Three strains of marijuana – “Moby Dyck,” “Space Queen,” and “Lemon Nigerian” – were cultivated in a commercial indoor growing facility according to the regulations and guidelines issued by Health Canada under an MMPR – Marihuana for Medical Purposes – license. The plants were initiated from rooted cuttings and provided with the nutrient regime for hydroponic culture as described elsewhere (Punja and Rodriguez, 2018). Lighting, temperature and growth conditions were provided in accordance with industry production standards to ensure pistillate inflorescence development (Small, 2017). The plants were first maintained in the vegetative stage for 2 weeks at a temperature range of 23–26°C with a 24 h photoperiod under a light intensity of 120 μmol m^–2 s^–1. They were subsequently transferred into a flowering room for an additional 6–7 weeks to induce pistillate inflorescence development under a modified photoperiod regime that consisted of reduced lighting duration and intensity (10 h photoperiod and 70 μmol ^m–2 s^–1 light intensity) (Figures 1A,B).

The sequential development of the female inflorescence in marijuana strains is illustrated in Figure 1. At the early stages of development, the clusters of pistils with protruding stigmas on terminal inflorescences were yellowish-white in color (Figures 1C–E). In some strains where pigmentation was a characteristic feature, the pistils and surrounding bract tissues developed a red or purple pigmentation (Figures 1F–H). With further maturation of the inflorescence, the stigmas shriveled and developed a brown-red color, while the ovules and surrounding bracts swelled and were covered by glandular trichomes that imparted a silvery-white appearance (Figures 1I–K).

During the 6–7 weeks flowering period of strains grown under commercial conditions, female inflorescences were examined at weekly intervals visually with the aid of a hand lens for the development of male anthers within the inflorescence (Figure 2). Around 1,000 plants in total were examined over the course of two repeated cycles of plant production in the study. Individual clusters of anthers appeared bright yellow and measured 2–3 mm in length (Figures 2A–D) and were formed within the bract tissues and surrounded by stigmas. In some cases, the entire female inflorescence was converted to a mass of anthers which emerged through the bracts (Figure 3). They were carefully removed with a pair of forceps, placed inside plastic petri dishes lined with moistened filter paper, and transported to the laboratory for microscopic examination. Scanning electron microscopic observations of anthers and pollen grains were made following preparation of the samples according to the procedure described by Punja et al. (2019). Pollen grains were released from anthers by suspending them in 5 ml of sterile distilled water for 2 min, from which 50 μl were plated onto 1% water agar (Bacto, Difco) and incubated at 23–25°C for 24–48 h. Percent emergence of pollen tubes was rated from 100 grains examined under an inverted compound light microscope (Zeiss). The experiment was conducted twice. In instances where seed formation was observed in the inflorescences of the three strains, they were collected at fruit maturity, counted, and set aside. Germination of a sample of 20 seeds from each strain was induced by placing them in a moist cocofibre:vermiculite (3:1, v/v) potting medium and incubating at 23–25°C for 20 days under supplemental lighting. The seedling tissues (young leaves), as well as anther tissues from hermaphroditic flowers, were used for DNA extraction (minimum of 15 seedlings per strain) as described below.

To examine the course of development of staminate flowers in genetically male plants, seeds of strains “Moby Dyck,” “Blue Deity,” and “Sweet Durga” were provided by a commercial seed producer which were produced from a controlled cross-fertilization cross. Seeds were placed in a moist cocofibre:vermiculite (3:1, v/v) potting medium and incubated at 23–25°C for 20 days under supplemental lighting as before. At least 20 seeds were germinated and 15 seedlings were obtained for each strain. The seedling tissues were used for DNA extraction as described below. The plants were grown for an additional 4 weeks under a 24 h/day photoperiod until staminate inflorescences were produced. These appeared as characteristic clusters of anthers followed by release of pollen (Figure 6). Individual clusters of anthers were carefully removed with a pair of forceps and brought to the laboratory for microscopic examination for the presence of pollen grains and for DNA extraction. In addition, scanning electron microscopic observations of anther and pollen morphology were made following preparation of the samples according to the procedure described above. To observe pollen grain germination and pollen tube growth on stigmas, pollen was manually collected from male flowers of strain “Sweet Durga” by tapping flowers gently over a piece of wax paper. The pollen was dusted onto pistillate inflorescences of strain “White Rhino” which had been collected the previous day, excised and placed in a humid chamber. A minimum of five replicate samples were included. After 72 h, tissue samples (n = 5) were prepared for scanning electron microscopy and germ tube production and growth on the stigmas that bore receptive papillae was determined at various magnifications from a minimum of 25 images.

DNA was extracted from leaf or anther samples using the Qiagen DNeasy^® Plant Mini Kit (cat. 69104, Hilden, Germany). The Quick-Start Protocol was performed with the following modifications: the tissues were ground in a microfuge tube with liquid nitrogen using a microfuge tube-sized pestle and 50 μl of Buffer AE was used to elute the DNA. Extracted DNA served as the template in PCR reactions using primers SFU1 (5′ GTGACGTAGGTAGAGTTGAA 3′) and SFU2 (5′ GTGACGTAGGCTATGAGAG 3′), the Qiagen Taq PCR Core Kit (cat. 201223), and Taq DNA Polymerase (cat. 201203). The PCR primers were designed from the MADC2 sequence from hemp (GenBank Accession No. JN426768.1) at positions 1–20 and 373–391, as per Mandolino et al., 1999). PCR conditions were as follows: one cycle of 94°C for 3 min, 40 cycles of 94°C for 30 s – 55°C for 45 s – 72°C for 2 min, one cycle of 72°C for 7 min, and hold at 4°C. The primers amplified a 540 bp sized DNA fragment in female plants, while in male plants, either two bands of 390 and 540 bp in size were produced, or just the 390 bp band was amplified (Punja et al., 2017). The 540 bp bands from the female plant samples were purified from their respective PCR reactions using NEB’s Monarch PCR & DNA Cleanup Kit (#T1030S, New England Biolabs, Toronto, Canada). The 540 and 390 bp bands from the male plant samples were separated on a 1% agarose gel then excised and purified using Qiagen’s MinElute^® Gel Extraction Kit (ref. 28604). The purified DNA was sent to Eurofins Genomics for sequencing in both the sense and antisense directions. Sequence files were processed and aligned, and the consensus sequences were extracted using Geneious Prime software by Biomatters Ltd. (Geneious Prime 11.1.5,¹ Auckland, New Zeland). All sequences from the different strains representing female and male plants were aligned using the Geneious Prime multiple alignment function. These sequences were also analyzed for conserved domains using NCBI’s Conserved Domain Database (CDD). In three female strains –Moby Dick, Space Queen and Copenhagen Kush- primers S22645strt (5′CCAATAACCCTCATCCCATTCC3′) and S22645end (5′ATTTCCAAAAGTGTGCGATTCC3′) were used to amplify beyond the region of the female ∼540 bp band. The primers were designed based on a high homology BLAST result between these sequences and Scaffold 22645 of the Purple Kush Genome (canSat3) available on The Cannabis Genome Browser.² Sequence files were again processed and aligned, and the consensus sequences were extracted using Geneious Prime software.

To compare the genetic variation among bands represented by the 540 bp size following PCR, an additional 10 strains of marijuana were chosen. They ranged from landraces (“Jarilla,” “Hoa Bac Silver,” and “Brazil Amazonia”), to autoflower strains (“Northern Lights,” “Acapulco Gold,” and “Snow White”) to commercially grown strains (“CBD Therapy,” “Space Queen,” “Pennywise,” “Girl Scout Cookie,” “Copenhagen Kush”). All of the strains were genetically female, i.e., grown for production of pistillate inflorescences (female phenotype). The plants were initiated from vegetative cuttings and were grown under hydroponic conditions or in the cocofibre:vermiculite mix and provided with the nutrient and lighting conditions for commercial production by a licensed producer as described above. Both leaf tissues and DNA extracted from these strains were stored at −20°C until used. Sequence comparisons were made among the 540 bp size band in female plants of these 10 strains, between the 540 bp band in female and male plants (strains “Moby Dyck,” “Blue Deity,” and “Sweet Durga,” among the 540 bp band in male plants (where present), between the 540 and 390 bp bands in male plants, and among the 390 bp band in male plants of different strains.

ISSR primers UBC 807, 808, 817, 825, 834, and 842 were used to assess the extent of genetic variation. Eight samples each from leaves of strains “Moby Dyck” (ID 1 in Table 3), “Space Queen” (ID 2 in Table 3), and “Lemon Nigerian” (ID 3 in Table 3) (representing the hermaphrodite-derived population of seeds) and eight samples each from strains “Blue Deity” (ID A in Table 3), “Sweet Durga” (ID B in Table 3) and “Healer” (ID C in Table 3) (representing the female:male cross-fertilized population of seeds) were included. PCR amplifications were performed in a volume of 25 μl. Each PCR reaction contained 0.1 μM of ISSR primer, one unit of Taq DNA polymerase, 200 μM of dNTP’s, 1.5 mM MgCl2, 20 ng template DNA, and 1 × PCR buffer. Amplifications were carried out in an Applied Biosystems 2720 thermocycler programmed for 1 cycle at 94°C for 3 min for initial denaturation, followed by 40 cycles at 94°C for 30 s, 55°C for 45 s, and 72°C for 2 min and then a final extension step at 72°C for 7 min (Punja et al., 2017). After amplification, each PCR reaction was subjected to electrophoresis on a 1.5% TAE agarose gel and visualized under UV light. Gels were viewed with a Life Technologies (Thermo Fisher Scientific, Toronto, Canada) E-Gel Imager with UV-light base (cat. 4466611). The sizes of the PCR products were compared with a molecular size standard (1 kb plus) DNA ladder. Only well-separated bands of 0.1–4.0 kb size with high intensity were scored as present or absent for ISSR markers. Data were scored as “1” for the presence and “0” for the absence of DNA bands in each sample. Each set of experiments was repeated to ensure consistency of results. A total of 25 loci were analyzed. These data were used to run statistical analyses in the program POPGENE version 1.32 (Yeh and Boyle, 1997). The observed number of alleles (Na), effective number of alleles (Ne), percent polymorphic loci (PPL), Nei’s (1973) gene diversity (H), Shannon’s index (I), and gene flow estimate (Nm) values were calculated using POPGENE v1.32. Nm was also calculated according to the formula, Nm = (1 - Gst)/4Gst (Slatkin and Barton, 1989). These statistics were calculated between groups (strains) and between populations (hermaphroditic and cross-fertilized). Hardy-Weinberg equilibrium and random mating were assumed for both hermaphroditic and cross-fertilized populations. All C. sativa samples were assumed to be independent.

The production system for marijuana plants is based on vegetatively propagated plants that are first grown under a 24 h photoperiod for 4 weeks and then switched to a 12–14 h dark:10–12 h light regime. The plants in Figure 1A have just been “flipped” to the reduced lighting regime. Figure 1B shows development of large terminal inflorescence clusters in some strains, e.g., “Hash Plant” that extend to a 1 m height above the leaf canopy. The sequential development of the female inflorescence in several marijuana strains is shown in Figures 1C–K. At the early onset of flower development (weeks 1–2 of the flowering period), young terminal inflorescences developed white hair-like stigmas (Figure 1C). In subsequent weeks 3–4, development of yellowish-white clusters of stigmas which were bifurcate at the tips can be seen (Figures 1D,E). This stage was the most receptive to pollination (authors, unpublished observations). In red and anthocyanin-accumulating strains, stigma development was similar over this time period, and at advanced stages of inflorescence development, the yellowish-white clusters of stigmas were accompanied by red or purple pigmentation in the style tissues or subtending bracts (Figures 1F–H). Maturation of the inflorescence (weeks 5–6 of the flowering period) was characterized by the curling and browning/reddening of the stigmas and swelling of the carpels that occurred in the flowering period (Figures 1I,J). The mature inflorescence close to harvest (weeks 7–8) with collapsed stigmas and swollen carpels is shown in Figure 1K).

Female inflorescences of three marijuana strains grown under commercial conditions were visually examined at weekly intervals. Beginning around week 4 of the flowering period, the appearance of individual anthers or clusters of anthers within the bract tissues adjacent to the stigmas was observed in hermaphroditic flowers at a frequency of 5–10% of the plants examined (Figures 2A–D). The anthers were visible in weeks 4–7 of the flowering period and were present until harvest. In rare instances, the entire female inflorescence was converted to large numbers of clusters of anthers (Figure 3). Scanning electron microscopic examination of the stigmas that were present in hermaphroditic flowers showed the papillae (stigmatic hairs) (Figure 4A), which in mature inflorescences originated from a central core (Figure 4B). Individual anthers that were produced in hermaphroditic inflorescences were shown to consist of an outer wall (epidermis and endothecium) with a longitudinal groove (stomium) (Figure 4C) which, upon maturity, expanded and dehisced to release pollen grains (Figure 4D). Bulbous structures presumed to be trichomes were also observed forming along the stomium of the anther (Figure 4E). When viewed under the light microscope, the anther wall and stomium could be seen and pollen grains were released into the water used to mount the sample (Figures 5A–C). Some pollen grains had collapsed when viewed under the scanning electron microscope (Figure 5D). Pollen germination was observed within 48–72 h on water agar and ranged from 10 to 30% (Figure 8A).

In genetically male plants, anthers were produced within clusters of staminate flowers that developed at leaf axils (Figures 6A–C) at around 4 weeks of age. At flower maturity in weeks 4–6, anthers dangled from individual flowers and were observed to release large amounts of pollen grains, which were deposited in yellow masses on the leaves below (Figures 6D–F, 7). Such prolific release of pollen was not observed from the hermaphrodite flowers. Scanning electron microscopic examination of the anthers produced on staminate plants showed the release of pollen grains (Figure 6G). Along the longitudinal groove or stomium, the formation of a line of bulbous trichomes (Figure 6H) that developed on a short pedicel (Figure 6I) was observed, similar to that seen in hermaphroditic flowers. When pollen from male plants was deposited onto female inflorescences (Figure 8B) and viewed at 72–96 h, various stages of pollen germination and germ tube development were observed (Figures 8C–F).

Within the hermaphroditic inflorescences in which anthers were found, seed set was initiated, and mature seeds were observed prior to the harvest period (Figures 9A,B). From each of 3 inflorescences bearing seeds, a total of 34, 48, and 22 seeds were obtained. The seeds were removed and placed in moist potting medium where they germinated at a rate of 90–95% within 10–14 days to produce seedlings (Figures 9C,D).

PCR analysis was used to identify specific bands which correlated with the male or female phenotype in commercial marijuana strains. Seedling tissues from strains “Moby Dyck” and “Blue Deity,” produced through cross-fertilization by a commercial seed producer, showed a band size of approximately 540 bp in female plants, while two bands (ca. 540 and 390 bp in size) or one band (390 bp), were observed in male plants. The resulting ratio of male:female plants in seeds derived from these latter strains was 5:7 and 9:5, respectively (Figures 10A,B). In a third strain “Healer,” however, which were seeds obtained from outdoor cultivation of marijuana, only two male plants were identified among 16 plants; the remaining 14 were female (Figure 10C). By comparison, seeds obtained from hermaphroditic inflorescences of strains “Moby Dyck” and “Space Queen” yielded seedlings that all showed the 540 bp band size corresponding to the female phenotype (Figures 10D,E); the male-specific 390 bp band was absent. PCR analysis of DNA isolated from anther tissues (A) from hermaphroditic plants showed that the banding pattern was of the 540 bp band (Figure 10F). In contrast, the banding pattern observed in staminate flower tissues showed the 390 bp band (data not shown).

Sequence comparisons were made using Geneious Prime among the 540 bp size band in female plants of 10 different marijuana strains, between the 540 bp band in female and male plants, among the 540 bp band in male plants (where present), between the 540 and 390 bp bands in male plants, and among the 390 bp band in male plants of different strains. The 540 bp band in female plants showed sequence similarities between 89.4 and 100% among different strains (Table 1). Strain “Girl Scout Cookie” showed the lowest sequence similarity (89.4–94.1%) to all of the other strains (Table 1), particularly to the autoflower strains “Northern Lights” and “Snow White.” Strain “Moby Dyck” showed 99.3% sequence similarity to “Copenhagen Kush” (Table 1). These variations in sequences could be due either to sequence divergence over time (SNPs) or base-calling errors introduced during sequencing. In comparing the sequences of the 540 bp band present in female and male plants, the range of sequence similarity was 87.3–98.1%, with strain “GSC” showing the lowest similarity (87.3–90.6%) to the male 540 bp band sequences of three strains that were included in this study (Table 2). The 540 bp band in male plants showed 95.1–97% similarity to each other while the 390 bp band sequence in male plants had 97.5–100.0% sequence similarity (data not shown). In comparing the 540 and 390 bp bands in male plants, two regions of about 176 bp were absent within the 390 bp band, suggesting an internal deletion had occurred (Figure 11). The 540 bp band had 95% sequence similarity with a SINE MADC2 sequence (GenBank Accession No. JN426768.1) and 86% similarity with a MADC2 male-specific sequence (GenBank Accession No. JF298280.1). The 390 bp band had 100% sequence homology in the aligned overlapping region with JF298280.1 and 88% sequence homology with JN426768.1. The sequence variations among the 540 bp band present in different female strains is likely due to the presence of SNP’s detected in these strains.

Conserved domain analysis of bands originating from female and male plants indicated the presence of an rve Superfamily integrase core domain (pfam 00665 present in the 540 bp band and pfam cl121549 in the 390 bp band size). When primers S22645strt and S22645end were used to amplify beyond the region of the female 540 bp band in three strains, a pre-integrase GAG domain (pfam13976) was found upstream of the rve Superfamily integrase core domain (Figure 12). Both domains are potential features of Class I LTR retrotransposons.

The six-primer set revealed a range of polymorphic bands within the populations of plants originating from hermaphroditic and cross-fertilized seeds. The percentage of polymorphic loci ranged from 44 to 64% for the hermaphroditic group of plants (n = 24) and 60–72% for the cross-fertilized group of plants (n = 24) from a total of 25 bands scored. A representative ISSR banding pattern is shown in Supplementary Figure S1. The range and mean values of statistical measures derived from data analysis for the hermaphroditic and cross-fertilized populations are presented in Table 3. To estimate the genetic variation between cross-fertilized and hermaphroditic populations and between the groups (strains) within these populations, the effective number of alleles (Ne), Nei’s (1973) gene diversity (H) and Shannon’s index values were calculated (Table 3). Hardy-Weinberg equilibrium, independence, and random mating were assumed for both hermaphroditic and cross-fertilized populations. Comparisons among the cross-fertilized groups (A,B, and C in Table 3) show overlapping mean values and standard deviations for their Ne, H, and I values. The Nei’s gene diversity (H) values for groups A, B, and C were 1.6 ± 0.5, 1.6 ± 0.5, and 1.72 ± 0.4583, respectively. This suggests that there was no observable difference in the level of genetic variation between the three cross-fertilized groups. The same observation can be made when comparing the three hermaphroditic groups (1, 2, and 3 in Table 3). With regard to mean values of the two populations, the hermaphroditic population did not differ in values of Na, Ne, H, and I when compared to the cross-fertilized population. This was confirmed by a Student’s t-test, in which all p-values were > 0.05 and therefore not significant (Table 3).

To determine if hermaphroditic populations as a whole are as genetically variable as cross-fertilized populations, the effective number of alleles (Ne), percent polymorphic loci (PPL), Nei’s (1973) gene diversity (H), and Shannon’s index values were calculated and compared (Table 3). The data showed no difference between these values (Table 3). The cross-fertilized populations had a combined Ne value of 1.7553 ± 0.2655 and the hermaphroditic populations had a combined value of 1.5799 ± 0.3403. In addition, the cross-fertilized populations had an H value of 0.4138 ± 0.1162 and the hermaphroditic populations had an H value of 0.3330 ± 0.1675. These values are not significantly different from each other. The results indicate there were no measurable differences in the population statistics used to assess genetic variation among plants after one cycle of self-fertilization as compared to cross-fertilization.