Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Improvement of the nutritional quality of sorghum and Pearl Millet

R. Jambunathan
International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Hyderabad, India


The discovery that the Opaque-2 gene in maize improves protein quality has stimulated great interest among breeders, nutritionists, and biochemists, and considerable progress has been made toward genetic improvement of plant protein quality in other cereals. Sorghum and pearl millet are two of the most important crops grown in the semi arid tropics (SAT). If sorghum and millets are to retain their place, and to increase, as major cereals for human food in the SAT where they are more productive and reliable than are other cereals, their grain quality is of paramount importance. The International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), created in July 1972 with headquarters in Hyderabad, India, has four main objectives. One of these is: "To serve as a world center to improve the genetic potential for grain yield and nutritional quality of sorghum, pearl millet, pigeon pea, chick pea and groundnut."

A stable, high yield of the crop is one of the main objectives of ICRISAT, At the same time, we will exploit whatever latitude exists for other nutritional characteristics such as protein, lysine, starch types, oil, minerals, etc.

Grain quality can perhaps be considered to consist of two main parts: (i) evident quality based on appearance, flavour, and cooking quality characteristics, and (ii) cryptic quality based on nutritional value. This paper will discuss progress that has been made in the improvement of the nutritional quality of sorghum and pearl millet.


Sorghum is largely a self-pollinated crop and is grown on more than 40 million hectares in tropical and temperate zones. About 83 per cent of the areas are in the semi arid tropics.1 Yields in the more developed countries are five times those in the tropics, indicating that much improvement is possible in the SAT

High-lysine gene

As in other cereals, lysine is the first limiting amino acid in sorghum. After screening more than 9,000 accessions in the world germ plasm collection, Singh and Axtell2 reported in 1973 that two sorghum lines of Ethiopian origin, IS 11167 and IS 11758, had exceptionally high lysine at relatively high levels of protein. Both lines were also high in oil percentage (Table 1). The protein efficiency ratio (PER) values obtained with IS 11167 and IS 11758 were 1.78 and 2.06, respectively, as compared with the PER of 0.86 obtained for normal sorhgum. Inheritance studies suggested that the increased amount of lysine in each line was controlled by a single recessive gene that could be easily transferred by standard plant breeding procedures.

Protein distribution in high-lysine sorghum

The improved nutritional quality in Opaque 2 maize is due to the decreased amount of prolamine (Zein) and the increased concentration of albumins, globulins, and glutelins, resulting in a larger amount of lysine in the whole kernel.

TABLE 1. Chemical Composition and Seed Characteristics of Whole Grain Sample of High-Lysine and Normal Sorghum Lines

Character High-lysine
Protein composition
protein % 15.70 17.20 12.70
lysine, g/100 g protein 3.33 3.13 2.05
lysine, %of sample 0.52 0.54 0.26
Chemical composition      
oil % 5.81 6.61 3.32
Seed characteristics
% germ 14.60 16.30 10.10
seed weight, g/100 seeds 2.78 2.45 2.75
Carbohydrate composition      
sucrose, % of sample 3.08 2.61 1.03
starch, % of sample 58.90 57.80 60.80

Source: Reference 2

Jambunathan et al.3 fractionated the protein of the two high-lysine (hi) sorghums and compared the distribution of proteins with that of F2 kernels obtained from a cross between Redlan (a normal variety) and high-lysine lines. The distribution of proteins in these F2 kernels and high-lysine sorghums, as shown in Table 2, indicates that high-lysine sorghums have a lower percentage of alcohol-soluble fractions and higher percentages of saline-soluble fractions, (albumins and globulins) and the distribution pattern is similar to that of Opaque-2 maize. It is clear that the high-lysine F2 sorghum kernels have protein distribution patterns similar to those of high-lysine sorghums.

As one of the ICRISAT's objectives is to improve the nutritional quality of sorghum, and the two high-lysine lines were available at the time when ICRISAT started to function, improvement of the nutritional quality of sorghum was included in the sorghum breeding programme in 1973. Because the kernels of the high-lysine sorghums are floury in nature, are partially dented, and have low seed weight resulting in low yield potential, attempts were made to transfer the shrunken high-lysine (hi) grain of the Ethiopian cultivars to photoperiod-insensitive genotypes with plump, well-filled grains. Another chemically induced high-lysine mutant, P721, discovered at Purdue University,4 was also used in the crossing programme at ICRISAT. The amino acid composition and fractionation data on P721 have been recently reported by Guiragossian et al.5

One of the functions of the Biochemistry and Common Laboratory Services Unit at ICRISAT is to assist breeders in selecting the desired cultivars and progenies with improved lysine concentration. Because we selected for vitreous plump grain, the screening method using a light-box for Opaque character could not be employed Therefore, in our laboratory, we have evaluated several methods for the rapid and accurate estimation of protein and lysine. Progress that has been made in identifying suitable methodology to screen thousands of samples is described below.


Protein estimation

We needed simple, rapid, inexpensive, and reasonably accurate methods. Several methods are available for the estimation of proteins; some of the commonly used procedures include the micro- or macro-Kjeldahl method, which is still used as a standard for crude protein estimation, the biuret method, the Lowry method, estimation of ammonia using the Technicon autoanalyser, and the near-infrared reflectance method.

When the laboratory started to function in 1974, we tried the biuret procedure6 for the estimation of protein in sorghum and obtained a correlation coefficient (r) of 0.91 with the micro-Kjeldahl method. However, when we used this method for routine screening, the differences between biuret protein value and micro-Kjeldahl value were especially large in the low protein range, as shown in Table 3. Later we tried the Technicon autoanalyser (TAA) method and a very high correlation coefficient of 0.99 was obtained between the micro-Kjeldahl and TAA protein values. During routine screening of samples using the TAA method, error percentage between TAA values and micro-Kjeldahl values determined on random samples showed less than 2 per cent variation over a wide range of protein content. Therefore, we are using the TAA method for the estimation of protein values in sorghum.

TABLE 2. Nitrogen Distribution in the Whole Kernels of Normal and High-Lysine Sorghums (Percentage of Total Nitrogen)

Fraction Redlan Redlan x
IS 11758
F2 kernels.
I Saline 10.0 15.3 26.0
II Isopropanol 15.7 26.4 10.3
III Isopropanol+2-mercapto ethanol 31.3 26.5 19.6
IV Borate buffer+ 2-mercapto ethanol 4.5 4.3 6.5
V Borate buffer+ 2-mercapto ethanol + sodium
dodecyl sulfate
29.3 22.5 27.2
Total nitrogen extracted (%) 90.8 95.0 89.6
Protein % 13.53 13.0 18.5
Lysine (g/100 g P) 1.56 1.85 3.27

Lysine estimation

For estimation of lysine content we used the dye-binding capacity (DBC) method.7 This method is based on the principle that the basic amino acids (lysine, histidine, and arginine) react in an acid medium with a monosulfonic acid azo dye (acid orange 12) to form an insoluble complex, resulting in a decreased intensity of the solution. The proportion of the dye bound is directly related to the total basic amino acids in the sample, and the unbound dye can be conveniently measured colorimetrically as per cent of transmission, expressed as Udy instrument reading (UIR). The UIR depends on the total quantity of protein in the sample and also on the basic amino acid concentrations in the protein. After protein determination by the TAA method, the weight of each of the samples was adjusted to contain 80 mg of protein, and the UIR value was obtained on the sample. The UIR value represents the total amount of basic amino acids in the sample. In order to speed up the analysis, and also to reduce the possible influence of sample size on UIR values, the procedure was slightly modified, as follows:

TABLE 3. Deviation of Biuret (B) Protein from MicroKjeldhal (MKJ) Protein Values in Sorghum Samples

No. of
Deviation of B
from MKJ (%)
5 - 5.9 5 12 - 22.7
6 - 6.9 6 44 - 17 9
7-7.9 7 69 -18.7
8 - 8.9 8 97 - 16.6
9 - 9.9 9 31 - 16.9
10 - 10.9 10 16 - 11.9
11 - 11.9 11 10 - 11.2
    279 - 16.6

UIR values were taken on a one-gram weight of sample (instead of adjusting the sample weight to contain 80 mg of protein) and the readings (UIR) were divided by the percentage of protein (P) in the grain sample to obtain a ratio (UIR/P). This ratio was compared with the lysine value determined by ion exchange chromatography using an amino acid analyser. Using 58 sorghum samples, which had a range of 1.34 to 2.98 per cent lysine and a UIR/P range of 2.32 to 4.57, we obtained a correlation coefficient of 0.93 between actual lysine concentration and UIR/P ratio. A regression equation was obtained using this correlation, and lysine values were predicted on routine samples by this method. We do check these estimated values by analysing selected samples with the amino acid analyser. We can now analyse, using the above procedures, about 140 samples for protein and lysine per day.

Based on laboratory values, ICRISAT breeders selected the best grain samples among progenies of random mating populations and crosses involving the Ethiopian hl gene. When the improvement of protein content and quality in the selected lines was followed year after year, it was observed that some of the selections showed a large fluctuation in protein levels (Table 4). Also, we could not increase the frequency of occurrence of such plants very much, and this led us to question whether the high-lysine (hi) gene was stable in a normal (plump seed) endosperm background. It was clear that we needed more data before this programme could be confined.

TABLE 4. Variation in Protein Content of Sorghum Lines from Season to Season


1974 Rabi

1975 Kharif

1975 Rabi

UDY N Protein
UDY N Protein
79337 2 7.2 30.0 1 7.8 34 1 12.0 24.5 20
79339-3 7.1 30.5 1 7.9 28 1 13.5 24.5 6
79337-4 7.2 31.5 1 7.2 28 1 12.7 23.0 8
79751-4 7.6 30.0 1 7.8 30 1 15.1 25.5 6
79337-2 7.2 30 0 1 7.0 32 1 13.6 23.5 6
79339-3 7.1 30.5 1 7.5 31.5 1 12.0 23.0 4

UDY-UIR per 80 mg protein. N-number of entries used to calculate the mean. Kharif-monsoon season, from June to October. Rabi-winter season, from October to March.

Therefore, a systematic study is now being conducted by a research scholar on the variation of protein and lysine due to location, management, and environmental conditions, and on the stability of the high-lysine gene under these varying conditions. Preliminary results indicate that the crosses involving P721 are promising in that they give rise to a much higher frequency of high-lysine segregants. This study is still in progress.

Polyphenolic compounds, also known as tannins, present in the grain of some sorghum cultivars substantially reduce the big-availability of protein and other nutrients, which indirectly has a major negative effect on the nutritional quality of grain sorghum. At the same time, many researchers have presented data supporting "bird-resistant" qualities associated with the brown (high tannin) sorghum in areas where bird damage is severed Weathering, deterioration of seed quality due to weather conditions, including pre-harvest seed germination, is reportedly less serious in high-tannin sorghums. More information needs to be obtained to understand the role of tannin in bird- and weather-resistance of sorghum grains.

A major source for improving the nutritional quality of sorghum is the germ plasm collection. The sorghum collection at ICRISAT exceeds 14,000 accessions. We have recently completed proximate and mineral analyses on 100 germ plasm collections representing the following types: with lustre; with persistent sub-coat; completely corneous; almost corneous; intermediate; almost floury; completely floury; waxy endosperm, and with white, yellow, straw, light brown, brown, reddish brown, light red, red, grey, and purple seed coat colours (Tables 5 and 6). Them were grown on red soil at the ICRISAT farm in the 1976 Rabi season. The wide range obtained in minerals and trace elements indicates that it is possible for sorghum lines to contain various amounts of these elements. It is recognized that the mineral composition of grain is influenced by the environment, soil, and management conditions. However, this observation draws our attention to the importance of analysing the advanced elite lines in the breeding programme for all the possible chemical constituents so that any cultivar having a very low amount of any of the important constituents can be identified at an early stage.

TABLE 5. Proximate Analysis of 100 Selected Sorghum Germ Plasm Samples

  Range Mean
Starch, % 55.6 - 75.2 70.8
Protein, % 10.6 - 18.5 14.1
Ether extract, % 2.1 - 7.1 3.3
Crude fibre, % 1.0 - 3.4 1.9
Ash, % 1.6 - 3.3 2.1
Sugar, % 0.8 - 4.2 1.3
Tannin, % 0.1 - 6.4 0.6
Lysine, g/100 g P* 1.37 - 3.39 1.7
100 seed wt. (g) 1.3 - 5.7 2.8
Grain hardness (kg) 1.8 - 10.2 6.5

* 1977 data.

Recently there have been several reports on protein-calorie malnutrition, and criticisms have been levelled against the protein quality work in food grain samples. 9,10 Studies conducted by the National Institute of Nutrition in Hyderabad, India, have shown that the primary deficiencies in the diet of people in India are mainly calories, vitamins, and minerals.! More research is required to determine the extent to which vitamins and minerals are heritable before screening and selecting for these constituents.

One of our current interests in sorghum is to determine the factors that affect/relate to consumer acceptance of products prepared from sorghum. Our target populations are sorghum and pearl millet consumers living in SAT regions. It is not uncommon to find farmers in indict growing a local cultivar in a small area for their own family use and another high yielding cultivar or hybrid in a larger area for selling on the market. Therefore, evident quality characteristics deserve an important consideration in a breeding programme, and efforts need to be made to screen for any characteristics that might be associated with the preparation of food products. There are certain characteristics that are preferred by people and are associated with good food products. Some of the desired characteristics in sorghum are shown in Table 7. These data were obtained from the responses received from the participants attending the International Sorghum Workshop held at ICRISAT in 1977.

TABLE 6. Mineral and Trace Element Analysis (mg/100 9) of 100 Selected Sorghum Germ Plasm Samples

  Range Mean
Calcium* 5.5 - 53.1 26.2
Phosphorus 226.4 - 620.2 397.0
Magnesium 150.0 - 293.0 189.6
Sodium* 1.0 - 24.3 7.6
Potassium 323.8 - 801.4 509.1
Iron* 2.6 - 9.6 5.9
Copper 0.1 - 3.2 0.8
Zinc 1.9 - 5.7 3.3
Manganese 0.2 - 3.5 1.6

1977 data.
*97 samples.

Our economists are making an attempt to relate the market price with some of the known characteristics that are said to be associated with better food products prepared in the traditional way. We are also analysing some of the market samples for various chemical constituents in order to find out whether any relationship exists between the market price and evident or cryptic quality.

We are also concentrating on the characteristics of sorghum for making chapati (an unleavened bread), and preliminary information has been obtained on evident quality characteristics by our sorghum-breeding section. Contrasting samples from the preliminary screening will be taken for chemical analysis. Very little work has been done in this area, and we will be directing attention in our Institute toward finding out more about the characteristics affecting cooking quality.

Pearl Millet

Pearl millet is a highly cross-pollinated crop that is extensively grown in semi-arid tropical regions of the Indian sub-continent and Africa. The protein content of pearl millet varies from 8 to 23 per cent, lysine from 0.9 to 3.8 per cent, oil 2.8 to 8.0 per cent, and carbohydrates 59.7 to 74.5 per cent. Due to its high oil content, pearl millet flour develops rancidity, especially when stored under humid conditions. Many of the observations made for sorghum also apply for pearl millet. The methods tried for sorghum quality evaluation were also tried with pearl millet samples. The methodology is the same as that described under sorghum; results are reported here.

Protein estimation

The biuret method gave a high correlation with micro-Kjeldahl values, but again we observed that large variations occurred between micro-Kjeldahl and biuret values when this procedure was employed for routine screening. Therefore, the Technicon autoanalyser method (which gave a correlation of 0.97 with MKJ method) was used for protein estimation in pearl millet.

Lysine estimation

We have obtained a correlation of 0.92 between the UIR values obtained on samples adjusted to contain 80 mg protein and actual lysine concentration, determined by ion exchange chromatography. As described in the sorhgum section, we are now relating the ratio of the UIR reading obtained on constant weight of the sample and the protein values (UIR/P) with actual lysine concentration.

TABLE 7. Preferred Characteristics in Sorghum

Colour White/yellow
Size Large/bold
Hardness Hard/corneous
Nutritional qualities High-protein, high-lysine,
low-tannin, high feeding value
Other characteristics More dough elasticity, good
injera, couscous, to, chapati,
sweet, to blend with wheat

Responses received from participants in the International Sorghum Workshop, Hyderabad, 1977.

High-lysine pearl millet has not been reported so far. Also, our breeders were interested in obtaining basic data on the relationship between protein, yield, seed size, etc. However, protein content in several composites showed a wide range-from 8 to 15 per cent. Environment seems to play a major role in influencing protein content, and the relationship between yield and protein content was not very strong. Therefore, it seems possible to increase the protein content of pearl millet cultivars without much affecting the yield. More attention will be paid to this possibility during the coming years.

As discussed for sorghum, we are also looking at the consumer acceptance characteristics in pearl millet. Table 8 shows a list of various preparations made from pearl millet. These were tabulated from the responses we received from participants attending the International Pearl Millet Workshop in September 1977. Therefore, our efforts in determining evident quality characteristics as well as in improving the protein content and quality of pearl millet will continue.


For the rapid estimation of protein and lysine in large numbers of sorghum and pearl millet samples, the Technicon autoanalyser method and the dye-binding capacity method were found to be most suitable.

We need more data to understand whether the high-lysine gene in sorghum is stable in a normal plump seed endosperm background.

The relationship between yield and protein in pearl millet was not very strong. Therefore, it seems possible to increase the protein content without much affecting the yield.

TABLE 8. Main Preparations Made from Pearl Millet Grain for Human Consumption

I Chapati >50%
To, tuwo  
II Nadida-thin gruel  
Kali-cracked grain eaten as porridge and dumplings  
Kichri, rabri  
Cuppasa-thick pancake  
Burkutu-liquor drink  
Gadogado-Nigerian pancake  
Waina-fried-type cake  

International Pearl Millet Workshop, 1977.

Evident quality characteristics in sorghum and pearl millet are very important. Attempts are being made to understand more about the factors that govern the cooking quality and consumer acceptance characteristics in food products.


1. J.S. Kanwar and J.G. Ryan, "Recent Trends in World Sorghum and Millet Production and Some Possible Future Developments," paper presented at the Symposium on Production, Processing and Utilization of Maize, Sorghum and Millets, Central Food Technological Research Institute, Mysore, India, December 1976.

2. R. Singh and J.D. Axtell, "High Lysine Mutant Gene (hl) that Improves Protein Quality and Biological Value of Grain Sorghum," Crop Science, 13: 535 (1973)

3. R. Jambunathan, E.T. Mertz, and J.D. Axtell, "Fractionation of Soluble Proteins of High Lysine and Normal Sorghum Grain," Cereal Chemistry, 52: 119 (1975).

4. D.P. Mohan, "Chemically Induced High Lysine Mutants in Sorghum Bicolor (L) Moench," Ph.D. thesis, Purdue University, West Lafayette, Indiana (1975).

5. V. Guiragossian, B.A.K. Chibber, S. van Scoyoc, R. Jambunathan, E.T. Mertz, and J,D. Axtell, "Characteristics of Proteins from Normal, High Lysine, and High Tannin Sorghums," J. Agric. Food Chom., 26: 219 (1978).

6. R.M. Johnson and L.E. Craney, "Rapid Biuret Method for Protein Content in Grains," Cereal Chemistry, 48: 276 (1971).

7. D.C. Udy, "Improved Dye Method for Estimating Protein," J. Amer. Oil Chemists' Soc., 48: 29A (1971)

8. W.W. McMillen, B.R. Wiseman, R.E. Burns, H.B. Harris, and G.C. Green, "Bird Resistance in Diverse Germplasm of Sorghum," Agronomy J., 64: 821 (1972).

9. D.S McLaren, "The Great Protein Fiasco," Lancet, 2: 93 (1974) .

10. P.R. Payne, "Nutritional Criteria for Breeding and Selection of Crops: With Special Reference to Protein Quality," Plant f oozes for Man 2: 95 (1976)

11. National Institute of Nutrition, National Nutrition Monitoring Bureau disport for the Year 1975, Indian Council of Medical Research, Hyderabad (1976).

UNU-WHP Fellows Invited to Enter Prize Competition

The UNU World Hunger Programme is planning a meeting in Bellagio, Italy, 28 April to 2 May 1980 to discuss how the UNU might best use the instruments of scholarship, research, advanced training, and dissemination of knowledge and its flexibility and capacity to deal directly with institutions and individuals to make a significant impact on the world hunger problem. An invitation is extended to all current and former UNU Fellows to contribute a written essay on this subject. Contributions are limited to 1015 double-spaced typewritten pages and may be written in English, Spanish, or French. They should be received at UNU Headquarters in Tokyo not later than 15 March, First prize will be US$500, and the essay will be published in the Food and Nutrition Bulletin. Other entries considered publishable will receive a prize of US$200 each.


Contents - Previous - Next