Ethanol Production Potential of Sweet Sorghum in North and Central Ukraine

2.1. Plant Material

A total of 41 Sorghum saccharatum (syn = S. bicolor var. saccharatum) accessions, deposited at the Bioenergetics crops collection in NBG, were used in this study (Table 1); 8 of which were used for additional investigation of bioethanol productivity. Besides, other species of Sorghum genus were also investigated, including one genotype of S. technicum (syn = S. bicolor var. technicum), one S. vulgare (syn = S. bicolor var. vulgare), one S. bicolor, two breeding lines (Breeding line 1 and Breeding line 3) of S. durra (syn = S. bicolor var. durra) and one hybrid line S. durra x saccharatum (syn = S. bicolor var. durra x saccharatum).

2.2. Conditions of Plant Cultivation and Experimental Design

The 34 accessions, used for morphometric measurements (listed in Table 1, excluding ETSSTSF-1; -3; -4; -5; -6; -7, cv. Bilotsukrova), as well as genotypes, used for comparison of different Sorghum species productivity, were grown on experimental plots at the M.M. Hryshko National Botanical Garden of the National Academy of Sciences of Ukraine (50°32’N, 30°33’E). The plot size was 4 m long and 4 rows wide, while the spacing between the rows was 70 cm and 10 cm between the plants. Each genotype was cultivated on a separate testing plot (n=4). The soil was classified as alfisol (рН 6.1-6.8, measured at 1:2 soil to water ratio). The average air temperature and precipitation in this area were +7.6^oC and 550-560 mm per year. Fertilizer grade 16–16–16 (N–P–K) at the rate of 60 kg/ha was applied to the test plots twice during the planting and 30 days after the planting. The scheme of plot allocation was randomized complete block design (RCBD) for 4 replicates.

The field research was conducted in the Agricultural Experimental Production Station “Hlevakha” of the Institute of Plant Physiology and Genetics of the National Academy of Sciences of Ukraine (50⁰15’N, 30⁰17’E). The soils there were also classified as alfisol (рН 6.0-6.7, measured at 1:2 soil to water ratio), with humus content of 2.5-2.6%. The average air temperature during the growing season (March-November) was +14.6 °C, while the annual average was +7.7 °C. Rainfall during the growing season varied from 432.3 mm to 623.8 mm. All variants of the experiment were cultivated on separate testing plots (n=4).

Breeding lines (ETSSTSF-1; -2; -3; -4; -5; -6; -7) were cultivated to identify their morphological differences and to establish their productivity rates, which were further used for ethanol productivity estimation. Plot sizes and rows/plants spacing were the same, as described above. Fertilizer 16–16–16 (N–P–K) was also applied at the rate of 35 kg/ha to the testing plots during the planting and then again, 30 days after the planting.

Cultivar Botanichnyi was used in a series of experiments, focused on thefertilization effect on plant productivity. In the control variant, no soil fertilization was applied. Bacterial inoculums - biopolycide (spores of Paenibacillus polymyxa, strain 6M) and phosphobacterin (mass spores of Bacillus megaterium), used for plant pre-treatment, were obtained from the Institute of Agroecology of National Academy of Agrarian Sciences of Ukraine. Seeds were pre-treated with biopolycide (0.5-1x10¹² of spores per 1 mL) at the rate of 15 mL per 1 kg of seeds. The inoculum was diluted with water to 1% solution and applied on the day of sowing. Phosphobacterin (1.0-1.5x10⁹ of spores per 1 mL) was applied also prior to sowing at the rate of 2 mL per 1 kg of seeds (inoculum was diluted with water to 0.2% solution). For a combined variant of pre-treatment, inocula were taken at the same rate per kg of seeds and were diluted in 1.5 L of water. For 16–16–16 (N–P–K) soil supplementation variant, the fertilizer (at the rate of 135 kg/ha) was applied to plots on the day of sowing and after 30 days from the planting date.

The effect of different rows/plants spacing was tested also using cv. Botanichnyi. The plants were sown in different manner (spacing between rows x plants): 70x30, 70x20, 70x10, 45x30, 45x20, 45x10, 15x30, 15x20, 15x10 cm. Prior to sowing, the seeds were treated with biopolycide, similarly to the way described above. In all mentioned testings, the scheme of plot allocation was also RCBD for 4 replicates.

2.3. Morphological Data Collection and Plant Productivity

Morphometric data was collected at two main periods: the flowering phase (when more than 50% of plants started flowering and shed pollen) and the seed ripening phase (30 days after the flowering) from 10 randomly chosen intact plants in each plot. The same procedure was repeated during the harvest (30 days after flowering of each cultivar starting approximately in mid-September). Plant height, stalk diameter, leaves and inflorescence size were recorded from 10 randomly chosen plants in each plot. Plant height was measured as a distance from panicle (or inflorescence) tip to the ground. Stem (stalk) diameter was measured at three different positions using a caliper. Plants were harvested manually. To do that, stems were cut at the ground level, and then panicles and leaves were removed. The weight of each plant organ was recorded individually for 10 random plants per plot. The total weight of stripped stalks and total biomass weight were recorded for each plot. Additionally, the weight of panicle (with seeds), the weight of seeds from one panicle, the weight of 1000 seeds and the total weight of seeds collected from each plot were recorded.

The stalks of sweet sorghum were crushed to extract the juice using a roller crusher. Extracted juice was then immediately stored at −20°C for further analysis. Specific gravity was calculated as juice weight divided by the volume of weighted juice. Total soluble solids of the juice were determined using a hand-held refractometer to measure Brix (sugar content in juice) immediately after squeezing of the juice.

2.4. Biochemical Analyses

Samples for biochemical analysis were collected at the above specified sweet sorghum growth stages. Dry matter content was measured for collected stalks. Stems were chopped, milled and immediately weighted. After that, the plant material was dried in a heat-oven at 105°C for 3 days. After drying to the constant mass, the resulting weight was recorded and used to calculate dry matter (as a relation of dry weight to fresh weight, expressed in %).

The total content of reduced sugars in stalks was evaluated by Bertrand’s method (with slight modification) based on the reducing effect of sugar of alkaline solution on tartrate complex with cupric ion. The cuprous oxide formed was dissolved in a warm acid solution of ferric alum. The ferric alum was reduced to FeSO₄, which was titrated against standardized KMnO₄; Cu equivalence was correlated with the table to get the amount of reducing sugar. This method was based on the alkaline solution of the tartrate complex of cupric ion [32]. Data were expressed in % of the weight of fresh stalks.

The amount of energy in the samples was determined with a calorimeter ІКА С 200 (IKA®-Werke GmbH & Co. KG, Germany). Apart from the plant material, collected during field growing, sorghum bagasse was also used to determine the energy value of biomass after the juice pressing (that was done only for genotypes AMBR-2, AMBR-5, cv. Botanichnyi, cv. Bilotsukrova, cv. Medove and cv. Yantar). Measuring of the caloric content was performed as follows: 0.1-0.2 g of dried plant material (measured exactly with the weight figure uploaded onto the calorimeter) was brought in oxygen bomb (decomposition vessel IKA C 5010/5012) and put in distilled water in the calorimeter. The procedure lasted for 13-15 min. Benzoic acid 723 was used for the standardization of the process. Data were expressed in kcal/kg.

2.5. Estimation of Potential Ethanol Yield

Several equations were used to estimate potential ethanol yield from the investigated sweet sorghum genotypes. Prior to calculating the ethanol output, some yield parameters were also estimated, based on the data of the above-described measurements. To determine the approximate yield of sugars (t/ha), the equation, described by Wortmann et al. [33], was used:

(1)

where 0.75 is the production index, introduced to account for potential losses at a particular technological stage.

Since some of the equations required the input of another type of crop productivity data, we used the following two equations, based on the previous ones (1):

(2)

(3)

where producing index (0.866 in both equations (2) and (3)) is derived from 0.75 production coefficient .

We also used three other different equations to estimate the ethanol output more accurately. The first one was successfully used to determine the influence of various crop cultivation factors on ethanol production [34] and originally described by Lipinsky [35]:

(4)

This equation (4) is based on stalk (stems) yield (t/ha) and requires measurements of soluble solids or sugars content, expressed in % w/w of the stems weight. The equation is based on the sugar to ethanol conversion rate (0.65 L of ethanol per 1 kg of sugar) and assumes 85% (0.85 index) of the production efficiency.

Two additional estimation methods were chosen since their accuracy was successfully tested against the real ethanol production experiments [36]. The first one was originally described by Smith et al. [37]:

(5)

This equation (5) requires sugar yield (t/ha) data that we derived by the method, described above in (1). The ethanol/sugar conversion of 0.665 (L/kg) was taken as an average of 0.648 (L/kg) of hexose conversion rate and 0.682 (L/kg) of sucrose conversion rate [36]. The same average conversion rate was also used in the previous estimation method (4). Also, this (5) equation was found to be more accurate when assuming the production efficiency of 80% (0.8 coefficient) [36].

The following equation was originally proposed by Somani et al. in 2003 [38]:

(6)

where SG – specific gravity of juice.

This equation (6) is based on juice productivity (t/ha) that we estimated with the formula (2). The ethanol/sugar conversion rate of 0.53 (L/ha) was used as in equation (5) but with 0.8 production coefficient.

To estimate the potential yield of ethanol, the calculation was done using our proposed equation:

(7)

where sorghum grain yield is expressed in kg/ha, taking into account the content of water-soluble starch in % (w/w).

A reference value of soluble starch content in equation (7) was used as 67.75% w/w, which was a mean rate, suggested in a respective sorghum collection study [39]. The ratio of 0.9 kg glucose per 1kg of starch reflects the efficiency of enzymatic hydrolysis of sorghum starch. This value is also a mean, based on two respective studies on sorghum grain starch hydrolysis and fermentation [40, 41]. Hexose to ethanol conversion of 0.648 (L/kg) was used, as well as an 80% production rate.

A theoretical ethanol yield from sweet sorghum bagasse (SSB) was calculated on the basis of existing SSB-to-bioethanol conversion rates. Therefore, the calculation was done as follows:

(8)

where kg of dry SSB is an amount of stems dry matter yield with a production efficiency coefficient of 0.85.

As reference values of SSB (8) to bioethanol conversion rates determined for technologies with different biomass pre-treatment methods: 163 g EthOH/kg SSB for SO₂-steam explosion and following S. cerevisiae fermentation were used [42, 43], 184 g/kg for acidic pre-treatment and fermentation with genetically engineered S. cerevisiae [44]; 275 g/kg for steam explosion with H₃PO₄ impregnation followed by E. coli SL100 fermentation, were observed [45]. Conversion of bioethanol weight to volume was done at the 0.789 cm³/g density (value for 99.5% bioethanol) [46]. In the final evaluation, a mean value of all SSB bioethanol calculation variants was used.

2.6. Statistical Analysis

All statistical processing of the obtained data was conducted using OriginPro 9.1 software. Deviations of all means were calculated as a standard deviation (SD). To identify the significance of differences in different morphological and productivity parameters between the studied genotypes, one-way ANOVA was used, which included the calculation of Fisher’s least significant differences (LSDs). The LSDs were used to identify homogeneous groups for values of particular morphological or productivity parameters (at p<0.05 significance level). In some cases, SD of means has not been denoted in tables (where it was considered not important for data representation), since one-way ANOVA was applied to all datasets.

Two-way ANOVA was conducted to identify the significance of influence of factor pairs (genotype and sugars content (Brix%) in juice) on the energy value change of sweet sorghum biomass before and after juice pressing (δkcal/kg). For verification of the significance of differences, LSDs were also calculated (at p<0.05 significance level).

Pearson correlation analysis was conducted using the same software and was applied to the data of morphology measurements of 34 accessions (mentioned above) during two plant growth stages (flowering and seed ripening). Correlations were considered significant at p<0.05 level.

3.1. Phenotyping of Sweet Sorghum Genotypes

A morphology screening study of the sweet sorghum diversity was conducted using 34 accessions from the collection of a total of 41, deposited at the Department of Cultural Flora of NBG. Specimens from Europe, Asia, Africa, and America were represented in the gene pool of forms and varieties of S. saccharatum (Table 1, first column). As a result of many years of plant introduction studies, we have been able to obtain seed-reproducing lines of all these genotypes. A gene pool of fully acclimated sweet sorghum lines has been created (Table 1, second column), which in some cases required crossing with local genotypes. This allowed us to create several breeding lines (Table 1, third column) with high adaptability to cultivation conditions and optimal productivity. Individual genotypes were further used as a basis for the creation of new cultivars, including one successfully registered – cv. Botanichniy.

Sweet sorghum is a new high-performance bioenergy crop in Northern and Central Ukraine. It is an annual plant, which passes all stages of organogenesis in one growing season under local climate conditions. The flowering of plants takes place in July-August, while seed ripening occurs from the second half of August to the end of September. The growing season lasts from 120 to 180 days depending on the genotype. Sweet sorghum is a thermophilic plant. Its seeds can germinate at a soil temperature of 10-12°C, while seedlings do not withstand temperatures below 0°C. It can grow and develop well at a temperature of 30-36 °C, easily withstanding high temperatures up to 40°C. Besides, sweet sorghum is a drought-resistant plant, therefore it can easily tolerate precipitation level [47], typical for temperate climate zone (including Forest-Steppe zone). Especially interesting from an environmental point of view is the cultivation of S. saccharatum at an industrial scale for the production of biomass and sugars on the soils contaminated with heavy metals [48].

The conducted screening of sweet sorghum germplasm allowed determining plant height (cm), stems, leaves and panicles weight (g), the total weight of the aboveground mass per plant as well as dry matter content (%) and biomass energy value (kcal/kg). Sorghum plants are able to grow more than 300 cm in height forming long-lanceolate leaves. The root system is fibrous, highly developed, which grows 150-200 cm in depth. Additionally, aerial roots are also formed. Depending on their height, the plants can be separated into three groups: first – 151-200 cm; second - 201-250 cm; third – 251 cm and above. Previously, for the creation of breeding lines, genotypes from the third group were used. During field trials, the highest plants were found in the plots of ETSSTSF-2 (breeding line), accessions AMBR-1-1, AMBR-3, ETSSTSFM-5 and cultivars Prohres, Kinelske-9 and Enerhodar.

Correlational analysis was conducted based on the phenotyping data that resulted from measurements at a flowering phase (Fig. 1) and at a seed ripening phase (Fig. 2). No significant correlations have been established between plant height and other parameters, including the aboveground mass. This may allow to suggest that, probably, stem diameter makes a greater contribution to the biomass accumulation. According to our results, the highest total biomass (aboveground mass) productivity was observed for accessions AMBR-1, AMBR-3, ETSSTSFM-5-1, ETSSTSFRR-2-9; breeding lines ETSSTSF-2, ETSSTSF-2-1, ETSSTSF-9; and cultivars Botanichnyi and Enerhodar. The strongest correlation was observed between total biomass weight and stems weight on both growth stages r=0.993-0.994 at p<0.01. Thus, the biggest biomass weight comes from the stems (approximately 71% at both stages), regardless of the genotype or general productivity of a particular line.

It is worth noting that significant (at p<0.01) correlations between stem and leaves weight were observed at both stages of sweet sorghum growth (r=0.922 at flowering stage and r=0.698 during seed ripening). At the same time, correlation of inflorescence weight with either leaves (r=0.471) or stems (r=0.395) weight was significant at p<0.05, but weak; while during seed ripening, these correlations were much stronger (r=0.624 with leaves and r=0.875 with stems) at low p-value (at p<0.01). Similar patterns of growth parameters were also observed in other countries [5, 22].

The energy value of dry biomass ranged from 2824.5 to 3911 kcal/kg during flowering and from 3394.2 to 3745.1 kcal/kg by the end of the seed ripening period. Interestingly, while flowering, a significant (at p<0.01) negative correlation (r=-0.622) was identified between biomass energy value and the aboveground mass of the plant, while no significant correlation between these parameters was observed. Additionally, no significant correlation was found between energy value and dry matter content during both growth stages.

Dry matter content of biomass, obviously, increased from flowering stage to the end of vegetation stage of plants. At the flowering stage, dry matter content varied from 18.34 to 34.36% and after the end of seeds ripening, it reached from 29.04-60.63%. The highest values were detected within breeding lines ETSSTSF-1-2 (60.63%), ETSSTSF-2-6 (56.41%); and cultivars Progres (55.72%) and Yantar (54.49%). Such high moisture loss could potentially negatively impact the juice output, thus reducing ethanol production efficiency. Cv. Botanichniy contained 44.07% of dry matter, which was very close to the mean value within the analyzed genotypes pool, i.e. 44.8%. On the other hand, these measurements were performed after the period of the supposed harvest of stalks for juice production; therefore, moisture content was expected to be higher at that time. Nevertheless, significant but not very strong negative correlations were observed between dry matter content and the above-ground mass weight (at flowering r =-0.463, p<0.05 and after seeds ripening – r=-0.638, p<0.01). More significant (both at p<0.01), but not stronger, negative correlations were observed between dry matter content and stalk weight, r=-0.513 at flowering and r=-0.626 during seed ripening. In general, loss of moisture by the plants during their vegetation (after flowering) led to a decrease in the total weight of stalks and biomass in general.

Important productivity traits for many crops are quality and quantity of seeds produced per plant. In Table 2, main parameters indicating seed productivity of several sweet sorghum breeding lines are provided. During the field trial, it was recorded that the length of the panicle usually reached 60-70 cm. Seeds of sorghum have an elongated form and differ in color, while the weight of 1000 seeds varies between 18.2 and 23.09 g. Typically, most of the weight of sorghum panicle is due to seeds, which comprise about 78.86% of the total panicle weight. The largest panicle weight (45.9 g) was found in the breeding line ETSSTSF-1, while the smallest (16.4 g) in ETSSTSF-4, respectively. ETSSTSF-1 delivered the highest yield of seeds per panicle (37.9 g, which is 82.57% of the total weight) and the highest weight of 1000 seeds was 23.09 g. Other breeding lines produced smaller panicles of about 30.37 g, containing on average 23.93 g of seeds.

3.2. Sorghum Species and their Bioethanol Producing Potential

Apart from S. saccharatum, there is a number of species, which could also potentially be used as biofuel (ethanol) feedstocks. Most of them feature either comparable or lower biomass productivity than sweet sorghum (Table 3). However, all Sorghum species included in this comparison are now considered to be subspecies or varieties of S. bicolor (The Plant List, http://www.theplantlist.org/tpl1.1/record/kew-443283). Therefore, there is a possibility to obtain ‘interspecies’ crosses, which was demonstrated in the example of the hybrid line (S. durra x saccharatum), obtained out of grain sorghum and sweet sorghum. On the other hand, data of molecular genetic fingerprinting of different Sorghum species suggest that S. bicolor, S. technicum and S. saccharatum are the most closely related; however, their representatives are sufficiently distinctive to form separate groups (clades) for each of these species [49].

The lowest biomass productivity was recorded within S. vulgare – 38.13 t/ha, while the highest one was found for sweet sorghum (cv. Botanichniy) – 82.74 t/ha and S. technicum – 79.86 t/ha. The hybrid plant possessed lower biomass accumulation – 72.96 t/ha. Similarly, as reported above, most of the biomass weight was observed in the stalks (approximately about 71%, regardless of the species). The content of sugars in stalk juice was significantly lower within all sorghum species (12.87-13.94%), except sweet sorghum (19.66-21.41%) and S. durra x saccharatum hybrid (19.1%).

Additionally, potential ethanol yield was estimated for each of the analysed species on the basis of the obtained field trial data (Table 3). The estimated EthOH output was not significantly different for S. bicolor, S. durra and S. technicum and was in the range of 1817.1-1986.9 L/ha. At the same time, sweet sorghum possessed significantly higher values of 3234.9-3538.8 L/ha, while the S. durra x saccharatum hybrid showed an intermediate value of 2635.5 L/ha, comparable to parental species. Other researchers reported similar low ethanol yields from forage sorghum of up to 1051 L/ha only [50]. Such low performance of other sorghums, in terms of ethanol yields, was expected, since these species (or subspecies) were not specifically bred for the accumulation of high levels of sugars.

Estimated grain yield values have also been identified for some of the investigated species (Table 4a). The highest productivity was received by Breeding form 3 of S. durra – 6.3 t/ha and sweet sorghum cv. Botanichniy – 6.4 t/ha, S. vulgare and Breeding form 1 produced 5.5 t/ha each. Concerning the conversion of grain to ethanol, the potential alcohol output can be increased by additional 1738.5-2023 L/ha. Regarding the general possibility of ethanol production, sweet sorghum holds the leading position compared to other investigated species/subspecies. Otherwise, S. durra x saccharatum hybrid line could be of interest for further breeding improvement.

As reviewed by Anglani, the nutritive value of sorghum is lower than that of other cereals due to the lower content of some amino acids and the presence of antinutritional tannins [51]. For this reason, Sorghum is not very good for breadmaking flour [51]. On the other hand, this opens possibilities for industrial utilization of its grain.

3.3. Approaches to Enhance Sorghum Productivity for Increased Ethanol Output

3.3.1. Breeding of New Sweet Sorghum Forms

The pool of breeding lines was created at NBG on the basis of acclimated sweet sorghum lines (Table 1). The most productive ones were selected for further research. Morphological characteristics of these breeding lines are provided in Table 4b. No significant difference was indicated for the height parameter. Although its mean value was 252.94 cm, lines ETSSTSF-1 and ETSSTSF-3 were slightly higher (271.2 and 267.6 cm, respectively) than other lines. Inflorescence (panicle) length varied from 22.2 cm (ETSSTSF-4) to 36.1 cm (ETSSTSF-1). In general, a bigger panicle can produce more seeds, but from the point of view of juice production, it is more desirable to have a higher plant with a small panicle to increase the size of the stalk. Another aspect that can impact juice yield is the stem diameter, which was found to be approximately 11.31 mm. The differences of this parameter were not significant for almost all investigated breeding lines, except the ETSSTSF-6. The highest number of internodes was detected in plants of ETSSTSF-4, while the difference with others was not significant. The average leaf length was 70.39 cm.

As the next step, ethanol productivity-related traits were evaluated (Table 5). The highest biomass productivity values were demonstrated by ETSSTSF-4 - 109.22 t/ha of total biomass, ETSSTSF-3 and ETSSTSF-5 – at 94.57-95.46 t/ha and ETSSTSF-1 – 76.37 t/ha, respectively. The stalk yield was lower and varied in the range of 29.0-77.5 t/ha. The highest dry matter content (and, therefore, the lowest juice content) was received within ETSSTSF-1 at 48% (dry matter content) and ETSSTSF-3 and ETSSTSF-6, both at 45%. The maximum juice yield reached 44.5 t/ha, while the highest sugar content was detected within genotypes ETSSTSF-1 (19.8%) and ETSSTSF-4 (16.3%). According to Adinurani et al. [32], the fresh weight of sorghum was 81.27–210 t/ha, the number of internodes was 7.11–11.77, the measured stem height was 181.29–301.28 cm, and the stem diameter was in the range of 1.27–1.40 cm.

Finally, the potential ethanol yield per hectare from each of the investigated breeding lines has been suggested. The estimation has been worked out using three different equations, according to the procedure described above. The mean value of three separate calculation methods of ethanol yield is presented in Table 4a. According to our estimation, the highest potential ethanol output can reach up to 3676.4 L/ha out of the juice of ETSSTSF-4. Lower ethanol amount could be yielded from ETSSTSF-1, ETSSTSF-3, ETSSTSF-5 – in the range of 2486.5-2698 L/ha. Similar calculations for the other genotypes have shown significantly lower levels of ethanol output. The mentioned genotypes that showed highest productivity and high sugar content were used as the basis for breeding of cv. Botanichniy (Fig. 3). The productivity evaluation of cv. Botanichniy is presented in the following sections.

Table 4b.

Genotype	Height, cm	Inflorescence Size, cm	Stem Diameter, mm	Number of Internodes	Leaf Length, cm
ETSSTSF-1	271.2±15.85a	36.1±7.82a	12.7±2.11a	7.0±0.67b	67.8±8.72a
ETSSTSF-2	246.4±18.56a	29.6±2.07a	11.1±4.07a	8.0±0.94b	77.2±6.83a
ETSSTSF-3	267.6±37.55a	31.6±3.63a	13.4±2.99a	8.4±0.84ab	71.2±7.97a
ETSSTSF-4	244.2±13.31a	22.2±3.08b	12.5±0.53a	9.5±0.53a	72.2±2.82a
ETSSTSF-5	245.6±15.18a	24.0±2.71b	10.6±2.50a	8.0±0.67b	59.6±12.38ab
ETSSTSF-6	246.4±14.09a	30.2±5.92ab	7.6±1.43b	6.9±0.74b	66.7±4.52a
ETSSTSF-7	249.2±14.47a	29.8±7.10ab	11.3±3.56a	6.8±0.42b	78.0±7.69a
Mean	252.94±19.97	29.07±6.14	11.31±2.9	7.8±0.93	70.39±8.91

Table 5.

Genotype	Biomass Yield, t/ha	Total Dry Matter Yield, t/ha	Sugar Content in Juice, %	Estimated Ethanol Yield, L/ha
ETSSTSF-1	76.37c	36.86b	19.8a	2698.0b
ETSSTSF-2	53.28d	20.35c	14.8c	1699.4c
ETSSTSF-3	95.46b	43.18a	15.1c	2486.5b
ETSSTSF-4	109.22a	35.76b	16.3b	3676.4a
ETSSTSF-5	94.57b	35.99b	13.9d	2547.1b
ETSSTSF-6	40.85e	18.32c	15.1c	1264.6d
ETSSTSF-7	47.51de	19.25c	13.1d	1312.3d
Mean	73.89±25.8	29.96±9.85	15.4±2.1	2240.6±869.1

3.4. Evaluation of Potential (Bio)Ethanol Productivity

Different methods of ethanol yield estimation can usually give only a general view of the potential productivity of technical alcohol production. The influence of the method of calculation factors (Smith et al. [37], equation (5)) on the final results is presented in Fig. (4a). The highest recorded productivity rate in this study was 9.5 t/ha, which was chosen as the reference value of sugar yield. This particular equation is based on the sugar productivity value, to which a conversion index is applied (0.665 L ethanol per 1 kg of sugars).

Also, the production efficiency coefficient was assumed to be 0.8 (80% of production efficiency), which takes into account losses during the real production process. Respectively, having applied the production efficiency rate (0.8) to the conversion rate (0.665 L/kg), the resulting conversion rate was 0.532 (L ethanol per 1 kg of sugars). Such an approach was used in equation (6) proposed by Somani et al. [38]. However, the conversion rate of 0.665 L/kg is a mean value of corresponding rates for sucrose (0.682 L/kg) and hexose (0.648 L/kg). The influence of different conversion rates on the final estimation is shown in Fig. (4a). Thus, the mean conversion rate suggests that the potential ethanol yield can be 5054 L/ha. Likewise, if the hexose coefficient is applied, this value can be 4924.8 L/ha, and in the case of sucrose, the value can become 5183.2 L/ha. Although such divergences may not seem to be significant in calculations, they could have serious implications in the case of estimating the efficiency of real large-scale production. It can therefore be concluded that for the sake of more accurate estimation, sugars composition for particular sweet sorghum genotypes should be investigated. This would enable the ultimate correction of the coefficients. Besides that, the equation (5) proposed by Smith et al. [37] was tested against real fermentation experiments (under laboratory conditions with 250 mL of juice) and proved to be accurate enough for estimations in the present form [36]. Otherwise, larger-scale production may require an even higher level of accuracy in calculations.

In Fig. (4b), the impact of sorghum productivity parameters on the estimation of final ethanol output is shown. This figure provides a graphical interpretation of the equation (6) proposed by Somani et al. [38]. This calculation method takes into account the fact that not all soluble substances in the juice are sugars, identifying their content as Brix%-3. However, usage of the constant value (of -3) for normalization of sugars content level leads to a significant distortion of the estimated ethanol yield. Thus, the constant value subtracted from low Brix% makes a more significant impact on the ultimate calculation result than the same operation done for high Brix% values (approx. ≥20%). Similar peculiarities of this estimation method were also discussed by Bunphan et al. [36]. This effect will be the most tangible in case of very high juice yields (blue zone on Fig. 4b). If extremely high ethanol yield values are considered, the data obtained with this equation will show linear dependence. Indeed, such high values (over 5000-6000 L/ha of ethanol) could take place extremely rarely or almost impossible in practice. The maximal actual ethanol yield in a soil supplementation experiment was 4941 L/ha, while the calculated mean rate was 4041 L/ha (excluding control- without soil supplementation). At the same time, the mean value for other different S. bicolor subspecies (or other species) was only 1396 L/ha caused by a concurrent combination of low juice yield and low sugars content in it. However, this mean figure excludes the value of S. durra x saccharatum hybrid, which demonstrates the average productivity compared to its parental species (subspecies).

Similar to the previous equation (5), this calculation method (6) was also tested against real ethanol production experiments and has shown a desirable accuracy [36]. In general, both equations (5 and 6) provided almost similar values of ethanol yield, except in the cases where sugar content was high (≥18%) as well as juice yield (over 30 t/ha). The difference was only 1.4-3.6% when the equation (6) provided higher values. In the cases when the actual input productivity data was higher than indicated, this difference rose to 4.8-5.7% or more. When the juice yield was below 20 t/ha and, at the same time, the sugar content was lower than 15%, equation (6) gave lower ethanol yield values than equation (5). The difference in this case was 3-6.1%. However, both equations (5) and (6) provided significantly lower values than equation (4), proposed by Lipinsky [35], which was based on stalks yield. The equation by Lipinsky [35] assumed higher production efficiency (85%) and used sugars % in stalks for calculation. It also used the same conversion rate of 0.665 L/kg, but did not take into account the potential losses during juice pressing since that formula dealt with stalk productivity parameters instead of juice or sugar yields. However, it also provided linear dependence of data derived from it, similarly to equation (5), thus dealing equally well with high or low yield values.

The equation (4), based on the stalks yield, was not tested in ethanol production experiments. Besides, the calculations by that method provided much higher ethanol yields – up to 5000-5500 L/ha [34] or even higher – above 7000 L/ha of ethanol form juice [66], what seemed to be, probably, rather overoptimistic, although theoretically possible. In the present study, the equation by Lipinsky [35] also provided significantly higher values of ethanol yield, compared to the equations (5) and (6). At the same time, the practical verification of equations’ accuracy, mentioned previously [36], showed significantly lower values of ethanol yields, calculated via equation (6) – only up to 3000 L/ha. However, these studies present the results of sorghum productivity trials under diverse climatic conditions (precipitation, mean temperature, etc.) that significantly impacted upon the resulting data. Moreover, humidity, precipitation and temperature fluctuations can also significantly impact on the sweet sorghum productivity since this crop is well resistant to semi-low watering conditions [47].

Thus, in the present study, the ethanol yield was calculated as a mean of the three mentioned equations (4-6). The value of 4416.4 L/ha was established as a final reference value of ethanol yield from sweet sorghum juice that is possible to obtain under the climate conditions of Northern-Central Ukraine (Forest-Steppe zone, temperate climate zone). This value is the mean of ethanol yield from the soil supplementation experiments (Table 8), except for the control.

Additionally to all the above-mentioned peculiarities of ethanol production rates estimation from sorghum juice, the great perspective is held by co-conversion of the whole plant biomass into ethanol and bioethanol. Potential yields may rise up to 10500 L/ha [45] or 11200 L/ha [67]. Fig. (4b) indicates a red zone of ethanol yields that can be achieved if sorghum juice, grains and bagasse are converted into ethanol. Such an approach could allow obtaining a huge amount of ethyl alcohol, compared with a conventional sweet sorghum juice fermentation, since the latter will require a rapid increase in plant productivity that cannot realistically be achieved at present.

Our estimation of potential co-conversion of different sorghum raw materials into ethanol is shown in Fig. (5). The ethanol yield from sorghum grain was determined at the level of 1973.2 L/ha (mean from the soil supplementation experiments, excluding control, Table 9), which was estimated by equation (7). The resulting yield of ethyl alcohol from both grain and juice feedstock was 6389.6 L/ha, which is very close to the values determined by Castro et al. for cv. UF15 [45].

To evaluate the potential amount of bioethanol that can be produced from sweet sorghum bagasse, another calculation method has been applied (equation 8), which is based on an average rate of conversion of dry SSB into bioethanol, involving various pre-treatment procedures. SSB amount was estimated on the basis of the data resulting from soil supplementation experiments (excluding control). Estimation of bioethanol yield suggested that it was possible to obtain additional 5033.5 L/ha of EthOH from SSB. However, the SSB/bioethanol conversion rates were very different for technologies involving various biomass pre-treatment methods. Thus, bioethanol production with acidic pre-treatment [44] may allow to produce only 4467.1 L/ha, while SO₂ steam explosion pre-treatment [42, 43] will result in even lower yields – 3957.3 L/ha. A novel approach that involves steam explosion impregnated with H₃PO₄ pre-treatment and following transgenic E. coli fermentation [45] could provide significantly higher bioethanol yields – up to 6676.4 L/ha. On the other hand, not all the technologies mentioned can be easily applied for industrial-scale production with the same efficiency.

According to our estimation, the total (bio)ethanol output that can be produced from the combination of juice+grain+SSB is 11423 L/ha. The calculated value is not significantly different from other mentioned estimations suggesting 10500-11200 L/ha of ethanol yield, resulting from such co-conversion approach [45, 67]. Moreover, another recent research revealed that it was possible to produce up to 10344 L/ha of bioethanol from sweet sorghum stems only if juice was not preliminary pressed [68]. Such evaluation is also consistent with our elaborations and other research findings since adding grain ethanol to the value estimated by Byrt et al. [68] could increase the calculated value by only 1000-2000 L/ha. Therefore, the resulting total amount of estimated (bio)ethanol yield will be close to either our estimation or the ones suggested by Barcelos et al. [67] and Castro et al. [45].

Apart from this, obtaining ethanol from a combination of juice+grain seems to be more realistic for industrial-scale production, since bioethanol production may have hardly-predictable economic efficiency. However, the productivity of SSB conversion into bioethanol can be further improved by using xylose-fermenting organisms, such as Trametes versicolor [69, 70] or Hansenula polymorpha [71]. Another approach, which could allow increasing the efficiency of ethanol production from sweet sorghum is associated with the genetic engineering of this species. As an example, manipulations with SWEET genes could be considered to enhance sucrose accumulation in stalks [72]. Mutant sorghum lines are also regarded as promising biofuel feedstocks with increased sugars content (ARS14 mutants) [73] or reduced lignin content (bmr mutant hybrids) [74].

Furthermore, recent advances in molecular genetics allow obtaining lipid-producing (in biomass) sorghum [75]. Co-conversion approaches for biofuel obtained from lipid-producing sorghum have already been proposed [76]. Additionally, a technology involving lipids transesterification with bioethanol for biodiesel production has been investigated [77]. As reported earlier [24], dry, pre-treated sorghum bagasse can be co-fermented with juice and with other biomass feedstocks such as finger millet (Eleusine coracana) [78]. Alternatively, sweet sorghum bioethanol can serve as an energy source for the production of conventional juice and grain ethanol. Finally, conventional breeding also offers promising opportunities. For instance, wild Sorghum species can serve as a donor of genes (allele variants) associated with high-stress tolerance [79].

Also, further improvement of sweet sorghum through its breeding can be extended using the existing broad gene pool, which was evaluated under the present study, as well as other sorghum germplasm, genetic diversity of which has already been characterized [49].