Tetraploid F2 hemp (Cannabis sativa) plants of genotype XXXX, XXXY, and XXYY. Scale bars = 80 cm.
Tetraploid hemp (Cannabis sativa) plant of genotype XXXY. (A) Whole plant. (B–D) Male flowering nodes. Scale bar = 80 cm.
Results of polymerase chain reaction allelic competitive extension assays for sex chromosome genotypes. Relative fluorescent units (RFUs) of FAM and HEX showing allelic discrimination of XXXX, XXYY, and XXXY sex chromosome genotypes among hemp (Cannabis sativa) F2 progeny from tetraploid F1 seed parents XXXX (A), XXXY (B), and control diploid XX (‘TJ’s CBD’ and ‘FL58’) and XY (‘Picolo’) plants. NTC = nontemplate control.
Three nodes bearing male flowers from triploid predominantly female hemp (Cannabis sativa) plants.
Click on author name to view affiliation information
Hemp (Cannabis sativa) is naturally diploid and dioecious with 1:1 female-to-male progeny. However, female-biased seed is desired for dual-purpose fiber and grain production to maximize yield. Based on investigations from the early 1940s, tetraploidy may be a way to produce female-biased seed. We evaluated sex phenotype ratios and genotype ratios by polymerase chain reaction allelic competitive extension using CSP-2 primers through F1 and F2 generations to ascertain the level of female bias from tetraploids. F1 progeny (n = 102) from crossing XXXX × XXYY had a phenotypic ratio of 7.5:1 female/predominantly female plants-to-male plants. Plants scored as predominantly female were very similar visually to those scored as females and had an equivalent number of flowering nodes, but they produced a small number (< 10) of male flowers that occurred on a small number of lateral nodes (two to five, with less than five flowers per node). F1 plants were allowed to cross-pollinate in the greenhouse, and F2 seed from one female and one predominantly female plant, with confirmed genotypes of XXXX and XXXY, respectively, were evaluated. Inheritance patterns for F2 progeny (n = 51) from XXXX × XXYY indicated preferential chromosome pairing behavior during meiosis for XY gametes, which resulted in 84% XXXY plants. Inheritance patterns for F2 progeny (n = 51) from XXXY × XXYY did not match those expected for tetrasomic segregation and were intermediate, with a ratio of 3:1 female/predominantly female plants-to-male plants. Genotype ratios fit a model in which XY ova fertilized by XY or YY sperm were inviable, possibly resulting from arrested endosperm development when the X:Y balance was ≤ 1. A small degree of self-pollinating of predominantly female F1 plants may have affected F2 inheritance patterns. Tetraploidy may be a useful approach for breeding female-biased seed resulting from a strong X-to-autosome balance system in hemp. Because the predominant F1 genotype from the XXXX × XXYY cross is XXXY, a female bias of 3:1 and as high as 5:1 could be expected for market-ready F2 seed, which is a significant shift from 1:1 for diploid hemp.
Hemp (Cannabis sativa) is a diploid (2x = 20) and dioecious species that produces offspring in a 1:1 female-to-male ratio (Warmke and Davidson 1944). The plant is grown for fiber, grain, and/or flower (Smart et al. 2023). Cultivation for both fiber and grain are referred to as dual-purpose production. Male plants contribute little to yield because they senesce before harvest, and they consume farmer inputs, thus reducing the efficiency and sustainability of the crop. Certified monoecious cultivars exist that generally produce a mix of monoecious and female plants but may also produce some male plants (Darby and Sullivan 2024a; Darby et al. 2022, 2023). Monoecious cultivars may or may not yield more than dioecious cultivars for dual-purpose production, but they are easier to harvest because of the lack of senescing male stems, which can clog harvesting equipment. The number of monoecious, female, and male plants per crop is affected by the breeding strategy used by the seed producer and/or the occurrence of uncontrolled pollination between seed generations (Darby and Sullivan 2024a; Smart et al. 2023). The degree of monoecy for diploid hemp plants can vary and is described by the Sengbusch scale, which ranges from first to fifth degree, where the first degree has 80% to 90% male flowers and the fifth degree has less than 10% male flowers (Bocsa and Karus 1996).
Despite the benefits provided by monoecious cultivars, there are drawbacks to monoecy (Bocsa and Karus 1996; Truta et al. 2007). Monoecious plants produce smaller grain as a result of self-pollination and do not necessarily yield more on a per-hectare basis (Forapani et al. 2001; Smart et al. 2023; Wortmann 2020). Historically, the monoecious trait is described as unstable and difficult to maintain without human intervention (Bocsa and Karus 1996; Salentijn et al. 2015). During seed production fully male plants and monoecious plants that are primarily male must be rogued, which is labor and resource intensive. This is likely a result of the presence of low numbers of XY males in the population, which produce far more pollen than the monoecious plants and thus can quickly restore the genotypic frequency of XY plants in the population through open pollination. We have confirmed the presence of XY individuals in a European monoecious cultivar using molecular markers that can distinguish XX and XY individuals (unpublished data). Recently developed dioecious cultivars for dual-purpose production produced yields comparable with existing monoecious cultivars in some university trials (Cornell Hemp 2024; Darby and Sullivan 2024a, 2024b; Darby et al. 2022, 2023). Although there have been advancements with dioecious seed, more emphasis must be placed on breeding female-biased populations to enhance yield (Zahl et al. 2024).
Tetraploidy has improved performance for crops including alfalfa (Medicago sativa), leek (Allium ampeloprasum), potato (Solanum tuberosum), rye (Secale cereale), and others (Sattler et al. 2016). Tetraploid plants are often vigorous and have larger flowers, fruit, seeds, and/or stems compared with their diploid counterparts. The impacts of tetraploidy for hemp have not been widely studied. Tetraploid and triploid hemp plants have been developed and exhibit larger leaves, stems, inflorescences, foliar stomata, and/or seeds than diploids (Bagheri and Mansouri 2015; Crawford et al. 2021; Kurtz et al. 2020a, 2024; Parsons et al. 2019). Tetraploid plants can be completely fertile whereas triploid plants are nearly sterile (Crawford et al. 2021; Kurtz et al. 2024; Ranalli 2004; Suchoff et al. 2024).
Warmke and Davidson (1944) crossed tetraploid female (XXXX) and male (XXYY) hemp plants and found a 7.5:1 female and “female–hermaphrodite”-to-male ratio among the F1 offspring (n = 94). They applied the term female–hermaphrodite to plants described as XXXY and “not sharply distinguishable” from female plants. We presume these plants produced a small number of male and/or intersex flowers; however, this information is not provided. Furthermore, the genetic background or origin of the tetraploid parents used in the cross is not provided. The early investigations of Warmke and Davidson (1944) suggest that tetraploidy may be an alternative breeding approach for developing female phenotype–dominant populations.
The objectives of our research were to cross XXXX × XXYY individuals and evaluate the sex phenotypes and genotypes of the F1 and F2 progeny. If tetraploidy can deliver female-biased populations, then it may warrant the development and evaluation of tetraploid cultivars. The results of this research will contribute to a greater understanding of sex determination in hemp.
Three female tetraploid ‘Kentucky Sunshine’ hemp plants from S1 seed were pollinated by a tetraploid male plant of ‘Kentucky Sunshine’ S1 × (‘Candida’ × ‘Wife’) also from seed. Tetraploid status of parental plants was confirmed by flow cytometry. Crosses were conducted from Apr to Jul 2024 in a growth chamber with a set point of 24 °C and a 12-h photoperiod. A pooled subsample of 225 F1 seeds was formed by combining 75 F1 seed per the tetraploid ‘Kentucky Sunshine’ mother plant. F1 seeds were stored dry in a plastic bag in a refrigerator at 4 °C until use.
On 18 Oct 2024, pooled F1 seed was sown in 50-plug trays and placed in a greenhouse with set points of 15.6/21.2 °C heating/cooling and an 18-h photoperiod provided by supplemental lighting using 1000-W high-pressure sodium lamps (Phantom HPS 1000 W; Hydrofarm, Petaluma CA, USA) and blackout curtains. On 1 Nov 2024, 102 seedlings were selected at random, potted in 2.78-L containers filled with peat-based medium (Metro-Mix 830; Sun Gro Horticulture, Arawak, MA, USA), and top-dressed with 5 g 15N–3.9P–10K controlled-release fertilizer (Osmocote Plus 15-9-12, 3- to 4-month formulation; Everris NA, Dublin, OH, USA). Potted plants were kept in the greenhouse and irrigated as needed by drip irrigation. Plants were fertigated once weekly with 20N–4.3P–16.6K water-soluble fertilizer (Peters 20-10-20; Scotts, Marysville, OH, USA) at 100 ppm N. On 12 Nov 2024, the photoperiod was reduced to 11 h to induce flowering. Beginning 6 Dec 2024, plants received fertilizer at every irrigation at the same rate. The experimental unit was one potted plant. Time to terminal flowering was recorded when a minimum of two pistils bearing stigmas were visible at the apical meristem. Time to anthesis was recorded when pollen release was first evident. Data were collected from 10 to 12 Dec 2024 on plant height, measured from the base to the top of the main shoot, and the number of flowering nodes per plant. The number of plants exhibiting the male, female, or female–hermaphrodite phenotype was also counted. From here on we refer to the female–hermaphrodite phenotype as predominantly female phenotype, because this is the current terminology (Shephard et al. 1999). For predominantly female plants, the number of nodes bearing male flowers and the number of male and female flowers per male flower-bearing node were counted. For male plants, the number of nodes bearing stigmas per plant were counted. The percentage of nodes bearing male flowers and female flowers per plant, and the percentage of male flowers per male flower-bearing node for predominantly female plants were calculated. Leaf samples for 15 F1 plants—five exhibiting the male phenotype, five female, and five predominantly female—were collected for sex genotyping as described in the next section.
F1 plants were allowed to open pollinate in the greenhouse. Pollen dispersal was facilitated by shaking male plants by hand daily during anthesis. On 31 Jan 2025, F2 seed was harvested from several female and predominantly female plants. F2 seed was not pooled. On 13 Feb 2025, F2 seed from one F1 plant confirmed to have the genotype XXXX and a second F1 plant confirmed to have the genotype XXXY were sown in 50-plug trays and set in a greenhouse with an 18-h photoperiod as described. On 27 Feb 2025, 51 seedlings per F1 seed plant, selected at random, were potted to 2.78-L containers as described. The experimental unit was one potted plant, and units were arranged in the greenhouse in a completely randomized design. Plants were irrigated and fertigated weekly as described. On 10 Mar 2025, the photoperiod was reduced to 11 h to induce flowering. Beginning on 26 Mar 2025, plants received fertilizer at every irrigation. Time to terminal flowering and time to anthesis were recorded as described. Data were collected as described from 7 to 9 Apr 2025. Leaf samples for all 102 F2 plants were collected for sex genotyping as described in the next section.
Sex chromosome allelic discrimination was conducted for 15 F1 and 102 F2 plants using DNA extracted from leaf tissue. DNA was isolated using a 96-well filter plate (AcroPrep Advance; Cyril, Marlborough, MA, USA) and the modified Cetyltrimethylammonium bromide method (Toth et al. 2020) or the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Genotyping by polymerase chain reaction (PCR) allelic competitive extension (PACE) (3CR Bioscience, Harlow, UK) was performed according to the manufacturer’s protocol, with the addition of five thermal cycles using a real-time PCR thermocycler (CFX Opus; Bio-Rad, Hercules, CA, USA). Data were analyzed using CFX Maestro version 1.1 software (Bio-Rad). The sex-specific CSP-2 primers, designed to be X chromosome specific for FAM and Y chromosome specific for HEX, were used for PACE (Quade et al. 2025; Toth 2022). The dioecious female hemp cultivars TJ’s CBD and FL58, and a dioecious male plant of the cultivar Picolo, were used as control genotypes XX and XY, respectively.
Phenotype data were subjected to the nonparametric Kruskal-Wallis test (P ≤ 0.05) using statistical software (SAS version 9.4; SAS Institute, Cary, NC, USA). Phenotype and genotype count data were subjected to the χ2 test (P ≤ 0.05) using computer software (Microsoft Excel for Microsoft 365 MSO version 2504; Microsoft Corporation, Redmond, WA, USA).
The numbers of F1 tetraploid plants exhibiting male, female, or predominantly female phenotypes were 12, 47, and 43, respectively (Table 1). Female and predominantly female plants produced an equivalent number of flowering nodes and were visually similar (Table 1; Fig. 1). Terminal flowering was reached 1 to 2 d earlier for predominantly female plants than female plants (Table 1). Predominantly female plants produced a small number (< 10) of male flowers per plant, which occurred at a small number of nodes (between two and five nodes). Male flower-bearing clusters were small (fewer than five flowers per node) and consisted of both male and female flowers, but ≥ 50% were male flowers (Table 1; Fig. 2). The predominantly female phenotype observed in our study was likely analogous to the female–hermaphrodite phenotype of Warmke and Davidson (1944). Predominantly female tetraploids might be considered fifth-degree plants on the Sengbusch scale for monoecy in diploids (Bocsa and Karus 1996). The ratio of female plus predominantly female plants to male plants observed for the F1 population was 90:12 (or 7.5:1), which corroborates the findings of Warmke and Davidson (1944) for tetraploid hemp and strongly deviates from the 1:1 ratio for diploid hemp (Table 2).
Citation: J. Amer. Soc. Hort. Sci. 150, 5; 10.21273/JASHS05523-25
Citation: J. Amer. Soc. Hort. Sci. 150, 5; 10.21273/JASHS05523-25
Tetraploid male plants were taller and produced more flowering nodes than female and predominantly female plants, and reached anthesis earlier than predominantly female plants (Table 1; Fig. 1). Similarly, diploid male plants are taller and flower earlier than diploid female plants (Flajšman and Ačko 2022). Most male F1 plants produced some stigmas, primarily on nodal clusters at or near the shoot apices. This was possibly stimulated by the accumulation of ethylene, produced by female and predominantly female plants, in the greenhouse. Hormones such as ethylene and hormone antagonist chemicals can influence sex expression in hemp and are used by growers for propagation and breeding (Galoch 1978; Mohan Ram and Sett 1982). F1 plants were induced to flower in December, a time of year in Connecticut when greenhouse venting, which would dissipate ethylene buildup, was not occurring. The parent plants used to produce the F1 population were derived from the cultivars Kentucky Sunshine, Candida, and/or Wife, all of which are easily masculinized with silver thiosulfate (an ethylene antagonist) and therefore highly responsive to ethylene concentration (Kurtz et al. 2020b, 2024; Lubell and Brand 2018; Mohan Ram and Sett 1982). Plants of ‘Kentucky Sunshine’ and ‘Candida’ from S1 feminized seed and ‘Wife’ from S1 to S4 feminized seed did not demonstrate monoecy (Kurtz et al. 2020b; McDonald and Lubell-Brand 2024), a dominant trait (Garcia-de Heer et al. 2024; Truta et al. 2007). Therefore, it is unlikely that monoecious genes influenced flowering.
Genotype analysis of 15 F1 plants of known phenotype (five each of male, female, and predominantly female) revealed there to be five XXYY, one XXXX, eight XXXY, and one undetermined, which may be the result of a PCR error or aneuploidy. An analysis of pollen mother cells from tetraploid XXYY melandrium (Melandrium dioicum) showed that 6% exhibited nondisjunction, which results in aneuploid gametes (Warmke and Blakeslee 1940).
Phenotype and genotype ratios for F2 progeny from both the XXXX × XXYY and XXXY × XXYY crosses differed significantly from those expected for tetrasomic inheritance from random assortment (Table 3) (Muthoni et al. 2015). The F2 progeny from the XXXX × XXYY cross produced 2 female phenotype plants, 45 predominantly female phenotype plants, and 4 male phenotype plants, but 3 were genotype XXXX, 43 were XXXY, and 5 were XXYY (Table 3; Fig. 3). The F2 progeny from the XXXY × XXYY cross produced 8 female phenotype plants, 31 predominantly female phenotype plants, and 12 male phenotype plants, but 10 were genotype XXXX, 28 were XXXY, and 13 were XXYY. Marker analysis matched genotype correctly with phenotype for 88% of the F2 plants. Of the 12 F2 plants with genotypes that did not match phenotype, six were XXXX but predominantly female, two were XXYY but predominantly female, two were XXXY but female, one was XXXY but male, and one was XXYY but female. These discrepancies may represent plasticity in sex expression resulting from environmental effects such as roots constrained in relatively small containers, fluctuating ethylene concentration in the greenhouse, a short photoperiod of only 11 h, and/or rapid growth in response to short days for plants that were relatively small (15–23 cm tall) and young (25 d post seed sowing), when photoperiod was reduced for flowering (Smart et al. 2023; Truta et al. 2007). The one XXYY plant that produced only female flowers may represent a sampling or PCR error or is aneuploid.
Citation: J. Amer. Soc. Hort. Sci. 150, 5; 10.21273/JASHS05523-25
F2 progeny from the XXXX × XXYY cross exhibited a greater proportion of predominantly female phenotype plants than F1 plants from the same cross of sex chromosomes (Table 2). F2 predominantly female plants produced twice as many nodes bearing male flowers and three times as many male flowers as F1 predominantly female plants (Table 1). Because F2 plants were induced to flower in March, when the greenhouse was venting and ethylene was dissipating, these findings may be attributable to less ethylene suppression of male flower production for F2 plants than for F1 plants. Lower levels of ethylene in the greenhouse during the F2 study could also explain why fewer F2, than F1, male plants developed stigmas (Table 1). We suspect that, among the F1 progeny, there was possibly a greater number of predominantly female plants than the phenotype counts indicate as a result of ethylene suppression of male flowering nodes.
We have provided a description for and genetic confirmation of hemp XXXY, first identified based on phenotype by Warmke and Davidson (1944) and referred to as female–hermaphrodite. Hemp plants of XXXY genotype are predominantly female, with > 92% all-female nodes (Table 1). Tetraploid hemp showed female bias with F1 and F2 progeny from the XXXX × XXYY cross that fit the 7.5:1 ratio described by Warmke and Davidson (1944), and F2 progeny from the XXXY × XXYY cross that fit a 3:1 ratio.
The inheritance pattern for the XXXX × XXYY cross indicates preferential pairing at meiosis I for XY gametes from the XXYY parent, which resulted in 84% XXXY progeny (Table 3; Fig. 3). Similar rates of XY gamete production from XXYY were reported for melandrium and acnida (Acnida tamariscina), at 89% and 85%, respectively (Murray 1940; Warmke and Blakeslee 1940). Cytological observations of pollen mother cells from XXYY melandrium distinguished seven different quadrivalent associations; however, only two of them occurred for a combined 82% of cases studied, and these produced XY gametes (Warmke and Blakeslee 1940). In melandrium, XX and XY gametes were produced in small quantities, resulting in only 3% XXXX and 2% XXYY progeny (Warmke and Blakeslee 1940). Similar results were reported for acnida (Murray 1940). We also found a low occurrence of genotypes XXXX and XXYY for hemp (Table 3).
The inheritance pattern for the XXXY × XXYY cross fits a model in which XY ova fertilized with XY or YY sperm results in embryo failure (Table 3). Embryo failure can occur when the ratio of X to Y for the endosperm is ≤ 1 (Janoušek et al. 1998). For a normal diploid XX seed, this ratio is 3X:0Y; for an XY seed, it is 2X:1Y. For our F2 genotypes from the XXXY × XXYY cross, ratios are 6X:0Y for XXXX, 5X:1Y for XXXY, and 2X:1Y for XXYY. In all these cases, the X to Y ratio is > 1, because the dose of X exceeds that of Y. The ratio of X to Y endosperm for an XY ovum fertilized by XY sperm would be 1X:1Y; YY sperm would be 1X:2Y. Arrested endosperm development caused by paternal excess has been described for the diploid × tetraploid cross in hemp and it results in triploid block (Kurtz et al. 2024).
In a scenario in which XY ova are inviable, we might have expected to find F2 inheritance patterns for the XXXY × XXYY cross like those for the XXXX × XXYY cross, but instead there were slightly more XXXX and XXYY genotypes relative to XXXY (Table 3; Fig. 3). It is possible that the XXXY genotype exhibits intermediate, between disomic and tetrasomic, chromatid segregation (Stift et al. 2008). Intermediate to highly preferential pairing for autosomal loci or linkage groups has been reported for tetraploid yellow cress (Rorippa hybrid), osage orange (Maclura pomifera), Moncada mandarin (Citrus hybrid ‘Moncada’), and bramble [Rubus sp. (Crane and Darlington 1932; Garavello et al. 2020; Laushman et al. 1996; Stift et al. 2008)]. To a small degree, our F2 inheritance patterns may have been influenced by selfing or cross-pollination among F1 predominantly female hemp plants.
Our findings support an X-to-autosome balance system for sex determination in hemp, whereby the ratio of X chromosomes to the autosomes is a primary indicator of sex (Akagi et al. 2025; Ming et al. 2011). The X-to-autosome balance for XXXX is 1 (1X:1A), and for XXYY is 0.5 (1X:2A). The same ratios exist for diploid XX and XY hemp, respectively. Genotype XXXY has the ratio 0.75 (3X:4A), which is halfway between 1 and 0.5; however, its phenotype is predominantly female. In hop (Humulus sp.), the closest relative of hemp, XXXY is 50% male and triploid XXY is predominantly male (Parker and Clark 1991; Shephard et al. 1999).
Additional evidence for the X-to-autosome balance system in hemp came from a preliminary greenhouse grow that included 26 XXXY and 2 XXY hemp plants (unpublished data). Both XXXY and XXY exhibited the predominantly female phenotype, but XXY had slightly stronger male traits than XXXY. The XXXY plants grew to 148 cm and had three nodes bearing male flowers, with only one to three flowers per male node. The XXY plants were 167 cm tall, had ∼25 nodes bearing male flowers, and nodes had ≥ 10 flowers per nodal cluster (Fig. 4). Hemp genotype XXY has an X-to-autosome balance of 0.66, which is intermediate, between that for XXXY and XXYY.
Citation: J. Amer. Soc. Hort. Sci. 150, 5; 10.21273/JASHS05523-25
These findings demonstrate that tetraploidy may be a useful breeding approach to produce female-biased hemp populations from seed. In monoecious hemp seed production, suppliers frequently sell the open-pollinated F2 seed, instead of F1 seed, to meet commercial production quotas (Berenji et al. 2013; Flajšman and Ačko 2022). A similar approach would likely be needed for tetraploid seed production. The predominant genotype from the XXXX × XXYY cross is XXXY; therefore, a female bias of 3:1 and as high as 5:1 could be expected for market-ready F2 seed, which is a significant shift from 1:1 for diploid seed. Although these results reveal tetraploidy as a viable method for producing female-biased hemp seed, the quality and yield of tetraploid grain and fiber would need to be evaluated.
Tetraploid F2 hemp (Cannabis sativa) plants of genotype XXXX, XXXY, and XXYY. Scale bars = 80 cm.
Tetraploid hemp (Cannabis sativa) plant of genotype XXXY. (A) Whole plant. (B–D) Male flowering nodes. Scale bar = 80 cm.
Results of polymerase chain reaction allelic competitive extension assays for sex chromosome genotypes. Relative fluorescent units (RFUs) of FAM and HEX showing allelic discrimination of XXXX, XXYY, and XXXY sex chromosome genotypes among hemp (Cannabis sativa) F2 progeny from tetraploid F1 seed parents XXXX (A), XXXY (B), and control diploid XX (‘TJ’s CBD’ and ‘FL58’) and XY (‘Picolo’) plants. NTC = nontemplate control.
Three nodes bearing male flowers from triploid predominantly female hemp (Cannabis sativa) plants.
Contributor Notes
J.D.L. is the corresponding author. E-mail: jessica.lubell@uconn.edu.
Tetraploid F2 hemp (Cannabis sativa) plants of genotype XXXX, XXXY, and XXYY. Scale bars = 80 cm.
Tetraploid hemp (Cannabis sativa) plant of genotype XXXY. (A) Whole plant. (B–D) Male flowering nodes. Scale bar = 80 cm.
Results of polymerase chain reaction allelic competitive extension assays for sex chromosome genotypes. Relative fluorescent units (RFUs) of FAM and HEX showing allelic discrimination of XXXX, XXYY, and XXXY sex chromosome genotypes among hemp (Cannabis sativa) F2 progeny from tetraploid F1 seed parents XXXX (A), XXXY (B), and control diploid XX (‘TJ’s CBD’ and ‘FL58’) and XY (‘Picolo’) plants. NTC = nontemplate control.
Three nodes bearing male flowers from triploid predominantly female hemp (Cannabis sativa) plants.
