessors, and undesirable

characteristics that previously had been suppressed

through careful selection were often expressed (e.g.,

lower potency, dull psychoactive effects, acrid aroma,

harsh taste, etc.). Many of the early NLD/BLD cultivars

offered hardy growth, rapid maturation, and tolerance to

cold, allowing sinsemilla to be grown outdoors in many

northern temperate climates. This revolutionized the

domestic marijuana market both geographically and cul-

turally by widening the scope and popularity of outdoor

sinsemilla cultivation. NLD/BLD hybrids not only

changed Western outdoor growing, but their rapid matu-

ration time, short stature, and high yields also pre-

adapted them to indoor growing as well. Tall and late-

maturing NLD cultivars were poorly suited for artiﬁcial

light growing, but shorter and faster ripening NLD/BLD

hybrids were ideal for commercial indoor production.

The combination of increased prohibition pressure,

improved lighting systems, asexually propagated indoor

cultivars, and high marijuana prices fueled the rapid,

very widespread strategy of cultivating sinsemilla indoors

in many regions of the world. The rampant success of

this agricultural model narrowed the selection criteria of

breeders toward developing only indoor cultivars, further

limiting the present-day genetic diversity of drug Canna-

bis. Although consumers and commercial growers

largely accepted NLD/BLD hybrids, serious sinsemilla

breeders began to view them with more skepticism, and

started to develop new drug cultivars for a wider variety

of applications (Figure 13).

IX. Present status of the Cannabis genome

As a result of millennia of dissemination and contact with

differing human cultures worldwide, Cannabis plants,

through natural and artiﬁcial selection, have evolved huge

genetic diversity which is expressed in a remarkably wide

range of phenotypes. Traditional farmers in geographically

isolated regions selected and maintained genetically unique

landraces that formed the foundation for breeding modern

hybrids. The Cannabis gene pool is amazingly diverse; how-

ever, breeders’well-intentioned domestication efforts have

been strongly affected by the repercussions of prohibition

and consequently its genetic diversity is increasingly being

attenuated.

The vast majority of European industrial hemp culti-

vars are based on the same few single plant selections;

and the drug Cannabis gene pool has reached a genetic

bottleneck caused by incest breeding between close rela-

tives, intentional and unintentional selﬁng, highly

directed selection for drug potency, and the preponder-

ance of vegetatively reproduced commercial populations.

The genetic diversity of Cannabis is constantly reduced

by persistent law enforcement seizures, accidental

Figure 12. Broad-leaﬂet drug (BLD) Afghan hashish cultivars

were introduced into Western Europe and North America in the

late 1970s, and hybrid crosses with NLD cultivars changed

domestic drug Cannabis production forever. This large BLD plant

is growing in a family garden in northern Afghanistan (from

Clarke and Merlin, 2013).

310 R. C. CLARKE AND M. D. MERLIN

agricultural mishaps, and the gradual culling of economi-

cally unfavorable clonal lines; commercially valuable

clones are propagated, and the remainders are destroyed.

If we project present trends into the future, genetic diver-

sity will continue to decrease, the lack of sexual recombi-

nation will lower the potential for evolving pest

resistance, and susceptibility to agricultural pests and

diseases will continue to increase. When pathogenic

organisms such as viruses, fungi, mites, and aphids attack

a genetically uniform, asexually reproduced crop, losses

are often extensive. Illegal drug Cannabis cultivation and

breeding have historically suffered from similar problem-

atic pest infestations that hindered and often over-

whelmed the successful cultivation of other monocrops

of commercially valuable plant species.

Devastating agricultural blights have indeed caused

great economic losses in several different crops, and are

directly associated with diminished genetic variability

leading to susceptibility to a pest or pathogen. Susceptibil-

ity to pathogens can arise in two ways. When a cultivar is

introduced to a new region it may be infected by intro-

duced alien organisms for which it has no naturally

evolved resistance (e.g., Afghan BLD cultivars parasitized

by gray mold in humid regions of North America). Alien

pests have also been introduced into new regions where

Cannabis crops have no natural resistance (e.g., the

broomrape parasite introduced to Kentucky from China

where it caused widespread damage to hemp crops, and

hemp russet mites introduced to Indiana along with South

Asian seed stocks). McPartland et al.(2000)provides

more detailed histories of Cannabis pests and diseases.

History presents us with several different serious

blight scenarios. The mid-nineteenth-century Great

Famine of northern Europe was caused by a virulent

strain of Phytophthora infestans, a parasitic and nonpho-

tosynthetic fungus-like organism, which is closely related

to brown algae; P. infestans commonly infects crop

plants, but not Cannabis (McPartland et al., 2000). The

virulent HERB-1 strain of P. infestans, that causes the

potato blight, originated in Mexico; it subsequently

spread into North America where it caused widespread

damage to potato crops, and then, in 1844, it traveled by

ship to Europe where the infection was rapidly dispersed.

Regions where a wider variety of potato cultivars were

being cultivated were not as severely affected, but the

majority of potatoes in Ireland were of the “Irish

Lumper”variety, which proved to be particularly suscep-

tible. The Irish potato crop loss in 1845 is estimated at

between one- third and one-half of cultivated acreage,

and in the next year three quarters were lost to blight.

Nearly a million people starved to death, and another

two million emigrated, many to the United States. Potato

cultivars are asexually multiplied by dividing tubers, so

every offspring is genetically identical to its parent plant,

and therefore all share the same susceptibility to pests

and diseases. Periodic failures of the northern European

potato crop persisted until the early twentieth century

when breeders produced potato cultivars resistant to

HERB-1. Drug Cannabis clones are asexually multiplied

by cuttings, and like potato cultivars all the plants within

each clone are equally susceptible to pests and diseases,

which, as we well know, spread rapidly in monoculture

crops (Kelly, 2013).

During the mid-nineteenth century, the Great Wine

Blight affected many vineyards and devastated the Euro-

pean wine industry. During the third quarter of the nine-

teenth century more than 40% of French vineyards were

destroyed, and the suffering economy also accounted for

increased emigration to North America. The blight was

caused by the phylloxera aphid (Daktulosphaira vitifo-

liae) that originated in eastern North America, where it

was a well-known grape (Vitis) root parasite. North

American Vitis rootstocks had been introduced to

Europe years before, but they did not carry the pest. In

the late 1850s, phylloxera-infested Vitis rootstocks were

transported across the Atlantic by steamship, allowing

the pest to reach Europe alive, where it spread rapidly

Figure 13. Female Cannabis ﬂowers produce a preponderance of

glandular trichomes or resin glands that secrete a cannabinoid-

and terpene-rich essential oil. Selection for high production of

resin glands and essential oil is the driving force in modern-day

drug Cannabis breeding, to the virtual exclusion of other agro-

nomically valuable traits (Photo by Todd McCormick from Clarke

and Merlin, 2013).

CRITICAL REVIEWS IN PLANT SCIENCES 311

through European vineyards of traditional Vitis vinifera

wine cultivars that had not evolved natural immunity to

a North American pest. Grape breeders crossed suscepti-

ble European wine cultivars with phylloxera-resistant

native North American species, but the offspring were

not particularly resistant to phylloxera and they made

poor wine. The solution lay in reintroducing phylloxera-

resistant rootstocks from North America (Vitis aestivalis

or other American native species) and replacing the sus-

ceptible V. vinifera rootstocks. Following considerable

resistance by traditional French wineries, endangered

French wine grape scions were successfully grafted onto

rootstocks of resistant North American species and the

European wine industry was resurrected (Ordish, 1987).

Traditionally, the vine plants of table and wine grapes

have been vegetatively propagated by grafting shoots, much

the way modern sinsemilla varieties are propagated by root-

ingcuttings.ArecentgeneticsurveybyMyleset al.(2010)

of grape accessions in the USDA collection came to the fol-

lowing conclusion: “We propose that the adoption of vege-

tative propagation was a double-edged sword: Although it

provided a beneﬁtbyensuringtruebreedingcultivars,it

also discouraged the generation of unique cultivars through

crosses. The grape currently faces severe pathogen pressures,

and the long-term sustainability of the grape and wine

industries will rely on the exploitation of the grape’stremen-

dous natural genetic diversity.”

In the early 1900s, the American chestnut blight,

caused by a wind-borne fungus (Cryphonectria para-

sitica), decimated American chestnut tree forests before

eventually spreading to Europe. This aggressive fungus

arrived in North America with infected Chinese orna-

mental chestnut trees that were resistant to the disease.

The blight is spread by wind-borne pathogenic spores

that moved the fungus quickly through dense native

chestnut stands, eventually killing an estimated 40 billion

trees; every chestnut tree in the contiguous forests of the

eastern United States succumbed to the disease in less

than 40 years (Freinkel, 2007). The only survivors were

apparently resistant trees or those in small outlying pop-

ulations that the disease did not reach. Subsequently, an

isolated, disease-resistant chestnut tree from Ohio that

had survived was crossed with three naturally resistant

Chinese cultivars. The resulting hybrids remained dis-

ease free, but were shorter than the average American

chestnut trees. Breeding efforts continued by backcross-

ing the hybrid offspring three times to the American

chestnut to transfer the genes for disease resistance to a

cultivar with the taller American chestnut phenotype.

Southern corn leaf blight (SCLB) is a fungal disease

found in many of the world’s maize-growing areas. This

blight is caused by Bipolaris maydis, which occurs in

three genetically variant races. Race T infects maize

plants with Texas male-sterile cytoplasm (Tcms), which

contains a gene causing susceptibility to the race T fun-

gus. Maize plants are monoecious and self-pollinating;

and before this susceptibility was clearly understood,

breeders encouraged a disastrous fungus invasion by

unwitting selection of maize plants infected with Tems

cytoplasm.

During the traditional multiplication of hybrid maize

seed, farmers remove the male ﬂowers (tassels) from the

female seed plants by hand so they will not self-fertilize,

and therefore will only receive pollen from monoecious

plants grown in the same ﬁeld. Removing tassels is time-

consuming and as a result it provides summer employ-

ment for students, but it is costly and prone to human

error; consequently, in the early 1960s, breeders devel-

oped lines with Tcms to be used as male-sterile seed

parents in hybrid maize seed production. Hybrid sowing

seed rapidly gained popularity, and by the late 1960s an

estimated 90% of the American crop contained Tcms

cytoplasm, making it extremely vulnerable to SCLB. By

1970, SCLB reached epidemic proportions destroying

15% of America’s corn production with losses estimated

at one billion dollars. The SCLB epidemic occurred

because female seed lines carried an unknown male-

linked trait that conferred disease susceptibility to a huge

percentage of the crop. As a result, farmers returned to

detasseling hybrid seed crops until breeders selected

SCLB resistant cultivars and found sources of male steril-

ity other than Tcms (Agrios, 2005). Male-

sterile seed lines could be used in monoecious hemp

breeding, and growing cuttings of sterile high-THC

females would guarantee that drug crops would remain

seedless wherever they are grown, offering another

advantage to sinsemilla growers.

In 1890, a wilting disease of banana was observed in

the “Gros Michel”plantation crops of Costa Rica and

Panama that reached epidemic levels throughout the

early 1900s. This precipitated catastrophic damage, forc-

ing banana plantations to move to new regions where

the disease had not yet spread. In 1910, the causal agent

of “Panama disease”in Cuban bananas was linked to the

soilborne wilt fungus, Fusarium oxysporum f. spp.

cubense (Foc wilt). Cannabis is also susceptible to closely

related Fusarium wilts (McPartland et al., 2000). The

entire dessert banana trade was based on the vegetatively

reproduced “Gros Michel”(Big Mike) cultivar that had

no resistance to the Foc wilt, and by the 1960s was

almost entirely wiped-out. A search for banana varieties

resistant to Foc wilt revealed specimens from botanical

gardens in the United Kingdom and the United Fruit

collection in Honduras. These were vegetatively repro-

duced by tissue culture, and by the late 1960s two closely

related “Cavendish”clones, “Dwarf Cavendish”and

312 R. C. CLARKE AND M. D. MERLIN

“Grand Nain,”were widely spread throughout the com-

mercial banana producing areas of Central America and

beyond. Presently, these two clones account for nearly

50% of global banana production (http://www.promusa.

org/TropicalCraceC4C-CTR4). Also in the 1960s, Pan-

ama disease was ﬁrst noticed in Taiwanese “Cavendish”

bananas. Subsequently, the Panama disease became a

very widespread and serious problem in various areas of

the world; it destroyed Taiwan’s banana industry, spread

through southern China, wiped out plantations in Indo-

nesia and Malaysia, threatens the Australian and Philip-

pine industries, and has also been identiﬁed in Middle

Eastern and East African plantations. Unfortunately,

Panama disease spreads easily and potentially endangers

monoculture banana production everywhere. Regretta-

bly, the banana industry has been slow to respond to the

renewed threat of Panama disease, and presently there is

no market replacement for the “Cavendish”cultivars.

Another pest infestation threatening plantations of

bananas is the virulent Foc wilt strain known as Tropical

Race 4 (TR4) that was not identiﬁed until 1994. TR4 is

now spreading rapidly, infecting susceptible plantations

of “Cavendish”clones, all of which are asexually repro-

duced and have no chance to develop resistance to TR4

(http://www.promusa.org/TropicalCraceC4C-CTR4).

The blights described brieﬂy above were caused by

different pests and diseases affecting various wild and

cultivated plant species, but their rampant spread

relied on one key factor—the underlying cause of all

these epidemics has been a lack of genetic diversity.

The “Irish Lumper”potato was an asexually repro-

duced clone; French varietal wine grapes shared com-

mon ancestors and the roots of all were susceptible to

phylloxera; genetically similar American chestnut trees

formed huge contiguous forests that all perished; the

American corn blight infected all hybrid maize culti-

vars that shared Tcms cytoplasm; and “Cavendish”

bananas are all members of two closely related clones.

These historically documented disasters highlight the

need to further understand the diversity and present

status of the Cannabis genome, and should give us

pause for thought.

By the dawn of the twenty-ﬁrst century, it was appar-

ent that Dutch sinsemilla varieties were predominantly

based on three founding Paciﬁc coast varieties: “North-

ern Lights,”“Skunk No. 1,”and “Haze”(Clarke, 2001).

Dutch, British, and Spanish companies are the largest

suppliers of drug Cannabis seeds today, mostly selling

multi-hybrid combinations of “old”varieties. In the

twenty-ﬁrst century, many “new”varieties have been

incorporated into the modern collective gene pool; how-

ever, many of them include reassorted alleles from the

seminal Cannabis varieties distributed far and wide

30 years ago. What appear as novel combinations today

reﬂect a breakdown of suites of characteristic landrace

traits, and the persistent reshufﬂing of individual land-

race genes in modern multi-hybrid cultivars. Over many

generations, sinsemilla cultivars have become hybridized

to the point of inbreeding and are increasingly similar

genetically to one another.

Because Cannabis can be a difﬁcult plant in which to

ﬁx traits through selective breeding, and only female

plants are of economic importance for drug production,

asexual reproduction or “cloning”by rooting vegetative

female cuttings provides an easy solution for rapidly

expanding commercial sinsemilla production. Under

this female vegetative planting regime, there are no male

plants to remove, the crop can be treated uniformly, all

the plants ripen at the same time, and they can be har-

vested together. If the electricity grid does not fail, grow

rooms can produce three to four crops per year, and

yield over 450 g (one pound) of dry ﬂowers per square

meter (about 10 square feet) each harvest. Male plants in

breeding programs can also be preserved in a vegetative

“library”under long photoperiod and induced to ﬂower

when pollen is required. Cloning radically changed Can-

nabis agriculture (and selective breeding pressures) by

making sinsemilla growing possible and proﬁtable for

hobbyists. However, in one generation, asexual cloning

ﬁxes traits forever (apart from artiﬁcially or naturally

induced mutations) and attenuates the genetic diversity

promoted by sexual seed reproduction (Figure 14).

Reliance on only a few select male and female cuttings

for seed production reduces genetic diversity. The number

of female clones used as seed parents in sinsemilla breed-

ing is now limited, but even fewer pollen parents are used

to make commercial seed. Seed varieties and clones are

most often named after their female seed parent, but

equally important is what plant was used as a pollen par-

ent. Male parents that consistently produce uniform off-

spring are much harder to ﬁnd than female parents,

which are often maintained as commercial cultivars. Indi-

vidual male plants are often selected for phenotypic traits

resembling females (e.g., increased vigor, short internodes,

profuse branching, dense inﬂorescences, strong aroma,

etc.). However, in order to identify which male clone will

produce the best offspring, each of these male clones

must be crossed with a range of female cuttings. The seeds

from each of these crosses must then be progeny-tested to

check for quality and consistency. For example, if ten pro-

spective male clones are each crossed with ten individual

female clones, and 200 seeds (approximately 100 will be

female) of each test cross are sown, about 10,000 female

plants will be produced, and then they must be screened

to determine which male and female clone combinations

produced the best offspring.

CRITICAL REVIEWS IN PLANT SCIENCES 313

A brief summary of the breeding of “Skunk No. 1”will

illustrate what is involved in developing a relatively true-

breeding seed cultivar. Several plants were grown from a

local California “skunk”variety, which was likely a

NLD and 1

4BLD hybrid (e.g., Colombian/Afghan £

Mexican) and then all the females were crossed with a

single select male. A female identiﬁed as skunk plant

number one was selected as the highest yielding and

most potent, and became the founder of all subsequent

generations. For the following nine growing seasons, at

least one branch of each of the (up to 100) female plants

was fertilized by pollen from a male selected from the

offspring of the previous year’s best female plant. Seeds

from the select females were sown the following spring.

After only two or three generations, “Skunk No. 1”was

relatively homogenous and true-breeding compared to

other hybrid lines that continued to segregate in the F

and F

generations, and “Skunk No. 1”was then deemed

ready for large-scale selection of parent plants with spe-

ciﬁc combining ability (SCA) for subsequent breeding.

A total of nearly twenty thousand plants were grown

in a common garden from the seeds of ten females from

the most promising lines selected from the previous gen-

eration, and the ten best females from among all those

lines were selected based on vigor, potency, type of effect,

ﬂower yield, high ﬂower to leaf ratio, resin gland

development, amount of branching and pest resistance,

as well as attractive ﬂoral aromas and ﬂavors. Ten males

were also selected for their vigor, pest resistance, female-

type growth form, and aroma. All 20 selections (male

and female) were reproduced asexually to preserve their

unique genotypes.

Ten clonal copies were made of each of the ten select

females. Pollen was collected from each of the ten select

male clones, and pollen from a single male was used to

fertilize a single copy of each female clone, resulting in

100 individual crosses. Two hundred seeds of each cross

were sown (20,000 total) and approximately 100 female

progeny of each cross were evaluated for phenotypic

consistency (homozygosity) of their favorable agronomic

traits (see above). Five female and three male clones

were selected and used for “Skunk No. 1”sowing seed

multiplication as well as hybrid cultivar development

(Figure 15).

Few sinsemilla seed companies invest as much time

and money in selecting superior male plants as they do

for specially selected females. Most simply select a male

plant based on visual characteristics alone without any

chemical analyses, and cross that individual male with

each of their female clones to produce commercial

hybrid seed. These seeds (all containing genes from a sin-

gle male parent) are then widely distributed and grown

to maturity, and from which female cuttings are then

selected for commercial sinsemilla production. Commer-

cial Cannabis seed production can be compared ﬁgura-

tively with historical human conquests. A dominant

invading male defeats all his rivals, procreates with select

captive women, selects one son as his chosen heir, and

marries his choicest daughters off to distant lands;

thereby ensuring his genetic dominance at home, while

spreading his genes far and wide.

More recently, “all-female”cultivars have been pro-

duced by transforming a female plant (XX) with hor-

mone applications to produce male ﬂowers with viable

female (XX) pollen. The offspring of female plants fertil-

ized with female pollen are all female. Every seed produ-

ces a useful female plant and there is no need to cull

male (XY) plants, which provides the advantages of asex-

ual propagation, but in the convenience of a seed. Female

seeds lower genetic diversity for the same reason as

female cuttings, because there are no male plants, and

therefore sexual reproduction and recombination are all

but impossible without making hybrids. Furthermore,

although “all-female”seeds are sexually reproduced, only

a few female plants can successfully be transformed into

high-yielding pollen parents. Often, only one trans-

formed female pollen parent is used to fertilize seeds on

all the recipient females, and thus all of the “all-female”

offspring share the genes from that single transformed

Figure 14. Uniform modern-day drug Cannabis crops are grown

from transplanted vegetative and asexually reproduced cut-

tings—either inside under lights, in greenhouses, or outdoors

under the sun—on various scales (from Clarke and Merlin, 2013).

314 R. C. CLARKE AND M. D. MERLIN

Figure 15. “Skunk No. 1”was among the ﬁrst nearly true-breeding NLD/BLD hybrid drug cultivars developed, beginning in the 1970s.

The “Skunk No. 1”single line was selected and bred through nine reproductive cycles to increase homozygosity and the resulting unifor-

mity. This was followed by selection in a common garden of individual breeding parents (ten female and ten male) that were asexually

reproduced before being intercrossed, and their offspring were later grown out and evaluated to determine their speciﬁc combining

ability (SCA). Five female and three male plants were preserved for further seed production.

CRITICAL REVIEWS IN PLANT SCIENCES 315

pollen parent. For example, in Spain, a single “White

Widow”NLD/BLD hybrid cutting that transforms easily

(as well as reliably producing all female offspring) is used

to pollinate almost all of the individual female NLD/BLD

clones for which “all-female”F

hybrid seed cultivars are

named. However, several seed companies also produce

single-selfed (S

) cultivars by crossing each transformed

female plant with other members of the same female

clone, with no apparent loss in vigor.

All-female cultivars have gained popularity in hobby

and indoor growing markets, and are also increasingly

used in commercial cannabis production regions. Well-

intentioned Westerners took modern NLD/BLD seeds to

Morocco in the 1980s and crossed their progeny with the

traditional Rif Mountain kif or local marijuana landrace.

Within ten years, traits from the “improved”Western

varieties were seen in nearly all the Rif populations. Pres-

ently, all-female cultivars are increasingly popular in

Morocco. The females are completely fertilized by male

plants growing nearby and the “better”improved seed is

sown the next year. In this way local landraces are

quickly swamped out by invasive genes, and are therefore

replaced by an unselected crossbred mix of Western vari-

eties that are poorly adapted to local environmental con-

ditions and processing methods. Mexico, Jamaica, and

Thailand have also lost many of their landrace popula-

tions (Figure 16).

In addition to “all-female”cultivars, breeders have

developed “auto-ﬂowering”cultivars that commence

ﬂowering when they reach an early stage in vegetative

development rather than waiting for decreasing day

length to trigger ﬂowering. In addition, all modern

“auto-ﬂowering”cultivars likely have a common ances-

tor, possibly the “Finola”NLH seed cultivar or another

variety which ﬂowers independent of photoperiod con-

straints. Now these two traits are combined in “all-

female/auto-ﬂowering”cultivars. Although originating

from two diverse gene pools (NLD/BLD and NLH), these

cultivars are presently reproduced using only a few

parents. Auto-ﬂowering plants cannot be maintained as

Figure 16. During the 1960s and 1970s, NLD landraces from diverse geographical regions which contained widely differing genetic

combinations were combined to form hybrid NLD sinsemilla cultivars. In the 1980s, several BLD landraces from Afghanistan were intro-

duced and thus added more diversity to the sinsemilla gene pool. By the 1990s, asexual propagation was commonly used to capture

heterozygous genotypes for use as commercial clonal cultivars. The widespread (yet narrowly focused) search for cultivars with unique

aromas and effects, combined with the proliferation of asexually reproduced cutting populations, obviated the need for sexual seed

reproduction, but also lowered the genetic diversity of drug Cannabis as a whole. Recently, inbreeding to produce all-female and auto-

ﬂowering seed lines has further restricted genetic diversity.

316 R. C. CLARKE AND M. D. MERLIN

clones because they will ﬂower and die even under long

day length, but could possibly be maintained in vitro and

regenerated into explants.

In the past decade, many novel cultivars have

appeared, and, with the exception of variations in aro-

matic terpenoid proﬁles, they are largely very similar in

the remainder of their phenotypic traits and agronomic

attributes. This has occurred not only because of the

incessant and thorough reshufﬂing of a diverse, yet lim-

ited, set of genes and their respective alleles; more

recently the homogenization of cultivars and their traits

has been produced by accidental incest between closely

related siblings or cousins. Many popular “female”clonal

cultivars will produce a few functional male ﬂowers

when they are stressed by any of a wide variety of factors

that are not far outside of optimum growing conditions

(e.g., changes in temperature, overabundance or shortage

of nutrients, ﬂuctuations in light cycles, etc.). The male

ﬂowers of these popular “female”clonal cultivars pro-

duce viable pollen and result in unintentional fertiliza-

tion of the neighboring “female”plants. Unstable female

sexuality becomes reinforced generation after generation

because plants with unstable sexuality are the pollen

parents. Usually only a few accidentally produced seeds

occur in sinsemilla (because it is usually seedless). How-

ever, among these, a high percentage will be sprouted,

most will be “female,”and they will strongly resemble

other popular cultivars, but with their own slight differ-

ences; consequently, the accidentally produced seeds that

do mature are likely to be brought into cultivation.

Along with the unfavorable tendency to produce male

ﬂowers and unwanted seed, modern varieties have some

other drawbacks: they are increasingly ﬁnicky about

growing conditions, are susceptible to pests (e.g., various

mites) and disease (e.g., gray mold and powdery mil-

dew), and are generally difﬁcult to grow. Increasing

numbers of pests and diseases lead to increasing use of

unapproved pesticides and plant growth regulators,

which can directly impact public health.

Without open-pollinated sexual reproduction, there is

much less chance for evolutionary change. In the past, as

early agriculturalists spread across new geographical

frontiers, they exposed Cannabis to new and evolution-

arily challenging environments. The tremendous genetic

diversity of geographically isolated wild populations and

traditional landraces of Cannabis produced by sexual

reproduction and genetic recombination presented novel

phenotypes for natural and human selection. Narrow

selection and strong founder effects have seriously atten-

uated genetic diversity, bringing us closer to potential

disaster. A ﬁnal blow may come from commercialization

of a small number of select varieties that satisfy limited

consumer preferences. Under the assumption of eventual

legalization, and modern monocropping paradigms

favored by “economically prudent”strategies to maxi-

mize proﬁts, fewer clonal cultivars will pass the con-

sumer ﬁlter, and it is likely that only these will be

proliferated. Variety registration to protect breeder’s

rights will favor asexually reproduced clones because it is

easier to demonstrate uniqueness, uniformity, and repro-

ducibility with clones than it is by breeding consistent

seed cultivars. Given the present-day economic and

political climates, and the increasing trend toward asex-

ual reproduction in both indoor and outdoor crops, this

scenario seems probable. If this outcome comes to pass,

it will further narrow the Cannabis gene pool to include

only the most commercially viable cultivars, and there-

fore, more of less repeating the history of varietal wine

grape selections and potentially leading to another major

crop disaster.

Whether varieties are lost through prohibition, custo-

dial neglect, or economic priorities, they become extinct

and are potentially gone forever. When indigenous farm-

ers maintain localized landraces, they are preserving

genetic diversity because landrace populations reproduce

sexually, and thus allow for genetic recombination,

mutation, and evolution under human selection. The

numbers and ranges of traditional agricultural societies

are diminishing worldwide, and many landraces are

already extinct, which does not bode well for the future

of the Cannabis gene pool. The loss of any genetic diver-

sity threatens unique genes and allelic combinations, as

well as lowering future evolutionary ﬁtness for the genus

as a whole and limiting the diversity of desirable prod-

ucts that crop plants may be able to provide.

The status of the worldwide Cannabis genome has

changed dramatically since the middle of the twentieth cen-

tury. The majority of traditional ﬁber, seed, and drug landra-

ces are no longer grown, and many of the few remaining in

situ landraces have been genetically diluted through inter-

breeding with introduced modern cultivars. Also, European

industrial hemp cultivars share much of the same germ-

plasm—dioecious cultivars were bred by crossing middle

Russian and Chinese landraces, and monoecious cultivars

all share “Fibrimon”as a pollen source—and overall the

genetic diversity of NLH is currently very limited (Hillig,

2005a,b). Commercial sinsemilla cultivars are asexually

reproduced, and almost all share varying amounts of both

NLD and BLD heritage. Founder effects from BLD landrace

selections introduced gray mold and powdery mildew sus-

ceptibility that cause millions of dollars in crop losses annu-

ally. Novel gene combinations still arise because the modern

drug Cannabis gene pool is based traditional landraces from

geographical origins that represent two genetically distinct

primary gene pool, NLD and BLD. Whenever a seed is

planted, it represents the sexual union of two parents, and

CRITICAL REVIEWS IN PLANT SCIENCES 317

creates an opportunity for new genetic combinations to

appear. Asexually reproduced cuttings are potentially

genetic dead ends—only sexual reproduction offers us a

truly brighter genetic future and the possibility for plant

improvement.

X. Seed banks

When a Cannabis landrace is not reproduced every ﬁve

to ten years, the stored seeds will most likely die and the

landrace may be gone forever. Seeds must be properly

kept in a gene bank and reproduced periodically under

ideal conditions. The past 50 years have seen the genetic

diversity of the Cannabis genome dwindle away. Indeed,

the vast majority of landraces may already be extinct,

and we therefore must be careful to preserve and multi-

ply what remains. As Watson and Clarke (1997) warned,

“Many local landrace varieties, the result of hundreds of

years of selection for local use, have been lost because of

Cannabis suppression and eradication, neglect on the part

of agricultural ofﬁcials and industry, anti-hemp propaganda

and the general trend (until recently) to reduce industrial

hemp breeding and research. Genetic materials are a living

heritage and we are their custodians. We must concentrate

our efforts to collect, preserve, characterize and utilize the

remaining Cannabis genetic resources before it is too late.”

As the worldwide reduction in Cannabis diversity

continues, the importance of genetic preservation

becomes more obvious. Unfortunately, no comprehen-

sive Cannabis germplasm collections exist. Most of the

few seed accessions are held by national gene banks that

may or may not share their valuable inventories with

breeders in other countries. The largest collection of

hemp germplasm is maintained by the Vavilov Institute

of Plant Research (VIR) in St. Petersburg, Russia. It pres-

ently numbers 563 seed accessions, including 23 possible

drug accessions from Afghanistan, Kazakhstan, Syria,

Turkey, and Uzbekistan, while the remaining are all

hemp and feral accessions from Armenia, Bulgaria,

Chile, China, Czechoslovakia, Estonia, France, Germany,

Hungary, Italy, Latvia, Moldova, Poland, Portugal,

Romania, Russia, Spain, Sweden, Ukraine, the United

States, and former Yugoslavia (Grigorev, 2015). Since the

late 1980s, political, technical, and ﬁnancial difﬁculties in

Russia have resulted in low population sizes and incom-

plete isolation, and consequently there has been consid-

erable loss of genetic diversity and purity in the VIR

collection (Hillig, 2004b). Many accessions may now be

so similar to each other that their importance to future

breeding programs could be diminished.

In 1992, the Cannabis germplasm collection at Wage-

ningen University in the Netherlands contained over 156

accessions originating from 22 countries and largely

sourced from other collections and research institutes

(De Meijer and van Soest, 1992; Gilmore et al., 2007).

Nearly half of these accessions are from the former

USSR and Hungary. The Institute of Natural Fibres and

Medicinal Plants gene bank collections in Poland contain

139 accessions of predominantly European origin, with

accessions from France, Hungary, and the Ukraine

contributing 54.7% of the collection (Mankowska and

Silska, 2015). The Yunnan Academy of Social Sciences

collection in Yunnan province, China holds approxi-

mately 350 accessions mostly of East Asian origin

(Salentijn et al., 2014) and the Ecoﬁbre Global

Germplasm Collection in Australia contains additional

Eurasian accessions (Welling et al.,2015). However,

comprehensive accession data are sorely lacking in

several of these collections and this limits their value to

breeders (Welling et al.2016).

In addition, the subterranean Svalbard Global Seed

Vault on the Norwegian island of Spitsbergen about

1300 km (810 miles) from the North Pole has a total of

43 Cannabis accessions that are duplicated in three other

seed banks. Five of these accessions, from North Korea,

Netherlands, Spain, Syria, and Turkey, may possibly be

Cannabis drug populations; 21 others are hemp acces-

sions from Argentina, Austria, China, Croatia, France,

Georgia, Germany, Italy, Poland, Romania, Slovakia,

Spain, and Sweden; and 16 accessions are of unknown

origin (http://www.nordgen.org/sgsv/). The world’s larg-

est seed repository is the Millennium Seed Bank housed

at the Wellcome Trust Millennium Building in West Sus-

sex, near London, which specializes in wild plants and

has only one Cannabis accession, which is from Slovakia

(http://apps.kew.org/seedlist/SeedlistServlet).

Given the importance of Cannabis as a traditional as

well as present-day crop plant, the biodiversity of this

genus (particularly among the drug cultivars) is sorely

under-represented in seed banks, especially in light of

recent research interest in medical Cannabis. If we take

into account this lack of diversity, in light of genetic

impurity and low seed numbers, there really is no reliable

reserve of Cannabis seeds. The primary goal of germ-

plasm preservation is the conservation of the entire

genome of each population. It is especially important in

open-pollinated, cross-breeding plants that the popula-

tion size is large enough to ensure that as many of the

alleles as possible within each gene pool are reproduced

in the seed. A minimum of 1000 plants for monoecious

accessions, and 2000 plants for dioecious accessions,

assures that 99% of the Cannabis alleles will be repro-

duced (Crossa et al., 1993). Unfortunately, the seed

reserves of many of the Cannabis seed bank accessions

consist of less than 1000 viable seeds (often only 500 or

less); therefore, genetic diversity is already limited by the

318 R. C. CLARKE AND M. D. MERLIN

number of archived seeds. The secondary goal of genetic

preservation is to reproduce the accessions in sufﬁcient

quantities to maintain a reserve for future reproductions

and public distribution.

A common goal of Cannabis breeders should be

establishing a more comprehensive core collection of

Cannabis seed accessions that have been exhaustively

characterized agronomically in the ﬁeld, and on molecu-

lar levels, genetically and chemically, in the laboratory.

Only then, can we see what diversity really is available

for researchers to work with in the future. This core col-

lection should be maintained with optimal reproduction

and storage methodology, and individual accession eval-

uations should be made accessible to breeders (Watson

and Clarke, 1997).

In the past 20 years the situation has only stagnated.

According to Welling et al. (2016), in their ﬁne review of

the present state of ex situ Cannabis germplasm

collections:

“Coordinated and comprehensive conservation and

characterization of ex situ Cannabis resources holds the

promise of preserving genepool diversity and enabling cul-

tivar development. However, the legal constraints imposed

by international narcotics conventions over more than

50 years have been inﬂuential in the fractionation and

erosion of publicly accessible Cannabis ex situ genetic

resources. The restrictions on legal exchange of bona ﬁde

research materials continues to limit the establishment of

physical and centralized ex situ core collections.”

XI. Present and future directions for Cannabis

breeding start here!

The primary evolutionary process that is presently fur-

thering domestication in Cannabis is basic Mendelian

breeding; this traditional system involves selecting sim-

ply inherited traits and increasing their homozygosity

through sexual recombination. Many of the primary eco-

nomic traits of Cannabis are simply inherited (e.g., stalk

height, seed size, and cannabinoid content). Cannabi-

noid biosynthesis is controlled by a narrow range of

alleles limited to only a few loci; heritability is extremely

high, which has favored successful breeding for high-

THC and high-CBD sinsemilla cultivars as well as low-

THC industrial hemp cultivars. High variability and

strong heritability also inﬂuence both ﬂowering and

maturity times, and breeding for earlier-maturing drug

varieties and later-maturing ﬁber varieties continues.

Selection for reduced sensitivity to day length could

improve the versatility of ﬁber hemp varieties making

breeding and seed multiplication easier. Late-maturing

or nonﬂowering ﬁber hemp varieties with low THC con-

tent would signiﬁcantly improve biomass yield and

extend the range where ﬁber hemp can be grown eco-

nomically into sub-tropical and equatorial regions

(Salentijn et al., 2014). Indeed, breeders are expected to

make further advances on all fronts as Cannabis cultiva-

tion becomes more widespread for a variety of uses.

There are many achievable goals awaiting twenty-

ﬁrst-century hemp breeders. Fiber hemp breeders will

certainly focus on raising ﬁber content and yield (e.g.,

increasing ﬁber percentage, extending vegetative period),

improving stalk and ﬁber quality for speciﬁc uses (e.g.,

paper pulp, composites, and ﬁne textiles), altering canna-

binoid content and composition (e.g., low THC percent-

age with high CBD yield), developing cultivars that are

easier to process (e.g., faster retting, easier peeling, and

cheaper dyeing), and increasing resistance to pests

and diseases. Hemp ﬁber is more easily extracted and

processed when the ﬁber bundles are high in cellulose

and low in lignins and pectins that bind the ﬁbers

together within the bundles. Sequencing the genes

responsible for the production of these compounds and

the ways they are assembled could assist in cultivar

development for the textile and paper industries

(Mandolino and Carboni, 2004).

Seed hemp breeders could develop cultivars with spe-

ciﬁc protein and fatty acid compositions tailored for spe-

ciﬁc whole seed, kernel, and oil markets, as well as easily

processed (e.g., harvesting, threshing, and hulling) culti-

vars, and should learn to control monoecy for hybrid

seed production (Salentijn et al., 2014).

Industrial hemp cultivation is presently restricted to

temperate latitudes, but its economic range could be

widely enlarged by developing cultivars adapted to lower

latitudes, or by selecting for day-neutral cultivars that

could be grown at any latitude. All commercial hemp

cultivars are presently adapted to growing at or above 45

degrees latitude, and will yield less ﬁber when they are

relocated as little as two degrees closer to the equator.

Closer to the equator, the day length is shorter earlier in

the summer, and consequently the Cannabis crop will

ﬂower prematurely when plants are small, which lowers

ﬁber and seed yield. European hemp cultivars cultivated

closer to the equator than within their normal latitudinal

farming range are also attacked by a host of alien pests

and diseases for which they have little resistance. As a

result, no hemp varieties approved by the EU can be suc-

cessfully grown closer to the equator than at about 40

degrees latitude. Limited traditional hemp production in

other regions of the world relies mostly on unimproved

local landraces (Clarke, 1995,2007; Watson and Clarke,

1997; Clarke and Gu, 1998).

Further advances in industrial hemp breeding are also

expected in the near future. Economically efﬁcient ﬁber

production relies on increasing the scale of production,

CRITICAL REVIEWS IN PLANT SCIENCES 319

reducing production costs, and tailoring production for

speciﬁc markets. These goals could be achieved in part

by breeding for increased yield of biomass along with

increased ﬁber yield and quality, reducing waste such as

dust, and improving ﬁber color. Field experiments on

small test plots indicate that hemp may have the poten-

tial to produce much higher stalk and ﬁber yields than

are presently achieved. Improved yields of both ﬁber and

seed varieties could be attained by continued selection

for optimum ﬂowering date, enhanced resistance to pests

and diseases, and improved canopy architecture.

Improvements developed by breeders for speciﬁc uses,

such as ﬁber size, aspect ratio, strength, stiffness, density,

surface characteristics, adhesive properties, ligniﬁcation,

and color, must be carried out in concert with develop-

ment of optimized agronomic regimes and speciﬁc proc-

essing strategies for woven, nonwoven, and composite

applications. Ideal hemp ﬁber varieties should have a

high proportion of ﬁne and easily extracted ﬁbers that

enhance rapid processing, as well as producing less

waste. However, ﬁber quality and ease of retting are con-

trolled by a complex set of genes which interact with

agronomic and processing conditions that result in a

variety of difﬁculties for ﬁber hemp breeders, including

lower heritability of ﬁber traits. On the other hand, seed

fatty acid and protein syntheses are controlled by fewer

genes encoding for speciﬁc products; this makes selection

and breeding for seed traits more straightforward

(Weightman and Kindred, 2005).

One of the primary goals of EU hemp breeders over

the past several decades has been the reduction of THC

levels. In 2001, the EU lowered allowable THC levels to

0.2%, although previously mandated at below 0.3%. This

legal reduction of allowable THC levels created problems

for hemp breeders faced with maintaining cultivar pro-

ductivity while further reducing already negligible psy-

choactive drug content. The lowering of THC levels

encouraged breeders to further increase inbreeding with

monoecious lines and was a major factor leading to

reduced genetic variability in present-day European cul-

tivars. Breeders are now working on “medicinal hemp”

drug cultivars that are high in CBD while remaining low

in THC. There is an ancient tradition of growing multi-

purpose Cannabis plants in various areas of China. How-

ever, very little had been cultivated during modern times,

until Chinese farmers recently began growing hemp

more intensely for ﬁber and seed as well as by-product

CBD production. Indeed, two high-CBD industrial

hemp cultivars should have been registered, with sowing

seed released by the Yunnan Academy of Agricultural

Sciences in 2016 (Salentijn et al., 2014).

Two predominant trends of Cannabis breeding have

appeared in recent decades. On the one hand, Cannabis

has been modiﬁed by industrial hemp breeders who have

lowered THC content and raised ﬁber content

while developing uniform, high-yielding cultivars. On

the other hand, sinsemilla breeders have also altered the

genetic makeup of drug varieties effecting quantitative

traits such as raising THC content, ﬂower to leaf ratio in

the inﬂorescences, and yield of ﬂowers. Intense selection

for potency explains in part why drug cultivars are “gen-

erally less polymorphic and heterozygous than hemp cul-

tivars”(Weiblen et al., 2015). Additional qualitative traits

in drug cultivars have also been altered recently such as

increasing branching, shortening internodes, and select-

ing for rapid maturation and early ﬂoral response to

inductive photoperiods.

The following four key evolutionary events differenti-

ate drug cultivars from ﬁber and seed cultivars: (1) ram-

pant and continued blending of the modern domestic,

hybrid drug gene pool; (2) exportation of “improved”

drug seed; (3) gene ﬂow of modern hybrid traits back

into traditional drug populations; and (4) the establish-

ment of unique, asexually reproduced clonal drug culti-

vars. Recent stages in the evolution of modern drug

Cannabis varieties have few parallels in the simple breed-

ing history and legal distribution of hemp.

Although industrial hemp breeders still develop ﬁber

and seed cultivars, the major thrust of Cannabis breeding

today is directed toward developing drug cultivars, espe-

cially those with enhanced medical efﬁcacy. Recycling of

public domain “genetics”continues, but sinsemilla

breeders are increasingly reexamining original landraces

and older hybrids to revitalize their breeding programs,

while the quest to collect traditional landrace parents

continues.

The present-day NLD/BLD hybrid gene pool, despite

its recent history of incest, continues to produce novel

allelic combinations. Breeders have recently developed

new sinsemilla cultivars with enhanced efﬁcacy in treat-

ing particular medical indications. The worldwide spread

of BLD genes into traditional NLD regions has nearly

brought an end to pure NLD landraces. In the face of

this degradation of in situ conservation, breeders are

developing new cultivars by growing some of the few

remaining pure NLD and BLD landrace seeds. NLD

landraces are valued in hybrid crosses to impart particu-

lar aromas and ﬂavors or to enhance potency and diver-

sify effects. Traditional NLD landrace varieties imported

from India, Kashmir, Nepal, Indonesia, Korea, Southeast

Asia and Mexico, as well as areas in western, central and

southern Africa, have regained favor because they

mature relatively early, but express fewer of the undesir-

able traits of BLD landraces (e.g., low ﬂower to leaf ratio,

fungal susceptibility, and unpleasant aromas). However,

many of these traditional NLD landrace varieties have

320 R. C. CLARKE AND M. D. MERLIN

become more difﬁcult to procure in recent years.

Breeders only recently learned that Afghan BLD landra-

ces produce CBD (largely bred out of modern hybrid sin-

semilla varieties) and new accessions from Central Asia

are occasionally also introduced. In addition, breeders

are also developing hashish varieties by selecting traits

such as large and easily removed resin glands, dry rather

than sticky or oily resin texture, and aromatic proﬁles

that persist through various processing protocols.

Because commercial-size marijuana shipments do not

often originate in traditionally noncommercial regions,

landrace seeds are usually collected in small numbers by

Figure 17. Cultivar development begins by combining two or more landraces through parental (P) generations to create a diverse

parental population. Select female and male siblings from the parental population are then crossed for several ﬁlial (F) generations to

form a synthetic multi-hybrid. The multi-hybrid seeds are subsequently grown out and female and male plants selected for breeding

are asexually reproduced by rooting vegetative cuttings. At this point, ﬁve different intentional and unintentional “breeding”paths can

be followed. The most common is to cross a female clone (1.4) with a male clone (2.1) to produce single-cross F

hybrid male and female

seed. A related technique using the same parents would be to backcross (BC) a select F

female (1.4 £2.1) to the same male parent

(2.1), sometimes for multiple generations, to produce backcross hybrid male and female seeds. These traditional breeding techniques

have been practiced for decades. The next three twenty-ﬁrst-century pathways rely on female plants to produce pollen. Plant growth

regulators, such as colloidal silver ions, are intentionally applied to female (XX) plants (1.1) to initiate male ﬂowers that will produce via-

ble female (XX) pollen. This pollen can be applied back to the same female clone (1.1) to make selfed (S) (1.1) all-female lines to increase

homozygosity in breeding, or it can be crossed to another female clone (1.2) to make (F

) all-female sowing seed. The accidental ﬁfth

path mimics the previous two, except male ﬂowers infrequently appear spontaneously in “female”clones (1.3), release pollen unnoticed

by the grower, and fertilize the same (1.3) or other clones unintentionally. This is the recent incestuous lineage of many modern-day sin-

semilla cultivars.

CRITICAL REVIEWS IN PLANT SCIENCES 321

travelers, and are relatively rare compared to imported

seeds from the major marijuana producers (Figure 17).

Sinsemilla growers prefer clones with a number of

economically desirable traits: (1) high yield of dry bio-

mass; (2) high proportion of ﬂowers as opposed to leaves

and stems; (3) preponderance of large glandular tri-

chomes; (4) high total cannabinoid content in the ﬂow-

ers; (5) reproducible proﬁle of the target cannabinoids

(e.g., CBD, THC, etc.); and (6) desirable suite of aromatic

terpenes—these are the major traits that breeders

attempt to accentuate. Similar to sinsemilla breeding, the

development of cultivars to supply pharmaceutical raw

materials starts with several promising hybrid crosses,

followed by serial inbreeding of select individuals from

the hybrid population and/or backcrosses to a parental

line. When successful, hybrid vigor is restored by making

crosses between selected inbred lines (de Meijer, 2004).

“Medisins”is a vegetatively reproduced “Skunk No. 1”

high-THC pharmaceutical cultivar registered by Horta-

Pharm in the Netherlands in 1998, and “Grace”is a

high-CBD seed cultivar registered by GW Pharmaceuti-

cals in the United Kingdom in 2004; both of these culti-

vars have been awarded plant breeders rights

(Weightman and Kindred, 2005). Cultivars can also be

developed that produce other cannabinoid compounds

of medical or industrial interest, and secondary metabo-

lite synthesis can be altered to provide enhanced protec-

tion from pests and pathogens.

Developing sterile cuttings of female varieties or seed

varieties that are infertile would guarantee that female

plants will be seedless even when grown near pollen

sources such as hemp seed ﬁelds, and offer an additional

advantage to commercial sinsemilla growers. More culti-

vars will be developed that produce economically

valuable amounts of rare cannabinoids (e.g., cannabi-

chromene (CBC), cannabigerol (CBG), etc.). Cultivars

could also be developed that produce predominately one

aromatic terpenoid compound or suites of varying terpe-

noids in ﬁxed amounts. These new cultivars may prove

to be of pharmaceutical interest, and competition may

develop between plant breeders that will enhance the

production of these rare cannabinoid compounds. In

addition, the bio-engineering manipulation of cannabi-

noid synthases by microorganisms may lead to innova-

tive production systems of these or other rare

cannabinoids. Among the major attractions of sinsemilla

cultivars are their diverse aromas and ﬂavors. However,

breeding for differing parameters may prove difﬁcult

because terpenoid synthesis pathways are controlled by a

suite of genes.

In the future, legal Cannabis enterprises will rely on

the protection of intellectual property rights. Uncommon

and readily apparent biological traits could be added

through breeding or genetic modiﬁcation (GM, see

below) to identify novel cultivars and protect the intellec-

tual property of the breeder. These traits might include

recessive morphological markers such as abnormal pig-

mentation (e.g., yellow stem, purple ﬂowers, or red

leaves); abnormal leaf shapes (e.g., hooked serrations or

webbed leaves); or unique chemical proﬁles (e.g., traces

of a minor unique aromatic terpenes or aldehydes) that

would make the cultivar phenotypically distinct. Genetic

ﬁngerprinting of DNA sequences will also be used to

identify cultivars.

Biotechnology in the form of GM has already

reached Cannabis. Japanese researchers transferred the

THCsynthasegenefromCannabis to cultivated

tobacco (Nicotiana tabacum L.) and induced it to con-

vert CBG (cannabigerol, the precursor molecule to

THC) into THC (Shoyama et al., 2001). German

researchers have engineered a yeast (Pichia)tomake

the same synthetic step and plan to insert more genes

for earlier biosynthetic steps along the pathway to

THC (Zirpel et al., 2015). Other agronomically valu-

able traits may also be transferred to Cannabis such as

enhanced pest resistance, increased yields of medically

valuable compounds, tolerance of environmental

extremes, and sexual sterility.

Recently, breeding lines that are focused primarily on

oleic acid production have been created using gene Tar-

geting Induced Local Lesions in Genomes (TILLING)

techniques; this method allows identiﬁcation of muta-

tions in a speciﬁc gene. Vegetable oils with high oleic

acid content are more stable at room temperature than

those rich in essential fatty acids (EFAs), alpha-linolenic

acid (an omega-3 fatty acid) and linoleic acid (an omega-

6 fatty acid), which are both commonly found in Canna-

bis seed. Bielecka et al.(2014) identiﬁed a suite of alleles

for “putative desaturase genes representing the four

main activities required for production of polyunsatu-

rated fatty acids in hemp seed oil,”and by gene TILLING

a chemically mutagenized population, they were able to

select plants that lacked the desaturase genes. Four back-

crosses and sibling crosses achieved homozygosity.

When grown in the ﬁeld, offspring of this population

produced seeds with a 70% oleic acid oil content and

total oil yields similar to parental lines; this “lays the

foundation for the development of additional novel oil

varieties in this multipurpose low input [Cannabis]

crop”(Bielecka et al., 2014).

The reaction to development and release of GM

organisms has been guarded in general and will be with

Cannabis as well (Russo, 2011). Cannabis presents a par-

ticularly high risk of transmitting GM genes to industrial

hemp crops and weedy Cannabis because it is wind-

pollinated, although so is GM maize that is widely

322 R. C. CLARKE AND M. D. MERLIN

grown. The EU has installed strict regulations to prevent

the accidental release of GM genes, and therefore

production of GM Cannabis in the EU may prove

impractical. However, nonfood industrial ﬁber and phar-

maceutical Cannabis cultivars may not receive as much

resistance from consumers and environmentalists as

food crops. Genes coding for cannabinoid biosynthesis

might also be transferred from Cannabis to organisms

that are less politically sensitive than tobacco such as sin-

gle-cell organisms. Transferring cannabinoid biosynthe-

sis genes from Cannabis into other plants, fungi, and

bacteria opens up the possibility of producing medically

valuable cannabinoids in industrial fermenters and cir-

cumventing Cannabis growing altogether. However,

Cannabis plants already produce extremely high yields

of target compounds. Dried female sinsemilla inﬂores-

cences can yield 20% of THC or CBD, which is very high

for a target compound. With the widespread, recently

increasing legalization of Cannabis cultivation and devel-

opment of high-yielding cultivars for cannabinoid

extraction, genetically engineered microorganisms will

face much competition for market share.

Molecular markers for suites of genes associated with

biosynthesis of individual target compounds as well as desir-

able agronomic traits would allow mass screening of juvenile

populations; this could be of great advantage to breeders (de

Meijer, 2004). Sequencing cannabinoid synthase alleles

would allow marker-assisted breeding for development of

single cannabinoid cultivars and could be especially useful

in recognizing rare cannabinoid variants (e.g., plants with

increased production of cannabigerol (CBG), tetrahydro-

cannabivarin (THCV), or cannabichromene (CBC)) for

pharmaceutical production (Mandolino and Carboni,

2004). Once the entire Cannabis genome is mapped and a

reference genome established, researchers can begin excising

and inserting sequences coding for each trait. Next-genera-

tion genome sequencing technology will allow breeders to

explore the diversity of the Cannabis genome, measure relat-

edness between individuals, identify gene sequences linked

to certain economic traits, screen thousands of individuals

for rare mutations, and use marker-assisted selection to

improve complex traits (Salentijn et al., 2014).

One important insight revealed by recent Cannabis

genome projects is how inbred the drug Cannabis genome

has become—most clones are closely related to all the

others. It will be beneﬁcial to breeders to choose parents that

are less related (i.e., ones with a greater genetic distance

between them) in order to encourage hybrid vigor and

reduce inbreeding depression. Initial studies aimed at foren-

sic identiﬁcation of illicit cannabis concluded that the

genome is highly heterozygous due to its obligate outcross-

ing with most alleles appearing at low frequency. With the

exception of certain major alleles shared by inbred cultivars

(e.g., “Fibrimon”relatives), alleles are shared across a wide

spectrum of phenotypes, and consequently there is a high

degree of variation even within inbred female cultivars.

These are all factors that have so far limited the application

of genome data for taxonomic, evolutionary, forensic, and

plant improvement research (Salentijn et al., 2014). Only

more recently have genetic differences between Cannabis

hemp and drug populations become more clear (Van Bakel

et al., 2011;Sawleret al., 2015).

XII. Summary and conclusions

Cannabis is an annual, sun-loving plant that thrives in open,

nitrogen-rich environments, including rubbish piles created

by humans (Anderson, 1967;Merlin,1972;Clarkeand

Merlin, 2013;alsoseeSmall,2015). Cannabis also has rela-

tively large, numerous, and easily sown seeds, and therefore

was preadapted for cultivation. Close associations between

humans and Cannabis stimulated its early cultivation, and

over time this eventually led to its domestication. As ancient

agricultural strategies took form, humans began to select

plants that provided more and better products. Some of the

earliest seed selections perished and growers must have

constantly collected from spontaneous populations to

replace, supplement, or modify extant varieties, resulting in

evolution of improved landraces. As their familiarity with

Cannabis grew, humans became more discerning with their

seed selections.

Because Cannabis grew well as a camp follower, later

selections were increasingly made from feral populations

near human settlements, rather than truly wild popula-

tions. Early semi-domesticated varieties also became nat-

uralized and interbred with nearby feral and/or wild

populations. Selection for differing economic traits con-

tinued as Cannabis spread well beyond its original puta-

tive range in Central Asia. Backcrosses between

cultivated and truly wild populations became rare,

whereas crosses between cultivated and feral populations

became more common. Eventually, isolation developed

between wild, cultivated, and feral populations which

evolved their own phenotypes. This scenario led to the

extreme variation encountered today in geographically

isolated populations, and accounts for the absence of a

ubiquitous, single spontaneously growing biotype

throughout Eurasia.

Profound evolutionary changes can occur rapidly in

annual plants such as Cannabis, and because it was likely

abundant in its original natural environment, repeated

selection was possible. Under such conditions, early

humans probably did not worry about improving yields,

collecting enough from nearby populations to satisfy

immediate needs. Only along with the advent of Canna-

bis management and eventually cultivation did an

CRITICAL REVIEWS IN PLANT SCIENCES 323

interest arise in increasing yields. As an intimate rela-

tionship between each plant population and the indige-

nous people developed, selection and controlled

breeding began.

The economically valuable traits of Cannabis such as

its strong bast ﬁber, edible seed, and psychoactive resin

are readily noticed and appreciated, and its ﬁrst use was

likely very early in human history in the regions where it

was ﬁrst encountered. The biology and ecology of Can-

nabis were determined by when and where it evolved—

temperate Central Asia during an ancient warm intergla-

cial period being the most likely choice for the origin of

Cannabis, a sun and warmth-loving, water and nutrient

hungry, dioecious, wind-pollinated, short-day annual

plant. Although this genus was undoubtedly useful to

early peoples in its naturally evolved state, human selec-

tion has played the major role in its evolution as a culti-

vated species and allowed Cannabis to thrive almost

everywhere it has been introduced. Selection and breed-

ing of Cannabis are the means by which humans exert

the most control over its evolution as a domesticated

crop plant, and lie at the heart of the human–Cannabis

relationship. Artiﬁcial selection has raised ﬁber content,

increased seed size and yield, and both raised and low-

ered cannabinoid levels. Cannabinoid chemotypes are

characteristic of ultimate geographical origin and usage

groups; and they are correlated with movements of Can-

nabis germplasm by humans to new environments where

they were affected by local cultural selection for particu-

lar products. Cannabis is now grown legally in many

regions of the world.

Presently, few if any imported drug Cannabis ship-

ments are as psychoactively potent as the best sinsemilla

grown in North America and Europe. However, seeds of

improved hybrid cultivars have spread to many Canna-

bis producing nations, and the potency of Cannabis orig-

inating from select locations is steadily increasing.

Because of continuing international pressure against

marijuana growers, and the highly inﬂated price of sinse-

milla, indoor cultivation will maintain popularity world-

wide in any location with a reliable electrical grid.

Instructional information is readily available on the

Internet, and it is easy to install and operate grow rooms

in attics, bedrooms, or basements. Under these crowded

circumstances, there is no tolerance for nonproductive

plants, and the single most productive female clone is

usually selected for all future cultivation. The use of a

single clone improves grow room performance, but pre-

cludes the possibility of seed production. Breeding is no

longer possible and variety improvement ceases entirely.

Vegetative propagation essentially freezes evolution, and

gene diffusion via cuttings has been more localized and

much slower than with seeds. Presently, due to the rapid

spread of indoor cultivation, growers are increasingly

likely to exchange vegetative cuttings of proven clones

rather than carefully bred seeds, and consequently the

genetic diversity in drug Cannabis cultivars has

decreased in the past two decades. As cannabis prohibi-

tion fades, production is rampantly proliferating in exist-

ing agricultural regions, and in the near future only the

most productive and proﬁtable clonal cultivars will be

grown. Cloning will have a lasting effect on sinsemilla

production and will further slow the evolution of drug

Cannabis.

Economic competition will increase with legalization

and normalization, and Cannabis will be further domes-

ticated as just another crop, with high yields and ease of

growing taking precedence over other traits. Although

traditional smoked marijuana will always have its fol-

lowers, and some connoisseur cultivars and their bou-

tique growers will prevail as they have in the wine

industry, the bulk of future production will probably be

extracted for use in edible and vaporized delivery sys-

tems. If target compounds in the future are derived from

a limited number of industrial drug cultivars, Cannabis

genetic diversity will once again suffer.

Present-day clonal sinsemilla cultivars remain geo-

graphically and functionally isolated from gene

exchange with their parental populations in Afghani-

stan or other regions of origin. However, many NLD/

BLD hybrid seeds were produced in Western countries

and some did interbreed with original ancestral popula-

tions. Well-meaning travelers visited many regions

where NLD landraces were still growing and gave mod-

ern hybrid seeds to local farmers who crossed them

with their traditional landraces, hoping for economic

beneﬁt. In the process, these traditional landraces are

becoming contaminated and eventually displaced by

the introduced hybrids. Original pure NLD landraces

have become rare in all traditional marijuana producing

countries(e.g.,Jamaica,Mexico,Morocco,andThai-

land). This situation makes the collection of traditional

landrace varieties increasingly difﬁcult, and conse-

quently their preservation and reproduction are more

signiﬁcant issues than ever before. It is important to

collect landrace building blocks on which future

breeders will depend for creating improved cultivars;

for the sake of genetic diversity, we must escape from

our reliance on clones and their inherent evolutionary

stagnancy. Save and sow your seeds!

References

Agrios, G. N. 2005. Plant Diseases Caused by Fungi in Plant

Pathology. 5th ed., pp. 137, 268, 467–468. Elsevier Aca-

demic, Amsterdam.

324 R. C. CLARKE AND M. D. MERLIN

Anderson, E. 1967. Plants, Man, and Life. (Original edition

1952). University of California Press, Berkeley.

Berenji, J., Sikora, V., Fournier, G., and Beherec, O. 2013.

Genetics and selection of hemp. In: Hemp: Industrial Pro-

duction and Uses, pp. 48–71. Bouloc, P., Allegret, S.,

Arnaud, L., Eds., CABI, Wallingford, UK.

Bielecka,M.,Kaminski,F.,Adams,I.,Poulson,H.,Sloan,R.,Li,Y.,

Larson,T.R.,Winzer,T.,andGraham,I.A.2014.Targeted

mutation of D

and D

desaturase genes in hemp produce

major alterations in seed fatty acid composition including a

high oleic hemp oil. Plant Biotechnol. J. 12(5): 613–623.

B

ocsa, I. 1994. Professor Dr. Ivan B

ocsa, the breeder of Kom-

polti hemp (interview by the JIHA). J. Int. Hemp Assoc. 1

(2): 61–62.

Boyce, S. S. 1900. Hemp (Cannabis sativa). Orange Judd Com-

pany, New York.

Clarke, R. C. 1977. The Botany and Ecology of Cannabis. Pods

Press, Ben Lomond.

Clarke, R. C. 1981. Marijuana Botany: An Advanced Study: The

Propagation and Breeding of Distinctive Cannabis. And/Or

Press, Berkeley.

Clarke, R. C. 1998. Hashish! Red Eye Press, Los Angeles.

Clarke, R. C. 2001. Sinsemilla heritage: what’s in a name? In:

The Cannabible, pp. 1–24. King, J., Ed., Ten Speed Press,

Berkeley.

Clarke, R. C. 1995. Hemp (Cannabis sativa L.) cultivation in

the Tai’an District of Shandong Province, Peoples Republic

of China. J. Int. Hemp Assoc. 2(2): 57, 60–65.

Clarke, R. C. 2007. Traditional Cannabis cultivation in Darch-

ula district, Nepal—seed, resin and textiles. J. Ind. Hemp 12

(2): 19–42.

Clarke, R. C. and Gu, W. 1998. Survey of hemp (Cannabis sat-

iva L.) use by the Hmong (Miao) of the China/Vietnam

border region. J. Int. Hemp Assoc. 5(1): 1, 4–9.

Clarke, R.C. and Merlin, M. D. 2013. Cannabis: Evolution and

Ethnobotany. University of California Press, Los Angeles

and Berkeley.

Clarke, R. C. and D. P. Watson. 2007. Cannabis and natural

Cannabis medicines. In: Marijuana and the Cannabinoids,

pp. 1–17. ElSohly, M. A., Ed., Humana Press, Springer,

New York.

Crossa,J.,Hernandez,C.M.,Bretting,P.,Eberhart,S.A.,and

Taba, S. 1993. Statistical genetic considerations for maintain-

ing germplasm collections. Theor. Appl. Genet. 86:673–678.

Datwyler, S. L., and Weiblen, G. D. 2006. Genetic variation in

and (Cannabis sativa L.) according to ampliﬁed fragment

length polymorphisms. J. Forensic Sci. 51: 371–375.

Faeti, V., Mandolino, G., and Ranalli, P. 1996. Genetic diversity

of Cannabis sativa germplasm based on RAPD markers.

Plant Breed. 115: 367–370.

Fleming, M. P. and Clarke, R. C. 1998. Physical evidence for

the antiquity of Cannabis sativa L. J. Int. Hemp Assoc. 5(2):

80–93.

Forapani, S., Carboni, A., Pioletti, C., Moliterni, V. M. C.,

Ranalli, P., and Mandolino, G. 2001. Comparison of varie-

ties using random ampliﬁed polymorphic DNA markers.

Crop Sci. 41: 1682–1689.

Freinkel, S. 2009. American Chestnut: The Life, Death, and Rebirth

of a Perfect Tree. University of California Press, Berkeley.

Gilmore, S., Peakall, R., and Robertson, J. 2007. Organelle

DNA haplotypes reﬂect crop-use characteristics and

geographic origins of Cannabis sativa.Forensic Sci. Int.

172: 179–190. doi: 10.1016/j.forsciint.2006.10.025

Grigorev, S. V. 2015. Personal communication. Dr. Grigorev is

the curator of the Cannabis germplasm collection at the VIR

in St. Petersburg, Russia.

Harlan, J. R. 1965. The possible role of weed races in the evolu-

tion of cultivated plants. Euphytica 14: 173–176.

Heiser, C. B. 1973. Introgression re-examined. Bot. Rev. 39(4):

347–366.

Hennink, S. 1994. Optimization of breeding for agronomic

traits in ﬁber hemp (Cannabis sativa L.) by study of parent-

offspring relationships. Euphytica 78:69–76.

Hillig, K. W. 2004a. A chemotaxonomic analysis of terpenoid

variation in Cannabis. Biochem. Syst. Ecol. 32: 875–891.

Hillig, K. W. 2004b. A multivariate analysis of allozyme varia-

tion in 93 Cannabis accessions from the VIR Germplasm

collection. J. Ind. Hemp 9(2): 5–22.

Hillig, K. W. 2005a. A Systematic Investigation of Cannabis.

PhD Dissertation. Indiana University, Bloomington.

Hillig, K. W. 2005b. Genetic evidence for speciation in Canna-

bis (Cannabaceae). Genet. Res. Crop Evol. 52(2): 161–180.

Hillig, K. W. and Mahlberg, P. G. 2004. A systematic analysis

of cannabinoid variation in Cannabis (Cannabaceae). Am.

J. Bot. 91: 966–975.

Iltis, H. H. 1983. From teosinte to maize: the catastrophic sex-

ual transmutation. Science 22: 886–894.

Kelly,J.2013.The Graves are Walking: The Great Famine and the

Saga of the Irish People. Henry Holt and Company, New York.

Kojoma, M., Seki, H., Yoshida, S., and Muranaka, T. 2006.

DNA polymorphisms in the tetrahydrocannabinolic acid

(THCA) synthase gene in “drug-type”and “ﬁbertype.”Can-

nabis sativa L. Forensic Sci. Int. 159: 132–140.

Lynch, R. C., Vergara, D., Tittes, S., White, K., Schwartz, C. J.,

Gibbs, M. J., Ruthenburg, T. C., deCesare, K., Land, D. P.,

and Kane, N. C. 2015. Genomic and chemical diversity in

Cannabis. bioRxiv preprint posted online Dec. 13, 2015.

doi: http://dx.doi.org/10.1101/034314

Mandolino, G. and Carboni, A. 2004. Potential of marker-

assisted selection in hemp genetic improvement. Euphytica

140: 107–120.

Mankowska, G., and Silska, G. 2015. Genetic resources of Can-

nabis sativa L. in the collection of the gene bank at

INF&MP in Poznan. J. Nat. Fibers 12: 332–340. doi:

10.1080/15440478.2014.928246

McPartland, J. M. and Guy, G.W. 2004. The evolution of Can-

nabis and coevolution with the cannabinoid receptor –a

hypothesis. In: The Medicinal Uses of Cannabis and Canna-

binoids, pp. 71–101. Guy, G. W., Whittle, G. W., Robson, P.

J., Eds., Pharmaceutical Press, London.

McPartland,J.M.andGuy,G.W.2010. THC synthase in Cannabis

has undergone accelerated evolution and positive selection

pressure. Proceedings of the 20th Annual Symposium on the

Cannabinoids. International Cannabinoid Research Society,p.

43. Research Triangle Park, North Carolina.

McPartland, J. M., Clarke, R. C., and Watson, D. P. 2000.

Hemp Diseases and Pests: Management and Biological Con-

trol. CABI, Wallingford, England.

de Meijer, E. P. M., van der Kamp, H. J., and van Eeuwijk, F. A.

1992. Characterisation of Cannabis accessions with regard

to cannabinoid content in relation to other plant characters.

Euphytica 62: 187–200.

CRITICAL REVIEWS IN PLANT SCIENCES 325

de Meijer, E. P. M. and van Soest, L. J. M. 1992. The CPRO

Cannabis germplasm collection. Euphytica 62: 201–211.

doi: 10.1007/BF00041754

de Meijer, E. P. M. 1995. Fibre hemp cultivars: a survey of ori-

gin, ancestry, availability and brief agronomic characteris-

tics. J. Int. Hemp Assoc. 2(2): 66–73.

de Meijer, E. P. M., Bagatta, M., Carboni, A., Crucitti, P., Moli-

terni, V. M. C., Ranalli, P., and Mandolino, G. 2003. The

inheritance of chemical phenotype in Cannabis sativa L.

Genetics 163: 335–346.

deMeijer,E.P.M.2004.ThebreedingofCannabis cultivars for

pharmaceutical end uses. In: The Medicinal Uses of Cannabis

and Cannabinoids,pp.55–69.Guy,G.W.,Whittle,B.A.,Rob-

son, P. J., Eds., Pharmaceutical Press, London and Chicago.

de Meijer, E. P. M. 2014. The chemical phenotypes (chemo-

types) of Cannabis. In: Handbook of Cannabis, pp. 89–110.

Pertwee, R., Ed., Oxford University Press, Oxford.

Merlin, M. D. 1972. Man and Marijuana: Some Aspects of

Their Ancient Relationship. Fairleigh Dickinson University

Press, Rutherford, NJ.

Millennium Seed Bank. (http://apps.kew.org/seedlist/Seedlist

Servlet) (last accessed 16-02-01).

M€

olleken, H. and Theimer, R. R. 1997. Survey of minor fatty

acids in Cannabis sativ

Cannabis Domestication, Breeding History, Present-day Genetic Diversity, and Future Prospects

Abstract and Figures