Hemp Sowing Seed Production: Assessment of New Approaches in North-Italy

Abstract

Bottlenecks occur as the cultivated area increases and the inappropriate mechanization methods hinder the stable supply of seeds for hemp multiplication. Moreover, the seeds ripen scalarly, and a timely stabilization is required to impede a germinability reduction. The study coupled the delayed sowing of a non-specific hemp variety to a harvesting system allowing for the collection of seeds with other threshing fractions. The final goal was the identification of an innovative route integrating agronomic and mechanical aspects for the improvement of the supply chain of seed production. To this aim, harvesting trials were carried out on the variety Futura sowed at the end of June in North Italy and were collected with a combine equipped with a separator developed for the recovery of threshing residues. The shortening of the growth cycle did not affect the plant height (173 cm on average). The effective working time of the combine was 57% of the total working time and the field efficiency was 1.14 ha h⁻¹, a good performance considering that in our work the seeds was threshed simultaneously to the harvest operation. Seed losses were found to be mostly at the expense of the mowing and threshing system (sector B) but remained below 5%. The separation system allowed for the rescue of 492.20 kg ha⁻¹ (DW basis) of high-value threshing residues.

1. Introduction

Hemp (Cannabis sativa L.) is a high-yielding multi-purpose crop historically cultivated mainly for fiber production [1,2,3]. In the last decade, after more than a half-century of ban from cultivation, the European hemp cultivated area has been increasing, encouraged by the positive impact on the environment, such as the low requirement for pesticides [4] and the increasing request of raw materials for the emerging bio-based economy [5]. Even so, the European hemp production chain is a niche economy, because of the missed constant upgrade in terms of variety selection, agronomic practices and harvesting techniques, which are required parameters for stable supplies of standardized raw materials.

In recent years, new technologies for increasingly sophisticated machines and new hemp varieties led to new scenarios [6]. The choice of a hemp harvesting system strongly depends on its final use: fibers (long or short); non-textile uses (paper, panels, substrates) where the fiber length is irrelevant; seed; double use (fiber and seed). Weaknesses of the hemp production chain in Italy, including seed production, comprise the need for research on mechanization and the lack of a national production chain [7]. One of the main general issues concerns the high biomass yields, as well as the specific mechanical characteristics of the crop which determine a huge load to the machine components. The mechanical harvesting of hemp seeds has other specific peculiarities. Hemp is characterized by the scalar ripening of the seeds and, in most cases, the seeds of the basal level of infructescence become scattering while the apical ones are not ripened yet. Scattering and seed loss can be exacerbated by adverse weather conditions, atmospheric agents, and mechanical head shaking during the harvest. In addition, when seeds are ready to be harvested the vegetative part is fully matured and, hence, presents the maximum biomass highly lignified [8]. Other critical factors to keep in mind are the volume of the infructescence and the moisture content of the seeds.

Beyond seed losses, harvesting techniques should focus on reducing mechanical damage to seeds, safeguarding their quality, and increasing their economic value in the sowing use. Harvest conditions, timely availability of the cleaning and drying plant as well as pre-treatment of the harvested material and storage conditions until processing are all important issues for producing high-quality seeds [9].

Recently, monoecious varieties have been selected to increase the crop uniformity through the reduction in sexual vegetative dimorphism and for facilitating seed harvesting [7,10,11,12]. However, the monoecious hemp varieties also require additional manual work for removing the male plants carrying the trait of dioeciousness. This means managing smaller fields where the use of simpler machines is advisable [8]. The statement is confirmed by the survey of Giupponi et al. [7] which observed as in Italy the seed harvesting is usually performed by a combine harvester developed for grain crops. This approach allows for a smoother inclusion of hemp into the most common farming systems and avoids the purchase of additional expensive machines. Nevertheless, certain modifications or device couplings are still necessary due to the specific features of the crop [13], e.g., long pieces from hemp stem or even loosened fibers can cause wrappings leading to considerable damages to rotating components of the machines [14]. Delayed sowing can be a valid approach for reducing the plant height and for favoring the use of conventional combines for mechanical harvesting [12,15]. However, the results in terms of seed production are highly dependent on the genotype and environment interaction. Ferfuia et al. [15] reported an increase in seed production of 55% and 17% with delayed sowing of monoecius cultivars compared to the sowing in conventional time. Delayed sowing at two sites in Belgium determined a decrease in seed yield by 0.023 t ha⁻¹ with a one-day delay in the sowing date [16].

Seed hemp can be collected in one, two or three steps. The single step involves the use of a self-propelled combine harvester equipped with or without specific heads and separation/cleaning systems for hemp [6,9]. The harvest in two steps includes a first passage with a machine for mower-windrower. After natural drying, the plants are collected with a combine harvester equipped with pick-up header. This system increases the homogeneity and quantity of ripened seeds [17]. Otherwise, the hemp plants were first cut with a high header only for seed harvest and then the remaining straw was cut and baled [18]. In other conditions, three separate steps were considered: first mowing, then transport and finally threshing in a dedicated area similar to onion seed harvesting procedures [19].

CREA (Consiglio per la ricerca in agricoltura e l’analisi dell’economia agraria) has been involved for years in the development of new technologies and solutions for emerging supply chains in the no-food sector [20,21,22]. For hemp, CREA focused its effort to refine agronomic and technical solutions for the development of a national hemp sowing seed supply chain, with the support of important Italian sowing seed companies [23].

The study combined the delayed sowing of a non-specific hemp variety for seed production to a harvesting system allowing for the concurrent collection of seeds and other threshing fractions, and to the seed treatment for cleaning and for stabilizing the moisture content. The working hypothesis about the delayed sowing relied on the “reducing” effect on biomass production to make the harvesting easier. For the concurrent harvest, we used a specific separation system developed by CREA which is under patent licensing. The final goal was to verify the efficiency and applicability of a different route where the integration of agronomic and mechanical aspects could improve the supply chain of seed production. As far as we know, this work is the first experience where both solutions were applied for seed harvest from a non-specific hemp variety.

2. Materials and Methods

2.1. Hemp Field

The study was conducted in 2018 on an experimental field located at San Pietro in Campiano (44°16′42.9″ N 12°11′18.5″ E), Ravenna (RA), within a commercial farm in Northern Italy. The site was regularly shaped with a length of 189 m and a width of 30 m for a total area of 5670 m². For avoiding the edge effect, ten meters from each header and 4.5 m from each side were removed to avoid the edge effects thus considering 3718 m² (169 m × 21 m) as the experimental area. The variety sowed was Futura 75 the most cultivated hemp variety in the last 5 years. It is a monoecious medium-late maturing variety selected in French and mainly used for dual-purpose, stems and seeds for industrial use.

The crop was sowed on the 25th of June, soon after the harvesting of triticale (Tritico x secale). Before sowing, the soil was harrowed in the first 10 cm and fertilized with digestate fractions according to the concepts of sustainability and soil preservation. The whole area (5670 m²) received 8.5 t of digestate of bovine slurry provided by the farm of Centro Ricerche Produzioni Animali (CRPA), Reggio-Emilia, Italy, involved in LIFE Seq-Cure EU Project. The slurry contained 3.8 g kg⁻¹ of total nitrogen (66% ammonia nitrogen), 0.5 g kg⁻¹ of phosphorus and 2.6 g kg⁻¹ of potassium; overall, a total of 57 kg ha⁻¹ of nitrogen, 7.5 kg ha⁻¹ of phosphorus and 39 kg ha⁻¹ of potassium were distributed. Thirty kilograms per hectare of hemp was sowed using a grain rows seeder with rows 20 cm apart, to obtain a target of 150 plants m⁻². Traditionally in Italy, the hemp is sowed in April but, for the objectives of this study, the date was delayed, to limit the development in height of the plants [24]. The sowing depth was 2.5 cm.

Water was supplied with two irrigations (25 mm each) with a self-propelled automatic sprinkler in August and no other fertilization were scheduled. The meteorological conditions were monitored by the regional network of environmental monitoring (ARPA).

2.2. Harvesting Methods

Before the harvesting, at the beginning of October we carried out a characterization of the crop. Three representative sampling areas of 1 m² were hand-harvested to monitor the development of the plant, the ripening stage of the inflorescence, and to define the optimal time for mechanical harvest of seeds. The representativeness of the selected areas was given by the presence of plants showing uniform growth and the same phenological stage. The plants were cut at soil level, and, in each plot, we counted the number of plants and measured the total fresh weight for estimating the potential biomass yield. The average height was determined on a sample of twenty individuals from each plot. Finally, plants for plot were selected as representative and analyzed for biomass partitioning considering the seeds, the infructescence (after manual threshing) and the stem. The seeds were separated manually from the remaining parts: after a first beating against a rigid surface, the infructescence was hand-grained simulating the work of combine harvester, while being careful not to remove green seeds that would not be separated by mechanical threshing. All the weights were determined with the precision balance Mettler Toledo (model PM460 Milan, Italy).

The harvest was performed at the end of October using the combined harvester New Holland CR9080 (

Table 1. Technical specifications and working set up of the combine harvester (CH).

Parameters	Value
Model	New Holland CR 9080
Engine power in kW	380
Rotor diameter in mm	559
Rotor length in mm	2638
Rotor speed in rpm	750
Fan Speed in rpm	660
Concave clearance in mm	45
Upper sieve clearance in mm	9
Lower sieve clearance in mm	5
Header Model	DBF TI
Rows (n°)	12
Length in m	5.75
Picker type	belts
Mower type	disc
Weight in kg	1950

). The machine was equipped with a sunflower header, with two counterrotating belts for a row, and with a separator developed by CREA for the recovery of threshing residues aimed at the extraction of secondary metabolites. The separator consisted of a hydraulic engine that rotates a propeller aspirating the products falling from the discharge valves and deviated them in a tube for discharging in windrow or in specific unloading bags.

The performance of the combine harvester was estimated according to the ASABE standards [25,26]. Considering that the machine operates in the field, it is important to determine its effective field capacity, which can be expressed as area capacity (Ca):

Ca = (S W Ef)/10

(1)

where Ca = area capacity (ha h⁻¹); S = field speed (km h⁻¹); W = implement working width (m); Ef = field efficiency (decimal). The yield efficiency accounts for the failure to utilize the theoretical operating width of the machine, the time lost due to the ability of the operator, and the field characteristics.

Following [24], during the harvest we estimated the amount of seed loss. To this aim, at the beginning of infructescence ripening, 17 plastic trays (1000 × 160 mm) were arranged in three groups within the machine working area (Figure 1): seven trays were positioned within the wheel track (B sector); five trays were displaced in correspondence of each side of the header (A and C sectors). The central position was used for monitoring mowing and threshing losses, whereas the side positions were used for estimating the mowing losses. The measurement was performed in three replicates corresponding to three different areas of the field placed at 60 m from field heads.

Seeds losses for dehiscence in the field were evaluated before the harvesting at the beginning (5th) and at the end (25th) of October, by the calculation of the number of seeds collected in each tray. Seed losses as kg ha⁻¹ were re-calculated starting from the average weight of 1000 seeds and the number of seeds recovered from the soil surface unit. The average weight of 1000 seeds was determined from dried and cleaned seeds distinguishing for two-class size: >2 mm and <2 mm by manual sieving. To determine the weight, each sample was mixed before the measurement and then 500 random seeds were weighted with a precision balance Mettler Toledo (model PM460 Milan, Italy). The measures were made in triplicate. All the samples were dried in the oven at 65 °C until achieving a constant dry weight without germinative energy damages.

2.3. Seeds and MOG Preservation

The harvested seed and the threshing residues, identified as Material Other Grain (MOG), were immediately sent to the drying plant (

Table 2. Characteristic of the drying plant.

Parameters	Value
Engine
Make/Model	CIMME X68 006330 E4 RD
Engine power in kW	7.5
Rotor speed in rpm	1450
Flow rate range in m3 min−1	125–355
Temperature range in °C	15–80
Heat generator
Model	Tecflam VDME30LM
Fuel type	LPG
Heat power in kcal h−1	180,000
Pressure in mbar	300
Low calorific value in kcal	2300
Output max in N m3 h−1	8
Generator’s fan
Power engine in kW	0.37
Rotor speed in rpm	2805

) of a seed company for cleaning and for stabilizing the moisture content. The drying process, which started 2 h after the harvest, was carried out at a temperature of 35 °C with high pressure dehumidified air to avoid any compaction and damages of the biomass. The duration of the drying process was 13 h. The main features of drying plant for sowing seeds are shown in Table 2.

After the operation of harvesting, cleaning and stabilization, the seeds were tested for the germination energy according to the Italian Ministry of Agriculture protocol “official methods of seed analysis” [27].

2.4. Statistics

To evaluate the seed losses of the combine harvester, we tested the null hypotheses that the seed losses were similar for each sector using the one-way ANOVA of the PAST software version 3.22 (2018, Øyvind Hammer, University of Oslo, Norway, https://www.nhm.uio.no/english/research/resources/past/, accessed on 30 September 2022). The B sector monitored mowing and threshing losses, whereas the A and C sectors estimated the mowing losses (Figure 1). Data deviated for normality after the Shapiro-Wilk test. Thus, we use the Kruskall-Wallis test for non-parametric ANOVA and Dunn’s test for the mean separation.

3. Results

3.1. Hemp Field

The site was characterized by a Mediterranean climate with rainy winters and hot, dry summers. Trends over the past ten years for the studied site did not show marked differences for temperatures, while average annual precipitation was fluctuating and difficult to predict (Figure 2). During years with low rainfall, e.g., 2006 and 2011, attainment of regular hemp growth required the use of emergency irrigation. In contrast, 2016–2018 saw an increase in rainfall toward maximum values.

The minimum and maximum temperatures in the month of October 2018 can be considered on the average for the period (Figure 3). The rainfall was concentrated during the winter months, but several spots of rain occurred just before the harvest period forcing the postponement of the harvest to 25 October.

It must be kept in mind that the excessive water content of the biomass lowers the harvester performance and increases the risk of both overload and fiber wrapping in the machinery parts. High water content also reduces the time of storage of the fresh biomass before the drying process to avoid the risk of fermentation [28].

The data collected on the hand-harvested plot returned the best yield scenario, in line with the data reported in previous works [3,28], without considering the losses occurring during the mechanized harvest. As showed in

Table 3. Yield components.

Parameter	U.M.	Average	Standard Deviation
Plant density	(n. m−2)	71.3	24.0
Plant height	(cm)	173.33	11.8
Total fresh biomass yield	(Mg ha−1)	18.8	3.3
Infructescence fresh weight	(Mg ha−1)	6.4	0.4
Moisture content of the infructescence	(%)	64.2	1.0
Seed yield	(Mg ha−1 FW)	2.0	0.2
Moisture content of the seed	(%)	24.8	2.0
Seed yield	(Mg ha−1 DW)	1.5	0.1
Seeds > 2 mm	(%)	77.7	3.4
Seeds < 2 mm	(%)	11.0	2.6
Impurity	(%)	11.3	1.0
Seed (>2 mm) yield	(Mg ha−1 DW)	1.14	0.16
Dry weight of 1000 seeds	(g)	15.9	0.7

, hemp is characterized by high intraspecific variability and competition [12], particularly in the dioecious varieties, resulting in large differences within the same area in terms of plant density. This characteristic is accentuated in a second crop sowing. The plant density varied from 49 to 95 plants m⁻² but did not negatively affect the total biomass yield. The biomass of panicles flowing through the harvester was estimated to be 6.4 ± 0.4 Mg ha⁻¹ (FW) with a humidity content of 64.2 ± 0.4% and corresponded to 34% of the whole total fresh biomass (18.8 ± 3.3 Mg ha⁻¹). The total fresh seed yield was 2.0 ± 0.2 Mg ha⁻¹ with 24.8 ± 2.0% of moisture. Of the total seed yield, 77.7 ± 3.4% were seeds suitable to be used as propagation material (diameter greater than 2 mm), corresponding to a yield of 1.14 ± 0.16 Mg ha⁻¹.

Our attempt to shorten the hemp cycle affected the plant height which was 173 cm on average, a value lower than the hemp sowed at correct time [10,11,29] and comparable in some cases with postponed sowing reported elsewhere [15,24].

3.2. Harvesting Performance

At harvest (25 October), the crop was in an advanced stage of development without any problems of lodging, brown leaflet and the trace of loosened seeds at the bottom of the panicle. Considering that in the single hand-harvested plots the plant height was on average 172.3 ± 22.3 cm, the header height and, consequently, the cutting height was set up at 130 cm. In few cases of lower plant height, the header height was lowered accordingly, remaining, however, always higher than 1 m. The working length of the header was reduced 0.5 m until 5.25 m due to operator’s problems to guarantee the full feeding of the header and to reduce biomass feeding and clogging risk. The gripping system of the double belt showed good efficiency of the head, guaranteeing the tightness of the cut tops up to the threshing system of the combine and avoiding the fall of the inflorescences on the ground. The combine harvester worked without setbacks or machine downtimes with an effective working speed of 3.5 km h⁻¹. The effective working time was 57% of the total working time. The remaining accessory time included 24% of unloading time, while the time for turning and adjustment (including the check of machine elements) accounted for 8% and 11%, respectively (Figure 4). The prolonged unloading time must be attributed to the high-water content of the biomass (60%) and to the difficulty of feeding the unloading system of the combine. The field efficiency (Ca) registered in this work was 1.14 ha h⁻¹, a value comparable with the results of similar studies [28,30]. However, it should be kept in mind the peculiar aspect of our work where the threshing of the seeds was carried out simultaneously to the harvest operation.

3.3. Seeds Losses and MOG Preservation

The monitoring of seed dehiscence (

Table 4. Natural and mechanical seed-losses (mean ± s.d.). Different letters indicate a statistical difference for p < 0.05 after the Dunn’s test.

Natural dehiscence on the 5th of October	248.0 ± 32.2
Natural dehiscence on the 25th of October	590.6 ± 210.3
Harvester’s dehiscence on the 25th of October
Sector A	34.8 ± 1.2 b
Sector B	39.3 ± 0.9 a
Sector C	35.0 ± 2.2 b
Mean	36.0 ± 2.8

) highlighted the importance of the identification of the correct harvest timing within the window of the gradual seed ripening in hemp panicle. Seed losses on the ground were 248.0 ± 32.2 kg ha⁻¹ (first survey) and 590.6 ± 210.3 kg ha⁻¹ (second survey). The rise in natural dehiscence was due to the progressive ripening of the seeds in presence of adverse weather condition. It should be also underlined the increase of standard deviation in the second survey indicating that, associated to the temporal variability, there was an additional spatial variability of the losses. The combined harvester losses, from three monitored sectors described in Figure 1 were on average 36.0 ± 2.8 kg ha⁻¹, accounting for 4.3% of the harvestable seed yield. The results showed the lower incidence of mower losses (sector A and C, Figure 1) with respect to the threshing losses (sector B, Figure 1): the difference in seed losses between the threshing part (sector B) and both the sides (sectors A and C) of the head was around 4.4 kg ha⁻¹ and was statistically significant. The head losses were 88.8% of the total mechanical harvesting losses.

Soon after the harvest, both seeds and by-products were delivered to a seed company operating close to the field and left in a dryer plant for 13 h at low temperature to avoid biomass deterioration (i.e., decline of seed germinability and enhancement of secondary metabolites stability). The seed yield was 523 kg (FW basis) corresponding to 331 kg on DW basis. With the presence of 10.8% of MOG, the cleaned seed yield was 295.3 kg (DW basis). On the other side, starting from 296 kg of fresh threshing residues the yield of by-products was 183 kg (DW basis). Therefore, at the hectare level, the estimated seed yield was 794.1 kg ha⁻¹ (DW basis), in addition to 492.20 kg ha⁻¹ (DW basis) of high-value threshing residues. It is important to underline as such results were achieved thanks to the separation system developed by CREA which allowed for the recovery of the threshing residues, which would have been otherwise lost in the ground. After drying, the seed germinability was still high (88%) and the weight of 1000 seeds (15.9 ± 0.7 g) was comparable with those achieved by the seed company.

The high humidity contents of both the seeds and the MOG by-products, and the increased seed losses for natural dehiscence are mainly attributable to the advanced harvesting season (October) and to the rainfalls that have occurred.

4. Discussion

A first insight from the present and other studies concerns the limiting effect that a two-month reduction (from 6 to 4) in the vegetative cycle following delayed sowing has on plant height. The late sowing had favorably reduced the plant height as observed also in other works [12,15,24]. In our study the plants remained abundantly below 2 m, contrarily to the data reported in Italy for the Po valley [29] and Friuli Venezia Giulia region [10] where the plants grown from April to September (6 months) exceeded 2 m [29] up to 2.3 m [10]. However, an exception can occur and plant height lower than 1.8 m has been observed by [10] in 2016. In a Greek experience on a three-year test with a reduced cycle from April to June, the plant height was lower than 1.8 m [31]. Within the approach of late sowing, a key factor is played by the choice of the harvest date. As demonstrated by [24], there is a tendency of plant height to increase as the harvest time is delayed. Thus, to maximize the effect of the height reduction associated with the late sowing, compatibility with favorable weather conditions is advisable to anticipate the harvest. However, other aspects such as weed control and intraspecific competition must also be considered when managing hemp as a second crop.

The increasing reduction in plant height only marginally affects seed production as the yield remains at acceptable values. Delaying the sowing of hemp compared to the ordinary cycle has yielded interesting results showing that late sowing does not affect the production and quality of the seed yield. The outcome agrees with the data reported by [15] in northeast Italy. The mean seed yield (0.78 t ha⁻¹) was little lower than in the present study, and, very importantly, the authors observed as the delayed sowing was positively correlated with seed yield and seed oil content. However, in a Belgian experience [16] the seed yield was higher (between 0.92 and 1.75 t ha⁻¹) but decreased accordingly to the delay of sowing date and plant of the late sowing had shorter inflorescences than plants from the early sowings. In this case the temperatures in the phase flowering-harvesting were lower for the treatment of late sowing (13.4 °C in mid-June). The stage of seed filling is susceptible to environmental constraints as heat or cold stress. At our latitude, delayed sowing can avoid heat stress during the grain filling, shifting the effect to the emergence-seedlings stage [15]. For seed yield, delayed sowing can be considered a useful approach for dual purpose as long as the genotype-environment interaction is evaluated.

The plant height reduction determined by late sowing can favorably affect the mechanical harvesting with the combines available for the most diffused crops. The field efficiency was 1.14 ha h⁻¹ a value comparable with the performances observed in traditional combine modified for seed harvesting of crop structurally resembling the hemp as cardoon (1.27 ha h⁻¹) [32]. Few works are available on the working performance of hemp seed harvesting. In these studies, the field efficiency varied from 1.41 ha h⁻¹ [8] to 2 ha h⁻¹ [33] until 2.9 ha h⁻¹ [34]. However, the reader must be aware that such results have been obtained working with prototypes [8], high-value machines designed for large-scale hemp cultivation [34] or modified combine equipped with a double head for hemp seeds and fibers [33]. Instead, the present study followed the approach of fitting the available conventional mechanization for sowing seed production, From this point of view, the outcomes demonstrated the feasibility in using machines and technologies developed for other crops and immediately available. Moreover, the (technically) simple integration of a specific separation system developed by CREA allowed for the recovery of threshing residues (mainly leaves and floral residues) having an interest for chemical industries. This solution too has been already addressed, but, again, by high-value machines [34] or by combine equipped with a double head [33] without considering to rely on the available farm mechanization. Of course, we can consider such a strategy not “specific” and just applicable to produce hemp seeds in cultivation where the plant height is below 1.5-1.6 m. Although this is an adjustment of the agrotechnique rather than a mechanical innovation, the final result (the reduction in the plant height) represents a valid support for the mechanization of hemp seed harvesting, while waiting for more specific solutions not available yet. Considering the scarcity of suitable mechanical solutions for hemp seed, the approach reduces the need of specific investments in machines hardly amortizable over short periods for a single crop.

Another important aspect of the supply chain concerns the timely stabilization of the harvested product. Renewed interest in hemp has led to a demand for facilities capable of handling and storing high-quality seed with a high degree of stability and germinability. This is mainly true for late maturing crops with wet grain and, as the case of hemp, high oil content (25-30%), the main contributor to the risk of oxidation and rancidity [35,36]. As stated by [36], high quality in harvested hemp seed is assured by an immediate aeration within 3–4 h from harvest. The lowering of seed moisture is required to allow for the preservation of hemp seeds [37,38,39]. As stated by [38], commercial storage of hemp seed in a viable state for one to several seasons requires the seed drying until 8% but preferably to 6% moisture content. In this condition, soon after the drying the germinability remains between 85 and 95%, provided that storage temperature levels not exceeding 15 °C are maintained [37,38,39]. Such an indication matched the route followed in the present study where seeds and by-products were dried by a plant of a seed company immediately after the harvesting producing stabilized seed with 88% of germinability. A logic corollary of the experience is that a fast stabilization is a key issue of the supply chain. This requires an efficient logistics and the availability of nearby cleaning and stabilization facilities, an aspect that must always be carefully planned. Closely related to stabilization processes is the sizing of the drying plant, which must be thoroughly evaluated based on the quantities of seeds and by-products harvested daily.

Finally, some consideration about the selling price of hemp seed is needed. Presently, there is no Italian market for seed hemp. In general, for the seed of other crops the producer’s price could be around fifty percent of the selling price. Considering for the hemp 4 euros per kilogram, a production of 500 kg, a level easily achievable in Italian area, could bring an additional profit of about 1000 euros ha⁻¹.

5. Conclusions

This work presents an innovative route for integrating agronomic and mechanical aspects aimed at making economically affordable an emerging supply chain such as seed hemp production. The delay of sowing shortened the growth cycle, but, in our experimental conditions, without reducing the plant height. However, the field efficiency (Ca) of harvesting was comparable with similar experiences and the threshing losses were below 5%. Thus, our hypotheses about a better handling of the biomass adopting a delayed sowing were confirmed. The main loss component was still associated to the combine header and the cutting systems, due to the ease of seed detachment from the panicle. Incidentally, the presence of a specific separation system made it possible to recover the MOG opening new possibilities for the utilization of a by-product otherwise considered as waste. The study confirmed also as the availability of an adequate cleaning and stabilization system close to the production site improved the quality of the collected fractions.