Yield stability of sweet sorghum genotypes for bioenergy production under contrasting temperate and tropical environments

Forty-three sweet sorghum accessions were grown in two contrasting environments; Nigeria (tropical environment) and Denmark (temperate environment). The objectives were to determine the interaction between genotype and environment on grain yield, fresh biomass and stem sugar


Introduction
Sweet sorghum (Sorghum bicolor L.) is a multipurpose crop grown for food, feed and fuel due to its high sugar level in the stem (Regassa and Wortmann, 2014).It is similar to grain sorghum but exhibits rapid growth, higher production, and wider adaptation.It has great potential for ethanol production.As a drought tolerant crop, it remains the most desirable alternative to other cereals.Sweet sorghum accumulates high amount of fermentable sugars in the stem and the ethanol from sweet sorghum is said to be cleaner than ethanol from sugarcane when mixed with gasoline (Belum et al., 2010).It compares well with sugarcane or corn when viewed from the perspective of energy balance between production and available extracted energy.Sweet sorghum produces eight units of energy for every unit of energy invested in its cultivation and production (Amosson et al., 2011).The crushed stalks or the bagasse could be used for cellulosic ethanol production and the grain may be used for ethanol production from the starch (Rajvanshi & Nimbker, 2003).Sweet sorghum is a potential biofuel crop as it is capable of producing high yields of ethanol from a combination of easily fermentable sugar and lignocellulosic bagasse.This is essential to meet the renewable fuel standard (RFS) which calls for production of 36 billion gallons or 144 billion litres of renewable fuel by 2020 (Stevens, 2014).
The biofuel produced from agricultural biomass provides a sustainable and eco-friendly energy option that fosters environmental sustainability as compared to other renewable sources.This led to economic consideration in production of sweet sorghum with emphasis on high grain yield, biomass as well as sugar yield.Plant breeding procedures require conduct of yield trials of crop genotypes in a number of environments.Such trials provide useful information on cultivar performance, adaptation and genotype by environment interaction, which are necessary for cultivar selection.Since yield and yield attributes are controlled by complex polygenes, their expression strongly depends on environmental conditions.Multi-environment trials (MET) are conducted to evaluate yield stability and performance of genetic materials under varying environmental conditions (Yan & Rajcan, 2002).A genotype grown in different environments will frequently show significant variation in yield performance.These changes are influenced by the genotype by environment interaction (GEI).Genotypes by environment interaction (GEI) sometimes complicate selection of superior genotypes (Ramagosa et al., 2013) making ranking of genotypes or correlation between genotype and phenotype difficult.Yield stability analysis therefore is an important step in developing cultivars for a wide range of environments or for specific location.AMMI analysis is used to determine stability of genotypes across locations using the principal component axis (PCA) scores and AMMI stability value (ASV).The purpose of the study therefore was to determine: (a) Interaction between genotypes and environments for grain yield, fresh biomass and degrees Brix.
(b) The yield stability of sweet sorghum genotypes across the contrasting environments (tropical and temperate environment) (c) To identify the best sweet sorghum genotypes for biofuel production under tropical and temperate environments.

Plant Material and Field Trials
The experimental material consisted of 26 sweet sorghum accessions obtained from ICRISAT (International Crops Research Institute for the Semi-Arid Tropics), India and 17 Dutch accessions obtained from the Netherlands which were grown during 2014 and 2015 cropping seasons.The field trial was conducted at Taastrup Campus (55°40′ N and 12°18′ E) of University of Copenhagen, Denmark (DNK) and National Cereals Research Institute, Badeggi (09°.04′N and 06°.08′E) in Nigeria (NGA).The Badeggi location had an average annual rainfall of 1,104 mm with Ferrosol type of soil while Taastrup had average rainfall of 313.9 mm between June and December in both cropping seasons.The experiment was laid out in randomized complete block design with three replications.A single row plot of 10 m × 0.75 m was maintained with inter-row and intra-row spacing of 75 cm and 30 cm, respectively.Two plants per stand were maintained after thinning with maximum of 33 plants per plot.In Nigeria, application of mixed inorganic fertilizer (NPK 15:15:15) was applied at the rate of 80 kg/ha to ensure top dressing at sowing and with urea fertilizer incorporated as side dressing.In Denmark, the fertilizer composition was 27% nitrogen and 4% sulphur.It was applied at four weeks after germination and at the initiation of shoot arrowing.Other cultural practices included weeding at four weeks interval until harvesting commenced.
Data on plant height was collected by measuring 8 plants per plot using meter ruler to measure from the base to the top of the shoot leaves.The degrees Brix (°Bx) was determined from 8 plants randomly cut per plot and the sugar concentration (Brix value) was measured at the third and seventh internode (from the top) using refractometer.Fresh biomass weight on the other hand was determined from all the plants per plot.Plants were harvested (stalk + leaves) without panicle and weighed using a weighing balance.Grain yield (GY) was determined after the harvested panicles with grains were dried and the seeds were threshed from panicle and weighed per plot using a weighing balance.A combined analysis of variance (ANOVA) was carried out using Genstat 8.1 version and the significant GEI was further substantiated by using different statistical tools.The means of all parameters were recorded.The Duncan Multiple Range Test was used for the comparison of means at 1% and 5% probability level.AMMI Model analysis was used to analyse genotype by environment interaction.The two locations and the two years form a combination of four environments.The environments are E1 (Denmark, 2014), E2 (Denmark, 2015), E3 (Nigeria, 2014) andE4 (Nigeria, 2015).
where, Yij = the measure mean of the i th genotype in j th environment; µ = the grand mean; α i = the main effect of j th genotype; β = the main effect of the j th environment; j = the interaction between i th genotype and j th environment.
The nature of genotype by environment interaction was investigated with the use of AMMI model (IRRI, 2007) which combines the standard analysis of variance with principal component analysis (Zobel et al., 1988).
Genotype stability ranking in these studies was calculated using the formula by Purchase (1997) as follows: where, ASV = AMMI stability value; SSIPCA = Interactive Principal Component Analysis sum of Squares 1 and 2; IPCA 1 and 2 Score = Interactive Principal Component Analysis 1 and 2 scores.

Results
The performance of each genotype considering the biofuel related traits such as plant height, fresh weight biomass, brix level and grain yield across two locations for two years is presented in Tables 1a and b.The mean comparison of the sweet sorghum genotypes across the two locations revealed significant differences in performance among the genotypes.The significant differences at p < 0.01 in each of the traits among the genotypes indicated variations in genetic constitution of the accessions.The plant height differed among the tested sweet sorghum genotypes across locations.Among the Dutch accessions the plant height ranges from 124 cm -167 cm in Denmark and 41 cm to 161 cm in Nigeria.The plant height of the ICRISAT accessions ranges from 127cm to 196cm in Denmark and 124 cm to 237cm in Nigeria.The genotypic effect and the contrasting effect of temperate and tropical climatic conditions influenced the plant height.In Denmark, the highest plant height was recorded on HI/Sn-PDI-R47 (167 cm) and F5.335 m10-1/6-6 (196 cm) of both Dutch and ICRISAT accession, respectively.The highest plant heights from Nigeria environment were obtained from H3-R9-32/3n (182.5 cm) and F5-3.SSN 10-2012-1 (236.8 cm) of both Dutch and ICRISAT accessions respectively.This indicated that the environments influence the height of the sorghum genotype.
The fresh biomass weight also differed significantly among the sweet sorghum genotypes across the two locations.The biomass weight tends to be higher among all the genotypes except H8-PD3-R51 which had the least performance across the locations.Fresh biomass weight is an important yield attribute to both juice yield, sugar content and ethanol production.According to Panhwar et al. (2003) it has a positive correlation with yield and other related traits.
The degrees Brix also differed significantly among the sweet sorghum genotypes across the two contrasting environments.High Brix values were recorded among the genotypes cultivated in Denmark than in Nigeria.This is an indication of genotype discrimination among the environment.Most of the high Brix values were recorded for the ICRISAT accessions (Table 1a and b).The results indicate that the genotypes with high fresh biomass weight are associated with high Brix values.Six genotypes: F5.3SSM10-15/5-1, F5.3SSM10-18/2-1, F5.3SSM10-20/2-1, F5.3SSM10-2/6-1, F7.5SSM09-1-1/7-1 and F7.5SSM09-1-1/9-2 recorded high fresh biomass weight with corresponding high degrees Brix.Combined analyses of variance of 43 sweet sorghum genotypes evaluated together in two years over the two contrasting locations (Denmark and Nigeria) are presented in Tables 2a and 2b.The analysis revealed that year (Y), environment (E), genotype (G), genotype and environment interaction (GEI) were significant at p < 0.01 and p < 0.05 in most of the biofuel yield attributes.The year and genotype interaction (Y×G) was not significant in biofuel attributes of Dutch accessions (Table 2b) but significant in grain weight and fresh biomass among the ICRISAT accessions (Table 2a).The genotype and GEI were significant in all the attributes except the brix level in ICRISAT accessions but significantly different in environmental effect.The implication was that there was genotype discrimination of the locations, which are similar to that in maize reported by Nzuve et al. (2013).
The significant interaction of genotypes with location (GEI) for plant height, fresh biomass weight, and grain yield except brix among ICRISAT accessions showed a differential behaviour of genotypes in the two locations.
The observed variation was purely environmental.Non-significant effect of the genotype by environment interaction (GEI) in Brix level among the ICRISAT accessions indicated that genotypes did not respond differently to the Denmark and Nigeria climatic conditions.That is the performances of the sweet sorghum varieties were not location specific.The higher sum of square for environment among the ICRISAT and Dutch accessions showed that environment were diverse, influencing the biofuel attributes except the brix value.Least sum of squares on brix level expressed non-environmental effect on this trait.Note.** and * signify significance at p < 0.001 and 0.005 respectively, df = degree of freedom and MS = mean square.

Stability Analysis
The additive main effect and multiplicative interaction (AMMI) models was used to evaluate the GE interactions and stability parameters of the sweet sorghum genotypes across the two locations in two years.The two locations and the two years formed a combination of four environments.
The AMMI analysis of variance for biofuel yields components tested at two contrasting environments for 2 years and the stability values are presented in Tables 3 and 4.
The observed variations in plant height, grain yield (panicle weight), fresh biomass and Brix value showed that the performance of the sweet sorghum genotypes were influenced by environment (E), genotype (G) and genotypes by environment interaction (GEI).Considering both ICRISAT and Dutch accessions, the AMMI analysis for plant height showed significant differences for treatment, genotype, environment as well as interaction (GEI).The genotype, environment and their interaction (GEI) accounted for 9.38, 25.20, and 20.81% of the treatment sum of squares respectively for ICRISAT and 17.75,22.43,32.28%respectively for Dutch accessions.The interaction sum of squares (interaction SS) was partitioned into IPCA1, IPCA2 and residuals.The IPCA1 and IPCA2 jointly accounted for 17.07% and 31.18% of the total variation due to interaction, with IPCA1 being significant in both accessions.The environment contributed largely to variation in plant height (Table 3 and 4).This also indicated that the performance of the plant height was location specific.
The analysis for grain weight showed that the genotype, environment, and GEI were also significant and each of the variation accounted for 14.54, 3.00, and 32.18% respectively for ICRISAT accessions and 24.85, 13, 01, and 26.26% treatment sum of squares for Dutch accessions.The environmental effect was the least in the observed variation in both accessions while the interaction was the highest.The high percentage of the interaction indicated high level of instability for both environment and genotypes.The variation due to interaction was also partitioned into IPCA1 andIPCA2 which were highly significant and jointly accounted for 24.32 and 25.07% of the variations due to interaction.
The fresh biomass weight for each genotype across locations was location specific and the variation was more of genotypic effect than the location for ICRISAT accessions.The AMMI analysis showed that G, E, and GEI effect accounted for 12.84, 7.32, and 26.88% for ICRISAT and15.63,23.80 and 27.14% treatment sum of squares for the Dutch accessions.The environmental effect was minimal on biomass weight considering ICRISAT accessions whereas it was enormous with the Dutch accessions.The variation due to G and GEI accounted for higher percentage and this indicated high level of instability between genotype and the environment.The IPCA1 and IPCA2 were both significant and accounted for 23.88 and 26.18% of the variation.
The AMMI analysis of variance for Brix value showed that G, E and GEI were significant and accounted for 9.72, 9.61 and 37.32% for ICRISAT and 24.10,11.53,and 31.61% for Dutch accessions as the Brix value total sum of squares.The variations were more of the genotypic effect and the interaction.The effect of GEI was higher showing that the brix level from each genotype responded differently in different environment.This also shows instability among the genotypes in sugar content.From the analysis the IPCA1 and IPCA2 jointly captured 33.34% and 30.26% of sum of squares interaction respectively of the two accessions (Table 3, 4).

Discuss
The Dutch accessions perform better in Denmark than in Nigeria.This is contrary to the report of Fix and Seebaluck (2013) that higher solar radiation at initial growth and during ripening in sugarcane help to accumulate more sucrose.The ideal genotype according to Yan and Kang (2003) is a genotype with high mean performance and absolute stability.The genotype must have high yield across the location with minimum GEI (Ezatollah & Mahsa, 2014).This report is confirmed by the results obtained on both fresh biomass weight and Brix value in which the genotype variances are greater than the variance due to interaction.The best performing genotypes with relative stability on each of the biofuel attributes (Plant height, fresh biomass weight, grain yield, brix levels) were about twenty-two genotypes among both ICRISAT and Dutch accessions.The performances were mostly location specific except few cases of sheared environments.From the available data collected the performance of the sweet sorghum is attributed to both genetic and environmental effects.Therefore genotypes adapted to specific locations or general stability across different environments has to be selected.

Table 1a .
Mean values for biofuel yield related components of ICRISAT sweet sorghum accessions across two contrasting environments Note.Mean values in the same column carrying the same letters are not significantly different at (p < 0.05%).Note.Means with the same letter (s) in the same column are not significantly different (p < 0.05) following separation by Duncan multiple Range Test.

Table 2a .
Mean square values of combine analysis of variance for sweet sorghum obtained from India (ICRISAT accessions)

Table 2b .
Mean square values of combined analysis of variance for sweet sorghum obtained from Netherland (Dutch accessions)