Introduction

The demand for the production of bioethanol from plant-based biomass materials is increasing globally. The demand is due to the fluctuating price of fossil fuels on world market combined with concern to protect the environment against the increase in greenhouse gas emissions resulted from the utilization of fossil fuels which contributed to the global warming. These necessitate gradually shifting from non-renewable energy source to plant-based biomass alternative energy source particularly in countries like Ethiopia to reduce its dependence on imported energy supplies and reduce the burden of foreign exchange.

Plant biomass-based biofuels can be derived from different sources like plant lignocellulosic material (bioethanol) or seed oil (biodiesel) or fermentable sugars (bioethanol) or starch (bioethanol). A number of crop plants are currently used as a source for biofuel production using their biomass. Sorghum (Sorghum bicolor) is among the preferred crops to be used for such purposes. Sweet sorghum is a type of sorghum which is among important biofuel feedstock for ethanol production globally (Murray et al. 2009). Moreover, fodder sorghum is promoted as a biomass crop for lignocellulosic-based ethanol production. The stalk of sweet sorghum is rich in sucrose which is used for the production of bioethanol via fermentation process. The sweet sorghum is characterized by reduced grain yield as compared to grain sorghum. The crop is categorized as the promising source of biofuel for the production of first (sucrose) and second (lignocellulosic material) generation of bioethanol. Sweet sorghum is among C4 crops which is characterized by significant level of directly fermentable stalk sugars. It has the potential to give high biomass under unfavorable conditions unlike many other C4 plants. The crop accumulates similar level of sugar to that of sugarcane (Murray et al. 2009). Under conducive environment, sweet sorghum crop has a significant potential of producing ethanol which is close to 8000 L of ethanol per hectare of land (Murray et al. 2009). Unlike sugarcane, sorghum is highly suitable for large-scale mechanization and seed-based production. Moreover, it is a relatively drought tolerance crop and requires low amount of water as compared to other types biofuel crops (Gnansounou et al. 2005). Besides, sweet sorghum has a huge adaptation potential ranging from extreme cold to high temperature environments. It is also highly suitable for both subsistence agriculture, semi-intensive and intensive agriculture which makes it an ideal crop to produce in various regions of the world. The utilization of sweet sorghum as bioenergy crop seems to be economically very visible as the cost of sugarcane production is three times more than the sweet sorghum cost of production (Reddy et al. 2005). Similarly, bioethanol production processing of stem sugar needs limited chemical reaction processes and less energy from the beginning to the final product than lignocellulosic-based technologies (Wang et al. 2009) hence presents a better alternative to lignocellulosic-based ethanol production.

Despite all these potentials, the global effort to improve the crop for bioenergy-related traits remains minimal. Little has been done in genetic diversity analysis of sweet sorghum genotypes using morphological and °Brix (Disasa et al. 2017) and molecular (Casa et al. 2005, 2008; Murray et al. 2009; Wang et al. 2009; Disasa et al. 2016) markers. Despite, Ethiopia/surrounding region is believed to be the center of diversity and origin for sorghum, not much attention has been given toward identification and mapping of QTLs conferring bioenergy traits using Ethiopian sweet sorghum germplasm. Hence, it is important to identify some novel genes/genomic regions that associated with various bioenergy-related traits.

Identification and mapping of economically important quantitative traits for the improvement of high stalk sugar-producing varieties is the major interest of many plant breeders. Several approaches can be used in identification and mapping of the trait of interest in crop plants. Most of the studies have largely been focused on linkage or family mapping of quantitative trait loci (QTL) which involves the generation of mapping populations, screening of markers for polymorphism, construction of linkage map followed by QTL identification and mapping. This is an ideal approach especially for the identification of QTLs linked with the traits of interest using population derived from biparental crosses.

Identification and mapping of QTLs for several traits have been previously reported in grain sorghum. Among these, QTLs conferring stay-green (Tuinstra et al. 1996, 1997,1998; Crasta et al. 1999; Xu et al. 2000; Tao et al. 2000; Subudhi et al. 2000; Kebede et al. 2001; Sanchez et al. 2002; Haussmann et al. 2002; Harris et al. 2007) and plant height traits (Lin et al. 1995; Pereira and Lee 1995; Rami et al. 1998; Hart et al. 2002; Klein et al. 2000; Natoli et al. 2002; Feltus et al. 2006; Ritter et al. 2008; Murray et al. 2008; Shiringani et al. 2010) are the most studied traits in sorghum. However, little attention has been given to identify and map QTLs conferring for bioenergy-related traits. Some of these works were based on selected traits derived from contrasting parents for stem sugar trait to identify and map stem sugar and other nonstructural carbohydrates (Murray et al. 2008). Limited attempts on identification and mapping of QTLs liked to stem sugar and other agronomic traits were also previously reported (Shiringani et al. 2010; Guan et al. 2011). Similarly, QTL mapping for eleven sugar-related agronomic traits were also conducted using 184 recombinant inbred lines (RILs) derived from two different parents (grain and sweet sorghum lines) through crossing (Ritter et al. 2008). All of the identified QTLs so far were from exotic germplasm, and there was no study reported regarding QTL mapping of °Brix and related-sugar content traits using Ethiopian germplasm. Therefore, the current study aimed at identification and mapping of QTLs associated with °Brix and other biomass-related traits in sorghum using Ethiopian sorghum germplasms.

Materials and Methods

Development of Mapping Populations

Effort has been made to generate F2 mapping populations by crossing one of the selected grain sorghum (Sorcoll 163/07) as a female parent and sweet sorghum (Gambella) as a pollen source. Sorcoll 163/07 was inbred line characterized by its drought tolerance nature with an average plant height (224.5 cm), stalk diameter (15.5 mm) and °Brix (6.8). Gembella is farmers’ preferred improved genotypes characterized by its sugary stalk with an average plant height (192.3 cm), stalk diameter (14.0 mm), and °Brix (14.4).

Planting was undertaken at Melkassa Agricultural Research Center (39°21′E, 8°24′N) during main rainy season in 2012. The field was supplied with irrigation as the area experiences frequent rain failure. All the necessary agronomic management practices were implemented. Crossing was carried out using hand pollination technique with a great care. The heads were tightly covered with appropriate paper bag in order to avoid contamination from neighboring pollen. Enough seeds from the crosses were harvested and used to produce F1 plants. Selfing of F1 populations was undertaken at Water Agricultural Research Center (40°21′E, 9°33′N) during offseason in the same year using irrigation facilities. The seeds from F1 populations were harvested and planted to produce F2 mapping population for sugar-related traits and identify molecular markers that are strongly associated with the traits of interest. A total of 200 F1 seeds were planted in green house of the National Agricultural Biotechnology Research Center based at Holetta (38°50′E, 9°05′N), Ethiopia in order generate F2 plants from which genotyping was undertaken. In order to test in multiple environments, each F2 plant was selfed to produce F2 derived F3 plants (F2:3).

Experimental Design and Phenotypic Evaluation

Sufficient seeds derived from each F2 population were planted across two locations; one at Melkassa Agricultural Research Center (39°21′E, 8°24′N) and the second one is at Dhera experimental station (39°19′E, 8°15′N) in Ethiopia. An α-lattice design was used with three replications for all the progenies along with the two parental lines in each trial. In both locations, a one-row plot of 5 m row length with the inter-row space of 0.75 m was used. Manual planting was undertaken followed by thinning to 0.2 m spacing after 3 weeks of emergence. Recommended amount of fertilizer was applied following the procedure used by Ayana (2001). Weed control and disease management was undertaken. The field was protected from bird damage.

Data were collected on three traits such as plant height, stalk diameter and °Brix. Plant height was measured as the length of the plant at physiological maturity stage ranging from the ground to the tip of the panicle. Stalk diameter was measured using appropriate digital caliper (World Precision Instruments, Shanghai Trading Co., Ltd). The measurement was taken at three locations along the height of the plant starting from 20 cm above ground. Stem juice was extracted from stalks harvested just before reaching full maturity stage. One to two drops of the extracted juice was applied on top of the hand-held refractometer (TEC++ DR. VOLKER SCHMIDT GM BH, Deutschland) in order determine the percentage of soluble solids (°Brix).

Phenotypic Data Analysis

Minitab v.16 statistical software (Minitab Inc. 2010) was used to estimate the normality of the data for each trait based on the data collected from two locations. Similarly, other major statistics including standard deviation and percent coefficient of variation (CV) of the three traits were analyzed using SAS software v. 9.2 (SAS inst. 2008). METAR (Multi Environment Trial Analysis for R) Version 2.1 software was used to calculate the broad-sense heritability, while the PROC CORR procedure was used to compute the Pearson correlation coefficient.

DNA Extraction and Polymorphism Screening

Young leaf tissues were collected from each parent, Sorcoll 163/07 and Gambella varieties in order to extract genomic DNA that will be used to screen polymorphic markers using the two selected parents. Promega Wizard® Genomic DNA Purification Kit (Madison, USA) was used to isolate the genomic DNA. Kit based DNA isolation was preferred in order to obtain good quality and quantity DNA molecules. The purified genomic DNA was stored in +4°C followed by quality checking using 0.8% agrose gel stained with GelRed® (Biotium, USA). The concentration of the genomic DNA was quantified using spectrophotometer. Based on the recorded concentration for each genomic DNA samples, the working DNA solution was further diluted to the concentration of 10 ng/µl (Disasa et al. 2016).

A number of literatures from previously published sorghum SSR markers mapping were used in order select the most polymorphic markers (Bhattramakki et al. 2000; Menz et al. 2002; Wu and Huang 2006; Li et al. 2009; Ramu et al. 2009). Nearly 300 selected sorghum SSR primers that are found to have full coverage of the ten linkage groups were identified. A total of 10 µl PCR volume was used with a combination of 30 ng/µl template DNA, 0.04 pmol forward primer, 0.2 pmol reverse primer, 1 × taq buffer, 2 mM MgCl2, 0.16 mM each of the four deoxynucleotide triphosphates (dNTPs),0.16 pmol for each four different fluorescent labels and 0.2 u Amplitaq (SibEnzyme Ltd, Russia). The entire forward primers were supplemented with fluorescently labeled detectors such as FAM, PET, NED or VIC (Applied Biosystems, USA) on the 5′ end (Schuelke 2000). These will enable to detect the amplified products independently.

Polymerase chain reaction was performed using a GeneAmp® PCR System 9700 thermal cycler (Applied Biosystems) adjusted for initial denaturation at 94 °C for 15 min, the second denaturation was programmed at 94 °C for 30 s, annealing temperature set at 50 °C for 1 min while extension of the product was adjusted at 72 °C for 2 min and final elongation at 72 °C for 20 min (Disasa et al. 2016).

To confirm whether the PCR product was successfully amplified or not, a 2.0 µl volume of the PCR products was obtained and checked using a 2% (w/v) agarose gel stained with GelRed® (Biotium). The result was visualized under UV. Amplified products were selected followed by co-loading of four PCR products together. The volume of each PCR products is depending on the strength and nature of fluorescent label used. The products were supplemented with Hi-Di™ Formamide (Applied Biosystems, USA) and internal size standard, GeneScan™ 500 LIZ® (Applied Biosystems, USA). Fragments with different sizes were separated using capillary electrophoresis techniques using an ABI Prism® 3730 Genetic analyzer (Applied Biosystems, USA), and fragment sizes scoring was done using GeneMapper 4.0 software (Applied Biosystems, USA).

Genotyping F2 Mapping Populations

A total of 192 F2 progenies and their parents were genotyped in these experiments. Fresh leaf tissues were collected from young seedlings, and genomic DNA was isolated using Promega Wizard® Genomic DNA Purification Kit (Madison, USA). The same PCR protocol and thermal cycler used to screen polymorphism among the parents was also applied here. A total of 81 SSR primers that show polymorphism between the parents, Socrcoll 163/07 and Gambella, were used to genotype the entire mapping populations. SSRs were amplified and separated based on fluorescent labeled forward primers. The entire polymorphic markers were analyzed in all mapping populations including parents using the ABI PRISM_3730 genetic analyzer (Applied Biosystems, USA). GeneMapper software, v4.0 was used to collect the data. The scored markers and F2:3 mapping populations were used to analyze QTL associated °Brix traits. Fragment scores were recorded as 2, 0 and 1 for Sorcoll 163/07, Gambella and heterozygous, respectively. Missing values were recorded as − 1 for individual marker across the entire populations.

Linkage Map Construction

Based on 76 successfully scored markers across 192 progenies and parents, the linkage map was constructed using IciMapping software v 4.0 (Li et al. 2007). Based on the logarithm of odds (LOD) threshold of 3.0, linkage groups were identified and map distances (in Centi-Morgan units) was computed based on the Kosambi mapping function (Kosambi 1944). The relationship between genetic distance in (m) and recombination frequency (r) is given by

$$m = 25 \ln \left( {\frac{1 + 2r}{1 - 2r}} \right)\quad {\text{Or}}\quad r = \frac{1}{2}\frac{{e^{4m} - 1}}{{e^{4m} + 1}}$$

The linkage map was compared and assigned to chromosomes according to sorghum censuses map (Mace et al. 2009).

The whole polymorphism screening of parents and F2 genotyping activities were undertaken at International Crop Research Institute for Semi Arid Tropics (ICRISAT-Nairobi), regional program, whereas DNA extraction was carried out at the National Agricultural Biotechnology Research Center of the Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.

QTL Analysis

Identification and mapping of the QTLs were based on three different approaches such as single-marker analysis (SMA), interval mapping (IM) and inclusive composite interval mapping (ICIM) using IciMapping software v 4.0 (Li et al. 2007). The results from IM and ICIM were compared. Finally, ICIM was used for the analysis. LS means of each of the traits were used for QTL mapping. Missing phenotypic data was scored as − 100.00.

Results and Discussion

Phenotypic Evaluation

For traits such as plant height and stem diameter, there was no significant difference between the two parents as well as the progenies in both environments (Table 1). However, °Brix value measured for Gambella was significantly (P < 0.01) higher than that of Sorcoll 163/07. The phenotypic value of °Brix showed continuous variation across the progenies in two locations and normally distributed with a moderate value of skewness and kurtosis (Table 1). Environmental means for °Brix were 8.9 and 9.1 in the two locations. The highest (15.8) and lowest (4.5) °Brix values were recorded among the progenies at Dhera and Melkassa stations, respectively.

Table 1 Phenotypic values and statistical analysis of °Brix, stalk diameter and plant height values in different environments
Full size table

The heritability of the traits such as °Brix, stalk diameter and plant height were 0.88, 0.41 and 0.65, respectively. °Brix showed a high heritability, whereas the stalk diameter and plant height showed moderate to high heritability. Pearson’s correlation coefficients analysis showed that there was no significant correlation between °Brix and plant height (r = 0.089, P > 0.05), but a slight negative correlation was observed between °Brix and stem diameter (r = − 0.320, P < 0.05). The result also showed that there was no significant correlation between plant height and stem diameter traits (r = 0.145, P > 0.05).

°Brix combined with other sugar-related traits are the most important traits for the utilization of sorghum crops as a bioethanol feedstock. Biomass-related traits such as plant height and stalk diameter are also important agronomic traits for the production of ethanol from sweet sorghum crop as they determine the overall sugar yield. Any positive correlation between these traits helps to improve the stalk juice yield of the crop. According to Murray et al. (2008), the total stem sugar yield per hectare is dependent on two traits, sugar concentration in the stem and stem juice yield per hectare. Increasing sugar concentration would be an important step to increase energy density and reduce processing and transportation costs. Identification and introgression of QTLs that are associated with both high sugar concentration and stem juice yield into elite varieties will be an important step. Therefore, QTLs that influence °Brix content would be good breeding target for biofuel improvement program in the future since °Brix had strong positive correlation with the sugar concentration and stem juice yield traits (Ritter et al. 2008).

Our result showed that no significant correlation between °Brix and plant height, which deviates from previously reported study whereby most of sugar-related traits showed significantly positive correlation with plant height (Ritter et al. 2008; Murray et al. 2009). The deviation may be due to non-contrasting parent (no significant variation between the two parents) for plant height trait and stem diameter used to develop mapping populations. The high heritability (H = 0.88) estimate for °Brix trait was in agreement with previous study conducted by Ritter et al. (2008) using the same trait (H = 0.86). However, the result is higher than the study reported by Murray et al. (2008), Shiringani et al. (2010) and Guan et al. (2011) which are reported to be H = 0.65, H = 0.59 and H = 0.76, respectively. This high heritability estimate nature of the trait has an implication that the trait is suitable for QTL mapping studies as compared to traits with low heritability potential.

Linkage Map Construction

Data from fragment analysis showed that among 304 polymorphic SSR markers tested, 81 (26.6%) found to be polymorphic between the two mapping parents. In contrast, 97 (31.9%) were unable to discriminate the two parents and considered as monomorphic markers and the remaining 68 (22.4%) markers did not produce distinct and reproducible patterns of polymorphism among the two parents. The polymorphic markers were fairly distributed across the ten linkage groups.

Out of the 81 polymorphic marker loci used for genotyping F2 mapping populations, five of them (6%) were failed to discriminate the F2 progenies and excluded from the analysis. Most of the 76 SSR markers used here showed good segregation pattern, which did not significantly deviate from the expected 1:2:1 segregation ratio. The distributions of markers more or less appear to be evenly distributed across the ten linkage groups except few numbers of markers on some of the linkage groups due to the difference in length of chromosomes (Fig. 1). Out of the 76 polymorphic marker loci, six of them could not assign to any specific linkage group and excluded from the final analysis.

Fig. 1
figure 1

Constructed genetic linkage map using 192 F2 mapping populations genotyped using polymorphic SSR markers

Full size image

QTL Detection and Analysis

Two major QTLs influencing the °Brix were detected on linkage group SBI-05 and SBI-06 at a LOD threshold of 3.5 using the data collected from Melkassa Agricultural Research Center. In contrary, QTLs influencing plant height and stalk diameter were not detected at similar LOD cut value (Fig. 2). The two QTLs associated with °Brix were found to be major QTLs and each contributed 40.0–44.7% of the total phenotypic variation (Table 2). The first QTL on the SBI-05 linkage group lies between Xiabt26 and Xisep0805 with a LOD value of 3.5, whereas the second QTL on the linkage group SBI-06, located between Xtxp289 and Xisep0511 with a LOD value of 5.0. The additive effect of each QTL on linkage group SBI-05 and SBI-06 was − 1.59 and − 1.93%, respectively. Similarly, the dominance effect was − 2.97 and − 2.67%.

Fig. 2
figure 2

Detected quantitative trait loci for °Brix trait in sorghum at LOD threshold 3.5 (at MARC environment)

Full size image
Table 2 Linkage groups, position and flanking markers of detected quantitative trait loci associated with °Brix trait evaluated at MARC and Dhera environments
Full size table

Seven significant QTLs that are associated with the °Brix content (LOD threshold > 3.5) were detected from the experiment evaluated at Dhera station (Fig. 3). Similar to Melkassa environment, no QTL was detected that is linked with plant height and stalk diameter traits. The QTLs were distributed across five linkage groups such as SBI-02, SBI-05, SBI-06, SBI-07, and SBI-10 (Table 2). The LOD values of the QTLs were ranging 3.5–5.5, while each QTL was explaining a phenotypic variation ranging from 17.2 to 44. 3%. The additive effect of each QTL ranged from − 2.5 to 2.3%, where as dominant effect ranged from − 3.0 to − 1.7% of the °Brix content.

Fig. 3
figure 3

Detected quantitative trait loci for °Brix trait in sorghum at LOD threshold 3.5 (at Dhera environment)

Full size image

A total of seven QTLs were detected using the combined data collected from two locations using LOD threshold > 3.5, whereas no QTLs influencing plant height and stalk diameter were identified (Fig. 4). Two of the QTLs conferring for total soluble solids were consistent across the two environments and located on the same linkage groups (SBI-05, SBI-06). The remaining QTLs were located on linkage group SBI-02, SBI-07 and SBI-09 with a LOD value ranging from of 3.5 to 5.4 (Table 3). Three QTLs were detected and mapped on linkage group SBI-07 with pairs of flanking markers Xtxp141–Xtxp270, Xisep0639–Xisep0639 and Xisep0639–Xisep0622 (Fig. 5). The three QTLs are considered as independent because the distance between them is greater than 20 Centi-Morgan (Shiringani et al. 2010). The QTL on the SBI-09 linkage group was detected between markers Xisep0449 and Xtxp265. Almost all the identified QTLs are considered as major QTLs contributed high percentage of phenotypic variance which ranged from 38.5 to 49.8% of the phenotypic variation with the additive effect ranging from − 2.3 to 2.1%. Besides, the dominance effect ranged from − 3.0 to − 2.3%.

Fig. 4
figure 4

Detected quantitative trait loci for °Brix trait in sorghum at LOD threshold 3.5 (combined locations)

Full size image
Table 3 Linkage groups, position and flanking markers of quantitative trait loci associated with° Brix trait evaluated at MARC and Dhera environments (combined locations)
Full size table
Fig. 5
figure 5

Map of QTLs controlling °Brix using F2 genotypic data and combined phenotypic data of F2:3 mapping populations using a LOD cutoff 3.5

Full size image

All of the detected QTLs were found to be major QTLs explaining phenotypic variation ranging from 38.5 to 49.8% and greater than most of the studies conducted in the past using grain and sweet sorghum genotypes (Ritter et al. 2008; Shiringani et al. 2010; Guan et al. 2011). However, direct comparison is difficult due to the difference in type of mapping populations and number of markers used in the study.

Two similar QTL located on SBI-05 and SBI-06 that are influencing °Brix traits were detected in both MARC and Dhera environments. Similar studies were reported by Ritter et al. (2008) who detected similar QTLs across two environments for many agronomic traits. This is a good indication of the high heritability nature of the trait. The repeatability of these QTLs has also an implication that they are useful for the introgression of these QTLs into elite breeding lines due to their stability in different environments. Similar result was reported by Ritter et al. (2008) on two linkage groups such as SBI-05 and SBI-06 with major QTLs. Our findings are also partly inconsistent with the previously reported result by Guan et al. (2011) who detected four QTLs out of which two of them located on linkage group SBI-02 and SBI-07, whereas the remaining two QTLs were detected on linkage group SBI-02 and SBI-07 across two environments.

Our findings also showed the repeatability of those previously reported QTLs at different times on various linkage groups. These major QTLs were located on SBI-02 (Shiringani et al. 2010; Guan et al. 2011), SBI-05 (Ritter et al. 2008), SBI-06 (Murray et al. 2008; Ritter et al. 2008), SBI-07 (Murray et al. 2008; Guan et al. 2011). This is a good indication that these QTLs are reproducible and important to be utilized in breeding program. Besides, we have identified a new major QTL on linkage group SBI-09. This QTL has never been reported previously and may be unique to Ethiopian sorghum germplasm. In contrary, our study failed to detect previously reported significant QTLs distributed on various linkage groups such as SBI-01 (Guan et al. 2011), SBI-03 (Murray et al. 2008; Guan et al. 2011), SBI-04 (Bian et al. 2006; Shiringani et al. 2010) and SBI-10 (Bian et al. 2006).

The major goal of many sweet sorghum breeders is to develop cultivars with improved stem sugars both in terms of sugar concentration and stem juice yield. QTLs that control these important traits are usually controlled by multiple genes and require dissection of the chromosomal region(s) using selected molecular markers. Successful identification and mapping of these novel QTLs plays a significant role to introgress the QTLs into locally adapted sweet sorghum genotypes. Major QTLs with high repeatability across locations were identified and mapped in this study. The high °Brix heritability observed in this experiment has implication that successful utilization of these novel QTLs in breeding program in the future to improve the sweet sorghum genotypes. Therefore, the current study will serve as a foundation to start advanced sweet sorghum breeding program in Ethiopia through marker-assisted breeding approaches to improve the quality of the sorghum lines for bioethanol industry.