Fossil fuels such as  coal,  oil  and  natural  gas  are  non-renewable, as well as environmentally  damaging  energy sources due to their formation (millions of years), consumption rates and polluting potential. They release CO₂ in the atmosphere during their extraction process. The price raise and the negative environmental impacts of the use of fossil fuels have reinforced the interest in using renewable forms of alternative energy. Extensive studies on bioenergy indicate that biomass conversion into energy is an important alternative, since it is a renewable source. Biomass is considered a CO₂ neutralizer because the carbon dioxide released during its combustion is recaptured by the new biomass growth. Thus, biomass is the only alternative energy source able to produce liquid, solid and gaseous fuels to replace the fossil ones (Simacek, 2008; Ohimain et al., 2014; Oliveira, et al., 2015b).

Elephant grass is a perennial C4-metabolism poaceae acknowledged as energy crop due to its advantages: fast growth, disease resistance, adaptability, little handling and easy propagation (Tsai, 2009). In addition, it is highly efficient in fixing the CO₂ from the atmosphere and has the potential to be used in biofuel production or in direct combustion (Rossi et al., 2014).

Nitrogen (N) and potassium (K) stand out among the macronutrients because of the key role they play in plant nutrition. Nitrogen is an essential constituent of proteins and it directly interferes in the photosynthetic process due to its participation in the chlorophyll molecule. Potassium is the cation showing the highest concentration in plants; this nutrient has relevant physiological and metabolic functions such as enzyme activation, photosynthesis and translocation of assimilates, as well as nitrogen uptake and protein synthesis. Thus, nitrogen and potassium are limiting factors for systems that make intensive use of cultivated soils (Taiz and Zeiger, 2012).

Nitrogen fertilization is the main N addition vehicle and one of the most important inputs in agricultural systems due to its increased performance in plant productivity (Lopes, 2007). Since elephant grass is a high productivity species, it is worth considering that its nutrient requirements are related to the production potential of cutting-system crops. According to Mistura et al. (2006), elephant grass responds to increasing nitrogen levels. The authors also highlight that potassium fertilization is of great importance, mainly when the plant is grown in cutting systems. Potassium removal due to plant cutting and transportation to other areas, other than the production site, often results in major nutritional imbalance issues in the soil.

The aim of the current study is to assess eight genotypes of elephant grass (Cubano Pinda, Vruckwona, IAC-Campinas, Capim Cana D'África, Cameroon,  CPAC, IJ 7139 and BAG-86) with energy potential subjected to three nitrogen doses (N1= 400, N2 = 1000 and N3 = 1600 kg N ha^-1) and to two potassium doses (K1 = 200 and K2 = 500 kg K₂O ha^-1), under the soil and climate conditions of Campos dos Gotacazes in  the Northern region of Rio de Janeiro State, by analyzing the morphoagronomic traits.

Location and experimental design

The experiment was conducted at the Agro-Energy and Waste Utilization State Research Center of PESAGRO located in Campos dos Goytacazes County-RJ, Brazil (latitude 21° 44’ 43’’ S; longitude 41° 18’ 29’’ W; 10 m altitude; datum WGS84). The climate in the region is hot and humid, and the mean annual temperature is 22.7°C. The results of the soil analysis conducted in the experimental site on January 8^th, 2014, were: water pH = 5.7; P = 7 mg dm⁻³; K = 121 mg dm⁻³; Ca, Mg, Al, H + Al and Na =  3.8, 2.5, 3.6 and 0.1 cmol dm⁻³; CEC = 10.2; and organic matter = 26.5 g dm⁻³.

The experiment was arranged in a randomized block design with three repetitions, according to a split-plot factorial scheme comprising three factors: Factor 1 (plots): genotypes - 8 clones; Factor 2: N - three doses (400, 1000 and 1600 kg N ha^-1); and Factor 3: K - two doses (200 and 500 kg K₂O ha^-1) (sub-plots). Each block comprised of 8 plots. The plots corresponded to parallel rows (12 m-long) spaced 1.5 m from each other. Each row was planted with one of the eight genotypes. The plot was divided into 6 sub-plots (2 m each) wherein the treatments corresponding to six N and K doses (400 x 200, 1000 x 200, 1600 x 200, 400 x 500, 1000 x 500 and 1600 x 500 kg ha^-1 of N x K) combinations were applied. An area of  1.5 m² was used to collect the samples to be analyzed through laboratory procedures, the central length (1.0 m ) of each sub-plot was taken into account.

The experiment was conducted on February 12^th, 2014. The soil was conventionally prepared using a harrowing grid and cultivation furrows (10 cm-deep) were opened. Culm fragments were arranged with 100 kg ha^-1 of simple superphosphate (P₂O₅) within the furrows, in a single row. Thirty days after the planting, 25 kg ha^-1 of urea (CH₄N₂O) and potassium chloride (KCl) were applied to the furrows.

The genotypes used in the experiment came from the Active Elephant Grass Germplasm Bank (AEG GB) of Norte Fluminense Darcy Ribeiro State University, located in Campos do Goytacazes County – RJ, Brazil. The genotypes were selected through visual analysis based on late flowering. Such property is desirable for long-term elephant grass cultivation aiming at dry matter production. The following genotypes were selected: Cubano Pinda (G1), Vruckwona (G2), IAC-Campinas (G3), Capim Cana D'África (G4), Cameroon (G5), CPAC (G6), IJ 7139 (G7) and BAG-86 (G8).

The standadized cut was performed in the culm base on March 29^th, 2014 (45 days after planting) in order to enable the uniform growth of the shoots. Two assessment cycles were held (two cuts after a year of growth). The fertilization was split in 6 applications in each   assessment   cycle,   according   to   the   rainfall.   The   first assessment cut took place on March 10^th, 2015 and the second one, on March 15^th, 2016. The experiment was conducted under natural conditions, without irrigation.

The following morphoagronomic traits were assessed before each cut: number of tillers per meter (NT), which were found by counting the number of tillers in the 1.5 m² area of the sub-plot; mean plant height (H), based on the mean height of the plants in the sub-plot, expressed in m and measured using a graduated scale; mean culm diameter (CD), expressed in cm and measured 10 cm from the ground level, using a digital caliper; mean leaf blade width (LBW), expressed in cm, based on the mean width of the center of three leaves.

The collected samples were fractionated, packed in paper bags and weighed after cutting. Subsequently, they were dried in a forced air circulation oven at 65°C. Seventy-two (72) hours later, the samples were removed in order to be weighed and to obtain air-dried samples (ADS). The ADSs were ground in Wiley mill (2 mm sieve). Two grams (2 g) of it were dried in an oven at 105°C, for 16 h, and then weighed in order to obtain the oven-dried samples (ODS). The dry matter rate DMR was estimated through the ratio between the ADS and ODS values. The dry matter production (DMP) resulted from the product between fresh matter production and DMR, and it was converted to t ha^-1.

Statistical analysis

The statistical analyses were performed in the Genes software (Cruz, 2013). The variance analysis of the traits assessed in the statistical randomized block design used the following statistical model, according to a split-plot factorial design:

Y_ijkl= µ + B_l + G_i + É›_a + N_j + G_iN_j + K_k + G_iK_k + N_jK_k + G_iN_jK_k + É›_b

Where, Y_ijkl = value of the i^th genotype, at the j^th nitrogen dose, at the k^th potassium dose and in the l^th block; μ = overall mean of the experiment; G_i = effect of the i^th genotype; B_l = effect of the l^th block; ε_a = effect of the error a, associated with the i^th genotype in the l-^th block; N_j = effect of the j-^th nitrogen dose; G_iN_j = effect of the interaction between the i^th genotype and the j^th nitrogen dose; K_k = effect of the k^th potassium dose; G_iK_k = effect of the interaction between the i^th genotype and the k^th potassium dose; N_jK_k = effect of the interaction between the j^th nitrogen dose and the k^th potassium dose; G_iN_jK_k = effect of the interaction between the i^th genotype and the j^th nitrogen dose and the k^th potassium dose; ε_b = effect of the error b, associated with the i^th genotype, with the j^th nitrogen dose and with the k^th potassium dose in the l^th block.

A multiple comparison test was conducted in order to study the contrasts between the means of the N factor within the K factor after the individual variance analysis, according to Tukey’s method, at 5% significance level.

The G factor had no effect (P > 0.05) on any of the traits assessed in the two cuts. Only the N factor had effect on the DMP trait in the first year of the experiment, whereas all factors (except for the G factor) and interactions had effect on the DMP trait in the second year of the experiment. Only the N factor had statistically significant effect on the DMR trait in the first year of experiment. Only the interaction between K and N had effect on the NT trait, in the second cut. Only the N factor had effect on the H trait, in the first cut; and on the CD trait, in the second year of experiment. Only the interaction between K and G had significant effect on the LBW trait, in the first year of experiment.

High coefficients of variation (CV) were found in most variables. According to Pimentel-Gomes (2000), the CV values cannot be standardized for all cultures since each culture has its own peculiarities, depending on the assessed traits. Thus, these values may be acceptable since the studied traits are ruled by many genes, as well as strongly influenced by the environment.

There was also the option of conducting a study focused on the effects of the N levels within K as it is a tipical procedure in experiments based on the factorial scheme. According to such scheme, a factor must be assessed in constrast to the other one. However, with regard to the characteristic wherein K has presented significant effect, either alone or in interaction, it was observed that the lowest K dose (200 kg ha^-1) was enough to genetrate the best outcomes. 

Table 1 shows that, according to the Tukey’s test (P < 0.05), there was statistically significant difference between at least two genotypes in traits such as DMP, DMR and CD at doses N3, N1 and N3, respectively, in the second cut; and between genotypes of the LBW trait, at doses N1 and N3, in the first cut. There was no statistically significant difference between genotypes in the other traits, in their respective cuts.

With respect to H, the genotypes did not differ from each other when they were subjected to different N levels within K. The general mean height was 3.32 and 3.03 m in the first and second assessment cycles, respectively. The lowest height (2.86 m) was found in the G8 genotype (BAG-86) at dose K1N2 in the second cut. The highest height (3.50 m) was found in the G5 genotype (Cameroon) at dose K1N3 in the first cut. The H trait was important because it was directly correlated with dry matter production (Xia et al., 2010; Menezes et al. 2014). These values are different from those found by Oliveira et al. (2015a), who found mean H 3.54 m when they assessed six genotypes of elephant grass grown under increasing N doses and cut at 9 months of age in Campos dos Goytacazes – RJ.

However, these results are similar to those found by Novo (2015), who obtained general mean H 3.04 m, and individual H means ranging from 2.79 to 3.19 m in elephant grass genotypes subjected to fertilization using increasing N and K₂O doses.

With respect to the CD trait, the G2 genotype (Vruckwona) at dose K1N3 differed from the other genotypes in the second cut, and presented 2.23 cm CD. The general means were 1.68 and 1.69 cm in the first and second cuts, respectively. These values corroborate those found by Rossi et al, (2014), who assessed the morphoagronomic traits of 40 genotypes of elephant grass in Campos dos Goytacazes - RJ. The authors found general CD genotype mean of 1.72 cm in 11-month-old plants, in the third cut. However, Santos et al. (2014) reported higher CD values in the second assessment cut in 300 day-old plants. They found means of 1.80 and 1.78 cm in genotypes subjected to N doses of 500 and 1000 kg ha^-1, respectively. On the other hand, the CD outcomes in the present study are above those found by Oliveira et al. (2015a), who found mean value of 1.59 cm, ranging from 1.27 to 1.83 cm for the herein described characteristic in 6 genotypes of elephant grass assessed for energy purposes at the age of 9 months.

The culm diameter showed positive correlation with dry matter production (Xia et al., 2010) and it had direct effect on such trait (Daher et al., 2004).

The LBW trait is concerned the leaf area used to capture sunlight and, consequently, its photosynthetic capacity. The general LBW means observed in the first and second cuts were 6.06 and 5.11 cm, respectively. Two genotypes shown values above the average. The G3 genotype (IAC-Campinas) has shown mean LBW 7.06 cm at dose K1N1. The G4 genotype (Capim Cana D'África) has shown mean LBW 7.00 cm at dose N3K1. These values are different from those found by Rossi et al. (2014), who reported mean LBW 3.95 cm in 11-month-old genotypes; and also from those observed by Oliveira et al. (2015a), who reported mean LBW 4.71 cm in genotypes fertilized with increasing N doses. 

Overall, the genotypes did not differ from each other when they were assessed for biomass production potential. The lowest K dose (200 kg ha^-1) was enough to generate the best results in characteristics that presented significant effect. The fertilization with increased N amounts did not promote DMP gains, and the lowest N dose (400 kg ha^-1) was sufficient to promote the highest values.

Likewise, as for the other traits assessed in the current study, although there was a genotype that showed statistically significant difference from any other genotype at a particular dose, the increasing N doses in the fertilization did not influence the performance of the genotypes. However, the results were consistent and met the expectations of energy production potential, since they assured the use of eight elephant grass genotypes as alternative biomass production sources.

The authors have not declared any conflict of interests.

The authors thank the Coordination for the Improvement of Higher Education Personnel (CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) for granting the financial support to the present research.