Fifty cuttings of the medical hemp cultivar ‘Trump’ were harvested from the same mother plant and propagated in 3” deep cells filled with a 1:1 mix of peat:vermiculite that was pH adjusted to 5.8 using hydrated lime (Ca(OH)₂). We used the medical hemp cultivar ‘Trump (T1)’ because it has high cannabinoids and a compact growth habit that is conducive to growth chamber studies where space is limited. After two weeks in propagation, 18 rooted cuttings were selected for uniformity and transplanted into 6.7 L plastic containers (#2 nursery pots) that were filled with a soilless mix of 6:1:1 peat moss, vermiculite, and rice hulls and randomly assigned to six groups. Each group consisted of three plants in three pots that shared a common leachate collection tray. Each group was randomly assigned to a P treatment (25, 50 or 75 mg per L) and placed in one of six leachate collection trays within an environmentally controlled growth chamber. Each P treatment was replicated twice (n=2). Plants were pinched to four nodes upon being moved into the growth chamber and grown under a vegetative photoperiod (18/6 hr day/night). After seven days of vegetative growth, the photoperiod was switched to an inductive photoperiod to promote reproductive growth (12/12 hr day/night). Plants were grown under an inductive photoperiod for 56 days (8 weeks). At harvest, the canopy area for each group was measured and used for yield and leached P calculations.

The extended photosynthetic photon flux density (ePPFD; 400 to 750 nm) was 600 ± 30 µmol m^-2 s^-1 during vegetative growth (18 hr/6 hr light/dark; DLI: 38.9 mol m^-2 d^-1) and 900 ± 50 µmol m^-2 s^-1 during reproductive growth (12 hr/12 hr light/dark; DLI: 38.9 mol m^-2 d^-1) from white+red LEDs (BIOS Lighting Inc., Icarus Vi) (Kusuma et al., 2021). The fraction of far-red photons (700 to 750 nm) was 1.5%. Because the far-red fraction was low, the classic PPFD and the ePPFD were within 1.5% of each other (Zhen et al., 2021). Lights were dimmed as plants grew to keep a constant PPFD at canopy height throughout each growth phase. Temperature was 25.9 ± 0.4/23.3 ± 0.2 °C (day/night) measured with a shielded fan-aspirated thermistor (Apogee Instruments Inc., model ST-100). Vapor pressure deficit (VPD) was 1.25± 0.06/1.02 ± 0.08 kPa day/night measured with a relative humidity and temperature sensor (Campbell Scientific Inc., model HMP45A). Environmental measurements were made every ten seconds and ten-minute averages of the data were recorded with a datalogger (Campbell Scientific Inc., model CR1000X). Fans supplied airflow of about 0.8 m per s at the top of the canopy measured with a hot-wire anemometer (TSI Inc., model 8330)

Nutrient solutions were mixed using deionized water and reagent grade salts (Sigma Aldrich). Table 1 shows the composition of the nutrient solution. The concentration of elements followed mass balance principles as described by Bugbee (2004) and Langenfeld et al. (2022).

Plants were irrigated daily with 2 L per hour drip emitters (NetaFim) to about a 15% excess. Drippers were tested at the beginning and end of the study to ensure uniformity. Leachate from each tray was collected and monitored daily for pH and electrical conductivity (EC). pH measurements were made with an Oakton pHTestr 10 BNC pH meter. EC measurements were made with a Hanna Dist 4 EC meter. The average leachate pH over the study was 5.4 ± 0.3 and the average leachate EC was 1.2 ± 0.6 mS cm^-1 (mean ± sd) across all treatments and reps. At weekly intervals, leachate P was measured using a colorimeter (LaMotte model Smart 3). The low range phosphate method was followed, and solutions were manually diluted to get within the 0.0 to 3.0 ppm PO₄ range (LaMotte method 3653-SC). Elemental P concentration was calculated by multiplying the measured PO₄ concentration by 0.326. The 0.326 multiplier is the molecular weight ratio of P to PO₄.

Plant height was recorded from the base of the stem to the tallest part of the plant at the beginning and end of the study. At harvest, plants were manually separated into flowers, leaves, and stems. The leaves that subtend the inflorescence (sugar leaves) were included in the flower fraction. Fresh weight was recorded, and plants were placed on well ventilated shelves at 25°C and 30% relative humidity. After 5 days, dry mass was recorded. The biomass at this temperature and humidity was between 12 and 14% moisture. Flower yield (g m^-2) was calculated as the total mass of flower per tray divided by the canopy area. Harvest index (HI) was calculated as the ratio of flower mass to total above ground biomass.

Approximately 1 g (dry) of leaf and 3 g (dry) flower tissue was sampled from each plant at harvest and analyzed for nutrient content. For element analysis, tissue was oven-dried at 80 C for 48 hours prior to analysis. The tissues from multiple plants in each tray were homogenized into one sample. Fan-leaves and flowers (with sugar leaves) from the top of the plant were used for analysis. Stems and roots were not analyzed. Tissue samples were analyzed using inductively coupled plasma optical emission spectroscopy (ICP-OES) at the Utah State University Analytical Laboratory (USUAL). USUAL is an accredited member of The North American Proficiency Testing Program (NAPT; naptprogram.org). Phosphorus use efficiency (PUE) was calculated as the total mass of P measured in the flower and leaf tissue divided by the total amount of applied P. Total recovery of P was calculated as the sum of P in the flowers, leaves and leachate divided by the total amount of applied P.

Flower material was sampled from dry flower buds at the top of the plant. Three flower buds (about 3 g dry) were sampled per plant. The dry material was ground to a fine powder using a stainless-steel coffee grinder and stored in plastic bags at 4°C until analysis. Approximately 100 to 300 mg of dry material was sampled and extracted with reagent grade methanol (10 mL methanol per 100 mg tissue). This was then vortexed for one minute and sonicated for 15 minutes. The sample was then diluted by adding 100 μL cannabinoid solution to 900 μL of reagent grade methanol and moved to a high-performance liquid chromatograph (HPLC) for cannabinoid quantification.

Sample extracts were analyzed using a Shimadzu model LC-2040C 3D Plus (Shimadzu Corporation, Kyoto, Japan) reverse phase high-performance liquid chromatography (HPLC) equipped with diode array detector and a NexLeaf CBX for Potency, superficially porous particle (SPP) 2.7 μm C18 4.6x 150 mm column (Shimadzu Corporation, Kyoto, Japan). The mobile phase was 0.085% phosphoric acid in acetonitrile (E₁) and 0.085% phosphoric acid in distilled water (E₂). Elution was performed by the following gradient: t_{0 min} = 70% (vv^-1) E₂; t_{3 min} = 70% (vv^-1) E₂; t_{8 min} = 85% (vv^-1) E₂; t_10 min = 95% (vv^- 1) E₂; t_{11.01 min} =70% (vv^-1) E₂. The flow rate was 1.6 mL per min. A reference standard containing the cannabinoid compounds of interest (Cayman Chemical Inc., Phytocannabinoid Mixture 11 (CRM)) was used to prepare the external calibration standards. All compounds were analyzed at 220 nm. CBDA had a retention time of 3.262 min; CBD had a retention time of 3.901 min; THCA had a retention time of 7.604 min; Δ9-THC had a retention time of 6.427 min; CBGA had a retention time of 3.519 min; and CBG had a retention time of 3.726 min. Minor cannabinoids were excluded from the analysis due to low concentrations. CBD, THC, and CBG equivalents (CBD_eq, THC_eq, and CBG_eq) were calculated as described by Westmoreland et al. (2021).

The study was a completely randomized design with three treatment levels (P input) and two replicates at each level (n=2). Each experimental unit consisted of three plants in three pots that shared a common tray for leachate collection. Flower yield was calculated as the sum of the dry flower from each tray divided by the growth area. For cannabinoids, each plant was sampled individually and the average of the three plants within an experimental unit was used for statistical analysis. For tissue element analysis, a representative sample of tissue from each plant was homogenized and analyzed as a single sample. Data were fit with a simple linear model in RStudio (R statistical software, version 4.1.0). Effects were considered significant at α = 0.05.

All plants were green and healthy throughout the lifecycle, and, in spite of frequent close inspections, there were no visual differences among treatments at any time. Dry flower yield was unaffected by P input (Supplementary Figure 1; p = 0.20). The dry flower yield was 649 ± 41.5, (mean ± SD) grams per square meter (g m^-2) averaged across all treatments. This resulted in a photon conversion efficacy of 0.26 ± 0.02 grams per mole of photons. PCE was calculated including photons from both the vegetative and flowering stage. Harvest index (HI) was 65 ± 0.02% averaged across all treatments. There was no effect of P input on HI (data not shown; p = 0.90).

We found no significant effect of P input on CBD_eq (p = 0.35), THC_eq (p = 0.38) or CBG_eq (p = 0.94) across P concentrations from 25 to 75 mg per L (Supplementary Figure 2). CBD_eq was 13.6 ± 0.58; THC_eq was 0.57 ± 0.03; and CBG_eq was 0.51 ± 0.02 averaged across treatments. The ratio of CBD_eq to THC_eq was identical among treatments at 23.8 ± 0.05.

High rates of P can affect the uptake of other ions but increasing P supply resulted in only small differences in the nutrient content of flowers and leaves (Figure 1). The nutrient content of the leaves were within published optimal ranges for all P treatments (Cockson et al., 2019; Landis et al., 2019; Kalinowski et al., 2020).

For the macronutrients – N was reduced from 3.1 to 3.0% in the leaves and from 3.9 to 3.7% in the flowers as the P input increased from 25 to 75 mg per L (Figure 1; p = 0.04). As expected, P increased from 0.66 to 0.89% in the leaves and from 1.02 to 1.13% in the flowers as the P increased from 25 to 75 mg per L (Figure 1; p < 0.01). This translates to a 35% increase in the leaves and 11% increase in the flowers in response to a 3-fold increase in P input. Interestingly, K content in the leaves and flowers was unaffected by P treatment, despite a 45% increase in K (from KH₂PO₄) as the P input increased from 25 to 75 mg per L. Ca increased from 6.0 to 6.2% in the leaves and from 1.3 to 1.6% in the flowers with increasing P input (Figure 1; p = 0.05).

For the micronutrients – Cu increased from 4.9 to 5.9 mg kg^-1 in the leaves and from 9.3 to 11.2 mg kg^-1 in the flowers (p = 0.02); Zn increased from 82 to 89 mg kg^-1 in the leaves and from 77 to 83 mg kg^-1 in the flowers (p = 0.04); and Mo increased from 0.27 to 0.43 mg kg^-1 in the leaves and from 0.46 to 0.51 mg kg^-1 in the flowers (p = 0.01) as P increased from 25 to 75 mg per L (Figure 1).

Nutrient content tended to be higher in the flowers than leaves for the mobile elements N, P and K, and higher in leaves than flowers for non-mobile elements Ca, B and Mg (Figure 1). Fe and Mn were the only elements that were not significantly different between leaves and flowers. Cu was twice as high in the flowers as in the leaves (Table 2). There was no interaction between P treatment and tissue type for any element tested (Figure 1).

P in the leachate increased to a maximum in all treatments at five weeks, followed by a decline or stable leachate P concentration for the final 4 weeks (Figure 2). As expected, increasing P input increased P in the leachate. At an input of 25 mg per L, leachate P concentration increased to a maximum of 28 mg per L, then dropped to less than 10 mg per L; at an input of 50 mg per L, P in the leachate increased to 75 mg per L and remained stable; at an input of 75 mg per L, P in the leachate increased to about 160 mg per L but declined to about 115 mg per L P at harvest (Figure 2).

The cumulative leached P was divided by the plant growth area to calculate grams of P leached per square meter. There was a linear increase in total leached P as input P increased from 25 to 75 mg per L P (Figure 3; p<0.001). Cumulative leached P increased by 5% for every 1 mg per L increase in input P, which translates to 12-fold increase in cumulative leached P in response to a 3-fold increase in input P.

At 25 mg per L, 65% of the applied P was in the flowers, 15% was in the leaves and 10% was in the leachate (Figure 4); the PUE was 80% and the total recovery of P was 91 ± 5.5%. At 50 mg per L, 35% of the applied P was in the flowers, 8% was in the leaves and 35% was in the leachate (Figure 4); the PUE was 41% and the total recovery of P was 82 ± 7.4%. At 75 mg per L, 25% of the P was in the flowers, 6% was in the leaves and 50% was in the leachate (Figure 4); the PUE was 31% and the total recovery of P was 85 ± 8.2%. The unrecovered P may have been in roots and stems, which were not analyzed.

Recent studies indicate that P in continuous liquid feed is adequate at about 15 mg per L for vegetative growth in greenhouse conditions with ambient CO₂ (Cockson et al., 2020; Veazie et al., 2021), but many studies, including this one, indicate higher levels of P are necessary for optimal reproductive growth (Bevan et al., 2021; Shiponi and Bernstein, 2021b). P uptake during reproductive growth may be much higher than during vegetative growth because developing inflorescences are enriched with P (Figure 1) and are a significant fraction of the biomass (Westmoreland et al., 2021). Although these inflorescences are a significant reservoir of P, this accumulation of P does not appear to be beneficial for yield or quality. This indicates that Cannabis has luxury uptake of P.

In medical Cannabis cultivation, growers actively prevent pollination and seed formation. This is significant because seeds of Cannabis contain high concentrations of storage P as phytic acid (PA) (Lott et al., 2000). PA is a critical source of P for developing seedlings in low-P soils, but it reduces the pool of active P in the plant and is unnecessary in medical Cannabis. The P in the flowers may be in the form of PA, but that is beyond the scope of this paper. Nevertheless, the high concentrations of P and PA in Cannabis seeds may explain the high uptake of P in medical Cannabis flowers.

Collectively, these data suggest a potential benefit of phasic P fertilization, where P supply is kept low during early growth and increased during late flowering to optimize yield (Correll, 1998). Future studies should examine increased P at different stages of reproductive growth.

Shiponi and Bernstein (2021b) studied two cultivars: a type I (THC dominant) and type II (roughly equal CBD and THC). There was no yield increase above 30 mg per L in the type I variety, but the yield of the type II variety increased by 30% from 30 to 90 mg per L. This interaction suggests the potential to genetically select Cannabis cultivars with a lower P requirement. Luxury uptake of K in fiber hemp has been reported (Finnan and Burke, 2013), but this was not observed in this study (Figure 3). This suggests there is the potential to select cultivars with less unproductive P in the flowers due to lower rates of luxury uptake.

We studied one cultivar in this study so it is difficult to draw conclusions about genetic variability in P uptake among Cannabis cultivars, but there is evidence from other crops to suggest potential differences that can be exploited through breeding. Cultivars of several crops vary in the accumulation of P. In groundnut, P in the seed ranged from 59 to 103 mg per 100 g, while PA-P ranged from 149 to 315 mg per 100 g (Hande et al., 2013). In soybean, P in seed ranged from 0.19 to 0.37 g per kg and PA-P ranged from 3.8 to 5.1 g per kg (Israel et al., 2006). In potato, the cultivar Pamella achieved maximum yield with no additional P application, whereas Desiree required 100 kg P₂O₅ per ha to achieve maximum yield (Daoui et al., 2014).

There are also differences in PUE among cultivars of crop plants in the field (Horst et al., 1993; Ciarelli et al., 1998; Rose et al., 2007; Powers et al., 2020; Beroueg et al., 2021). Differences in root architecture is a major contributor to this variability, with some cultivars more capable of increasing absorptive surface area under low P conditions. (Shenoy and Kalagudi, 2005; Beroueg et al., 2021). Cultivars can also vary in their ability to modify the rhizosphere by releasing organic acids and increasing the solubility of P in calcareous soils (Shenoy and Kalagudi, 2005), but this is not typically beneficial in controlled environment agriculture where adequate fertilizer is continuously supplied and substrate pH is tightly controlled.

Cannabis is becoming increasingly legal (Eastwood, 2020), leading to a rapid increase in cultivation. In northern California, the total area under cultivation increased by over 90% between 2012 and 2016 (Butsic et al., 2018). The environmental impact associated with this rapid increase is coming to the attention of cultivators, scientists and policy-makers (Mills, 2012; Wang et al., 2019; Summers et al., 2021; Zheng et al., 2021).

The effect of phosphorus on the environment has been well known for decades (Edwards and Harrold, 1970; Correll, 1998). Cannabis cultivation has been largely unregulated, but as legal cultivation continues to increase, there is a need to address the impact of over-application of fertilizer, especially P. Commercial cultivators can capitalize on the use of low P by marketing their product as sustainably grown.

There is a need for evidence-based recommendations for fertilization approaches that optimize Cannabis yield and quality with a focus on the associated environmental impact. In this study, an input P of 25 mg per L was sufficient for maximum growth and quality. The flowers accumulated high concentrations of P, but this did not result in higher flower yield or cannabinoid concentration.

Finally, we report a 12-fold increase in leached P in response to a 3-fold increase in input P. This study adds to the growing body of evidence indicating Cannabis does not benefit from excessive P fertilization and provides new insight into the accumulation and distribution among organs within the plant. Future studies should address genetic variation in P accumulation among Cannabis varieties and breeders should select for cultivars that accumulate less P in the flowers.

Controlled environment agriculture has the potential to reduce fertilizer and water use, but more research on precision nutrition is necessary to achieve this goal. Growers in field agriculture have been gradually adopting to strict regulations on fertilizer use; growers in controlled environments can set a new standard for sustainability for all types of crop production.