We selected nine varieties in 2022 based on their suitability in other parts of the United States and their global performance (Rahemi et al., 2021; Cornell Hemp report, 2020; Ferfuia et al., 2021). The varieties selected can be found in Table 1 . In 2023, a total of 16 varieties were screened, and the details of origin and use can be found in Table 2 . Likewise, their flowering and maturation times can be found in Table 3 . Likewise, for the planting season 2024, 10 varieties were selected for the varietal screening. The selection of the varieties was based on the previous year’s field performance and the availability of seeds for the planting season. Similarly, naturally growing (wild-type) American ferals were collected from different parts of Missouri. Their seeds were sorted and grown in the field. Among these, LHV-17, one of the best-adapted wild types, was selected for comparative studies. LHV-17 was then compared with domesticated varieties.

The field experiments were conducted at the George Washington Carver Farm, Lincoln University, Jefferson City, Missouri, in 2022, 2023, and 2024. The site is located at an elevation of 213 meters above sea level. The experiment was conducted using a Randomized Complete Block Design (RCBD) experimental layout in two replications. Because this was an early-stage, multi-year study to screen hemp varieties, using two replications was considered sufficient (Kempton and Gleeson, 1997). Planting was carried out using a Great Plains 1006 NT No-Till Compact Drill, with a plot size of 185.89 m² and row spacing set at 19.15 cm. Fertilizer was applied at a ratio of 75:20:30 nitrogen: phosphorus: potassium (N: P: K). Seeds were planted at a depth of 0.63 cm.

Weekly field data were collected for each season as indicated in Table 4 .

Variation in flowering time and seed shattering are some of the major bottlenecks for hemp cultivation, with economic challenges. To address these issues, we investigated candidate genes such as CsCENLP and CsDrrp2, which may be involved in developmental processes related to flowering and cell wall composition, factors that contribute to seed pod dehiscence and dispersal. Understanding the molecular basis of these traits is crucial for developing improved cultivars with uniform agronomic performance and reduced seed shattering.

Statistical analysis was performed using RStudio version 4.3.2. Data from the 2022, 2023, and 2024 seasons were examined separately using one-way analysis of variance (ANOVA). Differences between means for each year were separated using Tukey’s Honest Significance Difference (HSD) test. A probability level of P< 0.05 was considered significant for F-statistic values and mean comparisons. Graphing was done using the ggplot command from the ggplot2 package in R (R Development Core Team, 2020). Only for the varieties that were grown in all three years, treatment x season interaction was performed using two-way ANOVA in R Studio.

There was significant variation in germination percentage among the varieties in the greenhouse conditions (P< 0.001). Carmenecta had the highest germination percentage at 68%, followed by BVL3 and Gravity. On the other hand, Vega had the lowest germination percentage at 26.83% ( Table 5 ).

However, under field conditions, all the varieties displayed emergence percentages ranging from 66% to 93.5% ( Table 5 ). The variation among the genotypes at the field level was not statistically significant (P > 0.05). Gravity and Orion demonstrated the highest emergence percentage of 93.6%. In contrast, Altair had the lowest emergence percentage (66%), followed by Felina (70%) ( Table 5 ).

The growth pattern from the emergence until harvesting among the different hemp varieties is illustrated in Figure 1 . Grain type varieties continued growing till 5^th week of sowing, after which they transitioned to the flowering stage from the vegetative stage. After the flowering stage, grain-type cultivars Vega, Felina, BVL1, BVL2, Ferimon, Uso-31, and Altair showed minimal growth. Unlike grain-type varieties, dual-type and fiber-type varieties continued growing until the 10^th week, providing evidence of longer maturation time ( Figure 1 ).

Plants reach the final plant height (FPH) when they cease to grow vertically. FPH is a key plant performance trait that is correlated with plant biomass, lodging, and yield. Furthermore, it is essential for compatibility with mechanized harvesting. The study demonstrated the varietal effect on plant height among different hemp varieties (P value<0.001). The detailed plant height can be found in Table 1 . BVL3 had the highest plant height, 164.59 cm, among the dual-type varieties tested for the year 2023. Similarly, Altair had the highest plant height of 91.44 cm among the grain-type varieties for the year 2023. On the other hand, Vega (68.07 cm) had the shortest plant height for the same year ( Table 1 ).

For the planting seasons 2022/2023/2024, Futura 83 displayed the highest plant standing, indicating its robust field performance ( Table 1 ). On the contrary, grain-type varieties, namely Ferimon, Uso-31, and Vega, displayed the lowest plant standing scores in all these years. While the plant standing provides some insight into the overall qualitative traits of the plants, future studies should incorporate more physiological measurements, such as chlorophyll content and photosynthetic rate, for improved precision. Based on the number of plants per square meter, Santhica (190 plants/m²) and Bialobrzeskie (180 plants/m²) had the highest plant population (P< 0.01) ( Table 5 ).

All the tested varieties displayed some degree of weed infestation. The dual-type varieties, such as Futura 83 and Carmenecta, had lower weed infestations per plot (weed infestation scores of 2.08 and 2.5) ( Table 5 ). This still means weeds covered more than 45% of the plot, a considerable weed pressure. Conversely, the grain varieties were highly susceptible to weed infestation, with Ferimon, Uso-31, and Vega displaying the highest weed infestation. Weeds covered more than 80% of the plots grown with grain-type varieties. The major weed we found was cocklebur (Xanthium strumarium) and other weeds we found were Goose grass (Eleusine indica (L.) crab grass (Digitaria sanguinalis (L.) Scop and other monocots.

We found significant stem diameter variation across industrial hemp varieties in three years (P value<0.001). Jinma (17.63 mm) had the largest stem diameter among the tested varieties in all three years. Among dual-type and grain-type, BVL3 had the largest stem diameter of 10.66 mm for the planting season 2023. In contrast, Henola, a grain-type variety, had the smallest stem diameter, 4.42 mm ( Table 1 ). Like plant height, a discernible pattern was observed in the stem diameter across different hemp types. Fiber variety exhibited the largest diameters, followed by dual-type and grain-type varieties.

We observed that stomatal conductance (gs) increased with increasing apparent transpiration ( Figure 2A ). A positive linear relationship was observed between stomatal conductance and apparent transpiration rate, with an R² value of 0.57 (P< 0.001), indicating 57% of the variation in transpiration is explained by changes in stomatal conductance. Our result aligns with physiological expectations: higher stomatal conductance leads to more water loss through transpiration. Additionally, we found that the stomatal conductance decreased with increasing vapor pressure deficit ( Figure 2B ). These two variables had a weak relationship (R² = 0.28), meaning only 28 percent of the variation in the stomatal movement in the plants is due to Vapor Pressure deficit (VPD) (P value<0.05).

Seeds are considered mature when their moisture content reaches between 18% and 20%. We found that most of the grain varieties flower and mature earlier than dual-type and fiber-type varieties under the photoperiodic conditions ( Table 3 ). Additionally, male plants flowered 15 to 20 days earlier than female plants ( Table 3 ). Ideally, industrial hemp should be harvested when the seed moisture content is 20-30% for optimal yield. Most of the grain-type varieties matured in 60 to 75 days ( Table 3 ). However, two grain type varieties originated from the USA had an intermediate flowering (60–65 days) and maturation time (115–120 days). Likewise, dual-type varieties took 115 to 120 days to mature ( Table 3 ). Varieties such as Carmenecta, Gravity, BVL1, and BVL2 had a higher ratio of male plants; nevertheless, they had consistent good yield ( Table 3 and Table 6 ).

There was a statistically significant difference in the whole plant, above-ground, and below-ground biomass among the varieties studied (P value< 0.001). Based on this study, Futura 83 had the highest biomass, 111523.51 kg/ha, followed by Carmenecta. On the other hand, Vega had the lowest biomass ( Table 7 ). Likewise, aboveground biomass, which provides the biomass for actual use, also differed significantly among the evaluated varieties (P value<0.001), with the variety Futura 83 posing the highest above-ground biomass. Likewise, below-ground biomass was highest on the dual type variety Carmenecta, comparable statistically with the check fiber type variety.

We performed linear regression analyses on whole plant biomass against other key traits influencing this characteristic. We found that stem diameter, final plant height, and below-ground biomass strongly correlated with the whole plant biomass, with R-squared values greater than 70% ( Supplementary Document 1-File 9 ).

We conducted a multi-year varietal trial to screen suitable varieties for the Midwest production regions and to identify parental lines for breeding purposes. 2022 was the starting year for our research, and given the limited seeds available, we screened nine varieties. During the 2022 planting season, the highest grain yield among eight evaluated varieties was from Futura 83, a French-origin variety, with a grain yield of 2434 kg/ha. The field trial continued in 2023 with a total of 16 varieties, including Futura 83 and other good-performing varieties. During the 2023 planting season, the Futura 83 again achieved the highest grain yield of 2,793 kg/ha. There was an increase in the grain yield compared to 2023. Similarly, in 2024, among the 10 varieties, Futura 83 consistently had a higher grain yield compared to other varieties, but it was lower than in 2023, with a grain yield of 2581 kg/ha ( Table 6 ). Despite different climatic conditions in multiple years ( Supplementary Materials Document 1-file 2 ), Futura 83 had a consistent grain yield on average of 2602 kg/ha for three consecutive years, suggesting stable grain production. In contrast, grain varieties such as USO-31, Ferimon, and Bialobrzeskie had lower grain yield in all the planting years ( Table 6 ).

In addition to Futura 83, BVL3, a USA-origin variety, showed good grain yield in 2023 and 2024. The grain yield for 2023 and 2024 was 1800 kg/ha and 2079 kg/ha, respectively. Additionally, it has a high plant biomass (22.26 g/plant). The percentage of male flowers in Futura 83 is very low, the male percentage in BVL3 is 45%.

We found that the homolog of the GmPdh1 gene (CsDrrp2) is present in the selected industrial hemp varieties, including the wild-type hemp accession (LHV17) ( Figure 5 ). To further understand its gene expression pattern, RT-PCR was conducted to examine the GmPdh1 homolog gene (CsDrrp2) gene expression among various hemp tissues in the early and late reproductive stages, Futura 83.

The seed development in hemp undergoes the following stages (The figure is demonstrated in Supplementary Document 1 File 8 ).

Flowering stage (R1): The stage when the first female flower is seen.

Early seed development stage (R2): The stage when the seed begins developing from the female flower. No seed shattering was seen in this stage.

Early seed filling stage (R3): The stage when the seeds are developed but still soft and milky when crushed with a finger. There was no shattering in this stage.

Later seed filling stage (R4): The stage when the seeds become hard and begin shattering, detaching from the mother plant, and spreading away from the middle lamella. The shattering was mild in the beginning and severe in the later stage.

Mature seeds (R5): The stage when the seed has entirely shattered and lies on the ground.

We observed two dehiscence zones that induced shattering. One was in the bracts that contain each seed ( Supplementary Document 1 File 8 ). Another zone was in the chalaza placental region, where the seed is physically attached to the mother plant. In most cases, shattering is because of the dehiscence zone I in the bracts ( Supplementary Document 1 File 8 ).

Phenotypically, the early flowering variety, USO-31, shatters more than the later-flowering fiber variety Jinma and Futura 83 ( Table 5 ). Our field study suggests that the varieties should have 65–70 days for maximum grain yield and less shattering ( Table 5 ). We studied the expression levels of CsDrrp2 at different stages of seed development in Futura 83 and LHV 17. There was a higher expression of this gene in the reproductive tissues than vegetative tissues in both Futura 83 and LHV 17. However, there was not statistical significance of differential expression in LHV-17 ( Figures 6A, B ). Similarly, we found CsCENLP to be expressed in all the tissues, particularly more in flowers and seeds than root ( Figure 7B ).

We observed all the varieties to have better germination in the field conditions than in the greenhouse conditions. Light irrigation in the field followed by seed sowing might have contributed to better germination in the field. In addition, fertilized soil, on the contrary to just the cocopeat used in the greenhouse, might have favored the germination in the field. However, further study should be conducted for further verification. We found that the different varieties had distinct growth patterns and time frames to transition from one growth phase to another. Grain-type varieties have earlier flowering and maturation times than fiber-type varieties. Like our study, Rahemi et al. (2021) and Ferfuia et al. (2021) also found that grain-type varieties, such as USO 31, had earlier flowering and maturation than fiber-type varieties. These phenological differences among the varieties might be because of the distinctive degree day requirement for each variety and the climatic differences between Missouri and the country where it was bred (Ferfuia et al., 2021). These findings would be critical for regional cultivar selection; short-maturing types could be selected in the Northern part of the United States, and later-maturing types could be selected for the Southern Region.

We observed differences in plant population among the tested varieties. This might be due to a combination of factors, such as poor germination, adverse soil conditions, and biological threats like insects and pathogens (Van Roekel and Coulter, 2011). Like our experiment, Rahemi et al. (2021) also reported the grain varieties to have higher plant populations. The higher plant population found in Santhica and Bialobrzeskie v might be because of its smaller size and the intrinsic seed vigor traits that facilitate more germination and seedling emergence (Huang et al., 2019).

Weed infestation is one of the significant challenges in industrial hemp cultivation, as there are limited herbicides available for industrial use. Our study did not use herbicides, thereby looking for inter-varietal weed suppression. We found that fiber variety, Jinma, contrary to grain and dual types, could suppress weeds efficiently. The fiber variety grew vigorously and formed a faster, denser leaf canopy, depriving weeds of sunlight and nutrients, thereby suppressing their growth. Furthermore, planting time also affected weed suppression. Notably, the earlier planting facilitated enhanced plant growth, outperforming weeds in fiber and dual-type varieties. Our findings were consistent with the similar findings in other cropping systems, which showed that early planting improves crop dominance over the weeds through faster establishment and resource capture (Williams, 2006). Based on our findings, fiber-type hemp varieties are recommended for fields with persistent weed pressure, primarily when herbicide use is restricted. Furthermore, we found an inverse relation of weed percentage with regard to the plant population ( Figures 3, 4 ) and increasing seeding rate might potentially reduce the weeds; however, this requires further investigation for confirmation. Our weed intensity is based on a visual estimate and future research should include quantitative parameters such as weed biomass, nutrient competition, and additive findings.

Our findings on plant height and stem diameter were similar to those in previous research conducted in other parts of the United States. Notably, all the researchers found the fiber varieties to be the tallest, followed by the dual and grain types (Cornell Hemp Trial Report, 2020; Darby, 2021; Rahemi et al., 2021). Plant height (PH) is primarily influenced by stem elongation because of continuous cell division and the expansion of cells generated by the apical and intercalary meristems. In most plants, the plant height and diameter are polygenic traits (Wang and Li, 2008). In addition to genes, plant height is also regulated by plant hormones, including gibberellin (GA), brassinosteroid (BR), auxin (IAA), cytokinin (CK), and strigolactone (SL) (Tong et al., 2009). However, there has been a limited study conducted on the genetic architecture that determines the plant traits, such as plant height and stem diameter, in cannabis. While our study found the plant height to be associated as a varietal trait, further research should study what particular genetic mechanism and hormonal regulation contribute to the distinctive phenotype observed among the varieties (Petit et al., 2020b).

Grain yield is an important trait for the dual-type industrial hemp varieties (Alonso-Esteban et al., 2022). The grain yield among the tested varieties ranged from 328 kg/ha to 2,796 kg/ha. The average grain yield for the varietal trial was 1,283 kg/ha, which is higher than the national average of 896 kg/ha National Agricultural Statistics Service (2022). In a varietal screening study by Das et al. (2020), grain yields ranged from 27 kg/ha to 2,362 kg/ha, aligning with the range observed in this study. Similarly, Rahemi et al. (2021) reported an average grain yield of 771 kg/ha, which also falls within the range found in this study. As in the present research, Futura 83 demonstrated superior yield and performance traits in both New York and Vermont, confirming its suitability for cultivation across a wider geographic area (Cornell Hemp Report, 2020; Darby, 2021).

Grain yield is a complex trait. Hemp grain production varies among varieties and is influenced by environmental conditions such as soil type, moisture levels, and agronomic practices (Campbell et al., 2019; Campiglia et al., 2017; Poudel et al., 2020; Rahemi et al., 2021; Stack et al., 2021; Toth et al., 2020). In a study conducted across 14 cultivars in Latvia, the Czech Republic, France, and Italy, the grain yields were primarily determined by variety type and then by flowering time, which is consistent with our findings (Campiglia et al., 2017). We also found that grain yield was strongly influenced by variety and flowering time, as observed by Visković et al. (2024).

The climate study by Ferfuia et al. (2021) found that high temperatures during early grain filling negatively impact seed yield. While we did not study the impact of temperature on the yield, we found that early maturing varieties that had a maturation time of 55–60 days had lower yield. Dual-purpose varieties such as Futura 83 produced high grain and biomass yields, which has intermediate maturation times of 120 days. We observed higher biomass to be related to more grain yield, but this trait also additionally implies to higher fiber strength, assimilable sugars, making it more readily available for biofuel and fiber production (Williams et al., 2025). In addition to varietal effects and flowering time, grain yield is highly influenced by the Nitrogen dosage (Tang et al., 2022; Podder et al., 2024). The national average for grain yield is rising, probably because of the discovery of better agronomic practices, genetics, and screening stable varieties in the USA. However, no single variety has been identified that performs well in all the geographic regions of the United States, suggesting the paramount importance of this study (Williams et al., 2025). However, this study used only two replications per site-year, which may reduce the precision of traits such as flowering time and grain yield, especially given their strong environmental interactions. Despite this limitation, the consistent yield observed across multiple years suggests stable genetics in the Futura 83 variety. Nevertheless, future studies should incorporate more replications and multi-environment trials to allow for more robust statistical inference and to better support breeding decisions. Although our study did not delve deeper into the sexual dimorphism and grain yield, this factor might play an important role in this trait. While excessive male flower production can reduce seed yield in dioecious varieties, monoecious plants may produce inadequate pollination. Additionally, in hemp, monoecious might require allocating much resources to pollen production, reducing the resources available for seed development. On the contrary, in a variety of dioecious plants. In some cases, early male plants may senesce before females reach full reproductive maturity, enabling more effective seed development by females. While our study just explored the varietal effect on the yield, future research should explore how distribution of male and female flowers may impact yield parameters.

We studied the phenotypic variation of key seed compositional traits such as total protein, crude protein, crude fat, and crude fiber, which are key components for human food and animal feed. The total protein observed in the hemp was 26-31% per gram of seed, similar to the previously identified range of 25%-30% (Aluko, 2017; Sun et al., 2021; M. Liu et al., 2023). Likewise, the crude protein in industrial hemp was 23-27% per gram of seed on a dry weight basis. Our study was consistent with the research findings by Alonso-Esteban et al. (2022), which found hemp seeds to contain between 25% and 35% crude protein and between 20% and 25% crude oil. We found Ferimon and Futura 83 to have the highest crude protein, with 27% and 26%, respectively. These studies further provide evidence of hemp seeds as a rich source of digestible proteins for humans and animal feed (House et al., 2010; Alonso-Esteban et al., 2022). Furthermore, Edestin could be used as a substitute for blood plasma when plasma is lost (such as in trauma and surgery) (Horska, 2005).

A complex genetic network controls grain filling. Edestin 1 and Edestin 2 encode for edestin protein, the predominant protein in industrial hemp grain (Deguchi et al., 2020; Cattaneo et al., 2021; Sun et al., 2021). Seed storage protein (SSP) encoding genes are primarily activated during the grain-filling and maturation stages. SSPs accumulate in the embryo or endosperm of the seed, depending on the plant species. The endosperm remains a single-cell layer in dicotyledonous plants such as industrial hemp. At the same time, the embryo cells develop protein storage vacuole (PSV) structures to store SSP (Yang et al., 2022). Although significant development has been made in understanding seed protein and storage reserve accumulation in model crops, little work has been conducted on this topic in industrial hemp (Yang et al., 2022). The cause of the variation in seed proteins among the tested genotypes is unknown. The higher total and crude protein observed in Futura 83 and Ferimon 79 varieties may be attributed to their distinct genetic makeup, the environment favoring this cultivar, or the interaction of both genes and the environment. These high-protein-containing cultivars, such as Futura 83 and Ferimon 79, likely possess traits that enhance the expression of these seed storage encoding genes. Increased expression of these genes could result in higher seed protein content in these varieties. However, further investigation is needed to understand the mechanisms governing protein synthesis in hemp comprehensively.

One major challenge in hemp cultivation is the lack of uniform flowering and seed maturation. Some plants may remain vegetative within a single field while others reach full maturity. To exacerbate shattering, some flower clusters fully mature to seed within a single plant, while others are still in intermediate maturity ( Supplementary Material Document 1 File 8 ). One way to fix the seed shattering in hemp is to restore the flowering time and engineer it to flower uniformly. The hemp breeding goal should target an intermediate flowering time to reduce seed shattering. Flowering in hemp is often triggered by signals received by the meristematic tissue related to alterations in environmental factors such as light intensity, daylength, influencing flowering time and intensity in industrial hemp (Downs and Borthwick, 1956; Petit et al., 2020b). The plants flower prematurely if they experience daylight periods shorter than 14 hours (Hall et al., 2012; Cheng et al., 2022).

Thus, understanding the mechanisms governing flowering, sexual orientation, and their relation to seed development and shattering is necessary for hemp breeding and improvement (Toth et al., 2020; Asiamah et al., 2025). Our study found the flowering time was different among the tested varieties. However this trait in hemp is very photoperiodic and future research should include how the light intensity, day length and other environmental factors and study whether thresholds vary among varieties to better determine which traits are most strongly conserved (Hall et al., 2012; Zhang et al., 2025). In the meantime, our study remains valuable for preparedness in cultivation and harvesting. Further study is required to characterize the gene expression among different early, late, and intermediate flowering varieties to better understand the flowering patterns and maturity in hemp varieties, and this will help design molecular breeding strategies for flowering time and maturity traits. Additionally, future research should sequence the sex specific gel bands and compare with the available pangenome for the validation (Lynch et al., 2025). While CsCENLP has not been directly linked to flowering and shattering, this gene might regulate the timing and the place in the tissues of lignin deposition, which is a very critical physical phenomenon for shattering. Future transcriptomics research and histological research on CsCENLP mutants or overexpression lines could give a better overview of the relationship between CsCENLP and seed shattering.

In dicot plants, such as industrial hemp, seed shattering is caused by the development of a specialized dehiscence or abscission zone along the seed cover. During the seed development stage, cells in this zone differentiate into lignified and separation layers, which later undergo self-degradation, after which the seed cover splits open (Maity et al., 2021). Lignification is a necessary process involved in seed shattering. Alongside lignin, other key components of the plant’s secondary cell wall, such as cellulose and hemicellulose are related to cell wall strength and integrity, directly affecting pod dehiscence. Higher cellulose, hemicellulose, and lignin contents in pods are strongly associated with dehiscence (Hofhuis et al., 2016).

In soybean, GmPdh1 encodes a dirigent (DIR)-like protein. It enhances pod dehiscence by increasing the torsion of dried pod walls, generating a driving force for pod splitting under low humidity conditions (Funatsuki et al., 2014). The homolog of the GmPdh1 gene is the CsDrrp2, which we hypothesized to be associated with shattering. While the expression of the CsDrrp2 in the later seed development stage might associate with seed shattering, however, the exact role of this gene is unknown and we cannot provide its function based on the current result. It is unknown whether it causes lignin deposition in the dehiscence layer or cellulose deposition. The homolog of CsDrrp2 promotes lignification in the inner sclerenchyma of pod walls, the site of thick secondary cell wall formation (Funatsuki et al., 2014; Marsh et al., 2023). This gene may play a similar role in Cannabis sativa by promoting lignification in the inner sclerenchyma of the seed coat, as observed in legumes. Such activity could enhance the differential lignin deposition between the inner and outer layers of the seed coat. During consecutive drying cycles, these differences may generate dehiscence forces that lead to seed shattering. The future direction should be RNA sequencing of the highly shattering grain type varieties and the less shattering ones based on our screened varieties for seed development and shattering genes. Furthermore, additional seed shattering genes such as SHAT-5, SHP1 and SHP2 that are found to be associated with shattering should also be studied for further exploring the mechanism of shattering (Maity et al., 2021). Shattering occurs as a torsion in the bracts surrounding the seeds, pulling both the seed covers and exposing the seeds. Although our experiment included the bracts surrounding the seeds, further research should focus on identifying the specific tissue and the region where this gene is expressed. Furthermore, RNA in situ hybridization should be performed to determine the exact location of the expression of this gene and its role. Future work would be knocking out this gene and creating mutants to see if there are any quantifiable differences in the seed shattering.