SciELO - Scientific Electronic Library Online

vol.44 issue3Combined sampling for regeneration of plant genetic resources of monoecious species with natural pollinationCorn barriers in an integrated management strategy to control epidemics of papaya ring spot virus (PRSV-P) author indexsubject indexsearch form
Home Pagealphabetic serial listing  

Services on Demand




Related links

  • Have no similar articlesSimilars in SciELO



On-line version ISSN 2521-9766Print version ISSN 1405-3195

Agrociencia vol.44 n.3 México Apr./May. 2010




Analysis of grain density and yield characters in aromatic rice genotypes


Análisis de densidad del grano y caracteres del rendimiento en genotipos de arroz aromático


S. M. Shahidullah1,2*, M. M. Hanafi1, M. Ashrafuzzaman1, M. K. Uddin1, M. Razi–Ismail1


1 Institute of Tropical Agriculture, Universiti Putra Malaysia. 43400 UPM, Serdang, Selangor, Malaysia.

2 Bangladesh Rice Research Institute Regional Station, Sonagazi. Feni 3930, Bangladesh. *Autor responsable: (


Received: June, 2009.
Approved: November, 2009.



Aromatic rices (Oryza sativa L.) normally produce enormous spikelets of low grain weight resulting low yield. Grain weight and grain density are the vital indicators of grain filling and mainly dependent on genetic make–up. Detail study of grain density and yield components of aromatic rice is a crucial need for yield improvement. An experiment was conducted with 40 rice genotypes to assess the grain density of aromatic rices and to observe the relationships of different grain categories with yield characters. The percentage of high density grains, good grains, average grains, poor grains, partially filled grains, spikelet sterility, and thousand grain weight were found to vary in a great extent. Maximum 50 % high density grain was harvested in BR39 and a minimum 1 % in Khazar. Thousand grain weight ranged from 10 g to 30 g. The proportion of high density grains was positively correlated with thousand grain weight (r =0.57) and grain yield (r =0.57). The negative relations were found between spikelet sterility and grain yield (r = —0.47) and also between number of grains and thousand grain weight (r = —0.74). The grain yield showed the highest positive direct relation (0.23) with thousand grain weight, followed by high density grains (0.20).

Key words: Oryza sativa, aromatic rice, high density grain, grain yield.



Los arroces aromáticos (Oryza sativa L.) normalmente producen espiguillas enormes con bajo peso de los granos, que resultan en un bajo rendimiento. El peso de los granos y la densidad de los granos son indicadores vitales del llenado de los granos, y dependen principalmente de la constitución genética. El estudio detallado de la densidad del grano y los componentes del rendimiento del arroz aromático es crucial para mejorar el rendimiento. Se realizó un experimento con 40 genotipos de arroz para evaluar la densidad del grano en arroces aromáticos y para observar las relaciones de diversas categorías de granos con los caracteres del rendimiento. Se encontró que el porcentaje de granos de alta densidad, granos buenos, granos promedio, granos malos, granos parcialmente llenos, la esterilidad de espiguilla y el peso de mil granos variaron en gran medida. Un máximo de 50 % de granos de densidad alta se cosechó en BR39, y una mínima de 1 % en Khazar. El peso de mil granos varió de 10 g a 30 g. La proporción de granos de alta densidad se correlacionó positivamente con peso de mil granos (r =0.57) y con rendimiento del grano (r =0.57). Se encontraron relaciones negativas entre esterilidad de la espiguilla y rendimiento del grano (r = —0.47) y también entre número de granos y peso de mil granos (r = —0.74). El rendimiento del grano mostró la más alta relación directa y positiva (0.23) con el peso de mil granos, seguido de los granos de alta densidad (0.20).

Palabras clave: Oryza sativa, arroz aromático, granos de alta densidad, rendimiento del grano.



Aromatic rice is characterized by the presence of scent and often slender in shape. Since the dawn of civilization, thousands of locally adapted genotypes of aromatic rices have evolved because of natural and human selection (Singh et al., 2000). Aromatic rices constitute a small group that is usually used for special dish preparation in festivals and special occasions rather than in common and regular purpose. In addition, it is a high value cash crop for farmers (Singh et al., 2000). These rices emit aroma in fields, at harvesting, in storage, during milling, cooking and eating (Shahidullah et al., 2009). Aroma development is influenced by both genetic factors and environment. Pleasant aroma is a result of a large number of compounds present in specific proportion (Yajima et al., 1978). A popcorn like aroma component 2–acetyle–1–pyrroline, is an important flavour component of several aromatic varieties (Laksanlamai and Ilangantileke, 1993).

Most of the aromatic rice varieties are low yielding because of its traditional plant type, low grain weight but with enormous spikelets. Grain density is normally higher for larger grain size, however it is also dependent on grain filling. Different rice genotypes with same grain size are found to have different grain density (Yamamoto et al., 1991). This indicates that poor grain filling found in the cultivars with enormous spikelets could be genetically improved through utilizing available genetic resources. It was also suggested that the rate and duration of grain filling in rice affect final grain traits such as weight and density (Yang et al., 2008; Jongkaewwattana and Geng, 2001; Wang et al., 1999).

Information obtained from correlation coefficients can be augmented by partitioning the relationships into direct and indirect effect for a given set of relationships a prior cause–and–effect. In such situations, the correlation coefficients may be confounded with indirect effects due to common association inherent in trait interrelationships. Path coefficient analysis has proven useful in providing information that describes prior cause–and–effect relationships, such as rice yield and components (Vlek et al., 1979). The relationships among different components and on path analysis for wet–seeded and irrigated rices has been evaluated (Wells and Faw, 1978; Jongkaewwattana and Geng, 2001; Huan et al., 1999). However, such type of studies on grain density of aromatic rices have not been performed. So, the objective of this research was to evaluate the relationships among grain density and yield components and to determine the effect of the factors on grain yield of aromatic rice genotypes.



The experiment was conducted at the farm of Bangladesh Rice Research Institute (BRRI), Gazipur, in transplant Aman (lowland rice culture in wet season), July–December, 2005. Forty rice genotypes composed of 32 local aromatic, five exotic and three non–aromatic rice varieties as standard checks, were selected for this research (Table 1). Non–aromatic varieties were BR28, BR39 and Nizersail. Local improved Nizersail is a standard photoperiod sensitive variety whereas BR28 and BR39 are modern varieties released for extensive commercial cultivation in Bangladesh. Exotic genotypes were collected from Pakistan (Basmati PNR346), Nepal (Sarwati and Sugandha–1) and Iran (Khazar and Neimat). The rest of the materials are representing their distribution throughout Bangladesh. Forty rice genotypes formed the treatment variables and were assigned randomly to each unit plot of 5 m X 2 m dimension.

Thirty day–old seedlings were transplanted on the 15th August, 2005, according to a randomized complete block design (RCBD) with three replications. Transplanting was done at the spacing of 20 cm X 20 cm. A fertilizer rate of 25–35–10–3 kg ha–1 of P–K–S–Zn as triple super phosphate, muriate of potash, gypsum and zinc sulphate, was applied at final land preparation (BRRI, 1995). Because of wide genotypic variation in phenological development and yield potential, varieties differed enormously in attaining panicle initiation (PI) stage and in the requirement of nutrient elements. For this reason, nitrogen was top–dressed as urea in 2–3 splits instead of the common dose with fixed time routine. The amount of urea and time of application were determined with a leaf colour chart (Ladha et al., 1998). The dose of total N application was 40 to 75 kg ha–1.

All the panicles of randomly selected four plants (hills) in each plot were used as a sample to count the number of grains (Gomez, 1972). A random sample of 1000 well–developed, whole grains at 14 % moisture content was weighed on an electronic balance to determine of grain weight. Spikelet sterility was calculated from four plant samples per plot:

Plants were harvested at crop maturity (116 and 154 d after sowing depending on genotypes). All the plants of a 5 m2 sample area were cut at base. After threshing and cleaning, the fresh weight of grains was recorded and adjusted to 14 % moisture content:

where, FW= fresh weight of the grains; MC=% moisture in the fresh grains.

Rice grains of a variety are graded according to the density:

Twenty gram samples from the whole plot harvest were used to sort out grains of different specific gravity. The solutions of different specific gravity were prepared using sodium chloride in tap water. The specific gravity of tap water is 1.0. To raise the level of specific gravity by 0.02 30 g salt L–1 water are required (Rao et al., 1985; Rao, 1991; Venkateswarlu et al., 1986); with this rate, solutions of 1.06, 1.14 and 1.20 specific gravity were prepared. The necessary adjustments were made with salt to maintain the specific gravity levels while standardizing with hydrometer. The sample was placed in a beaker containing 1 L normal tap water, and stirred. The floating grains were taken out with a plastic mesh and pressed with the finger to identify chaff and partially filled. The submerged grains were transferred to 1.06 specific gravity solution and the floated grains were separated. The process was continued for 1.14 and 1.20 specific gravity levels and the floated grains were counted in each of solution levels for specific grades.

Data on 10 characters related to grain density and yield variables (% HDG, % good grains, % average grains, % poor grains, % partially filled grains, % sterility, number of panicles m–2 , number of filled grains panicle–1 , 1000–grain weight, and grain yield) were subjected to several statistical analyses to interpret the results. The ANOVA and some descriptive statistics were performed through IRRISTAT Windows 4.01 and Microsoft ® Office Excel ®. Genotypic (GCV) and phenotypic (PCV) coefficients of variation were estimated according to Burton (1952) as follows:

where = genotypic variance; where = phenotypic variance; = population mean.

To understand the association between any two variables simple correlation (r) was calculated from average data:

where COVxy = covariance between the characters x and y; = variance of the character x; = variance of the character y.

Path analysis was performed according to Singh (2000); a series of simultaneous equations are constructed using the estimates of simple correlation coefficients (r):

where r12, r13 etc., are the estimates of simple correlation coefficients between, traits 1 and 2, 1 and 3 etc., and P17, P27 are the estimates of direct effects of trait 1, 2, on the dependent variable i.e. trait 7 (grain yield in this case). After placing the values of correlation coefficients in the equations, direct and indirect effects of component traits are estimated by the process of elimination.



Genotypes characterization

Ten characters belonging to grain density and yield variables are shown in Tables 1 and 2. High density grain (HDG), having specific gravity higher than 1.20, is a highly desirable trait in a rice cultivar. Rice genotypes varied to a great extent for the number and percentage of HDG: minimum (0.51 %) for Khazar and maximum (49.61 %) for BR39. Khazar possessed a long slender grain with a thousand grain weight over 22 g (Table 2). In spite of larger grain size, it seldom bears high density grain. Grain density is dependent on genetic makeup, but it is influenced by nutritional factors to some extent. Translocation or upward movement of nutrients may affect grain filling and hence grain density (Murthy and Murthy, 1982). Our results agree with the findings reported BRRI (1995), that BR14 showed 31–36 % high density grains, whereas Pajam and BR5 had 1–5 %; Tulsimala had no high density grain at all. In the present study, enormous variations were found in the genotypes for the portion of good grains (specific gravity = 1.14–1.20), from 3.97 % to 51.36 %. A notable observation is that Khazar holds the poorest position regarding good grains. Maximum percentage of good grains was produced by Begun bichi with short bold grain type having 10.92 g of thousand grain weight (Tables 1 and 2). The average grade of grains with specific gravity 1.06–1.14 was also found to vary in a large extent. The maximum portion of average grain was produced by Baoi jhak (32.71 %) and the minimum by BR39 (5.04 %). The range of poor grains was 3.09 % (BR28) to 25.64 % (Khazar). This data indicates the best grain filling in the modern variety BR28. The highest level of partially filled grains was observed in Gandho kasturi (14.97 %). The genotypes Gandho kasturi and Benaful hold the highest thousand grain weight, over 30 g (Table 2). Grain filling of the genotype might be affected by its extreme late maturity (data not shown).

Yield components

Number of panicles per unit area is the most important component of rice yield and it accounts for 89 % of the variation of grain yield (Miller et al., 1991; Yoshida et al., 1972). In our study, number of panicles m–2 varied from 90 (Khazar) to 286 (Nizersail). The total number of filled grains per panicle (the aggregate of different grades of spikelets) also differed markedly, between 70 (Benaful) and 187 (Kamini soru). The highest spikelet sterility was 44 % in Khazar and the lowest 8 % in Gandho kasturi (Table 2). Combined effect of spikelet sterility as well as lower grain density leads the cultivar to produce the lowest grain yield (1.42 t ha–1 ). Among the characters studied, thousand grain weight showed the largest variation: lowest, 9.99 g in Rajbhog; highest, 30.48 in Gandho kasturi. In aromatic rices a lower thousand grain weight is preferred; by natural selection, most of the aromatic rice races show a low yield, as well as lower grain weight. Grain yield is important for a producer; in our study it ranged between 1.42 to 4.21 t ha–1 . Aromatic rice is considered as the best in quality; so, its lower yield could be accepted to satisfy consumers' demand (Singh et al., 2000).

The genotypic and phenotypic coefficients of variations for each character are shown in the Tables 1 and 2. The higest GCV was 36 % for thousand grain weight, followed by spikelet sterility (33 %) and the percentage of good grains (32 %). The lowest value of GCV was observed for the number of panicles per square meter (15 %) followed by grain yield (16 %) and the percentage of partially filled grains (24 %). The GCV for the rest of the characters ranged between 25 to 30 %. Higher GCV in a character gives a better opportunity for a cross combination to obtain a wider variation. Most of the characters showed little differences between PCV and GCV which indicated negligible influence of environment on the expressions of these characters. However, spikelet sterility showed slightly higher differences between GCV and PCV, indicating comparatively higher influence of environments on the expression of the characters. Low values of GCV and PCV have been reported for plant height and panicle length of wheat (Das and Rahman, 1984). Amin et al. (1992) observed closeness of PCV and GCV for a few characters and a large difference between PCV and GCV for others.

Relationships among the traits

Relationships among the different grades of spikelets and yield related variables were determined through simple correlation coefficient (r) and 10 characters were subjected to correlation matrix (Table 3). Among the 45 correlation coefficient values, 3 were positive and 7 negative (p <0.05). The highest positive r value (0.58) was recorded between the number of filled grains per panicle and the percentage of good grains. The percentage of high density grains was positively correlated with thousand grain weight (r = 0.57) and grain yield (r =0.57). Neither the number of panicles m–2 nor the percentage of partially filled grains were correlated to any other characters. The percentage of high density grains showed negative relationship between average grade of grains (r = —0.74) and the percentage of poor grade of grains (r = —0.52). Finally, grain yield showed a significant positive relationship with high density grain and a negative with spikelet sterility (r = —0.47) and with the percentage of poor grains (r = —0.49). Positive relationship of high density grain with grain yield and thousand grain weight, and negative relationship between the number of grains and thousand grain weight and also between spikelet sterility and grain yield are shown in Figure 1. Different levels of correlation have been reported for grain yield and related components of modern, lowland and winter rice varieties (Bai et al., 1992; Manuel and Palanisamy 1989; Vange et al., 1999).

Effects of traits on grain yield

Most contributing six variables were subjected to path analysis where correlation coefficients were partitioned. Direct and indirect effects were quantified and they are shown in Table 4. Thousand grain weight was found to have the highest correlation (0.23) with grain yield, the percentage of high density grain was second (0.20) followed by the number of panicles (0.17). The most negative correlation ( 0.32) was found between spikelet sterility and grain yield. In this analysis it is observed that a considerable portion of effects were unexplained and remained as residual effect (0.69). Kumar et al. (1998) showed direct and indirect effects of component characters on yield, for some rice cultivars across different locations in winter.



Grain density of aromatic rices vary over the range of genotypes. Higher grain weight is a crude indicator of heavier specific gravity. However, grain density may not follow the pattern of grain weight. Grain yield showed the highest positive correlation with thousand grain weight, followed by the percentage of high density grains.



Amin, M. R., D. C. N. Barma, and M.A. Razzaque. 1992. Variability, heritability, genetic advance and correlation in durum wheat. RACHIS 11: 30–32.        [ Links ]

Bai, N. R., R. Devika, A. Regina, and C. A. Joseph. 1992. Correlation of yield and yield components in medium duration rice cultivars. Environ. Ecol. 10(2): 459–470.        [ Links ]

BRRI (Bangladesh Rice Research Institute). 1995. Annual Report for 1995. Publication No. 126. Bangladesh Rice Res. Inst., Gazipur, Bangladesh. 253 p.        [ Links ]

Burton, G. M. 1952. Quantitative inheritance in grasses. Proc. 6th Int. Grassland Congr. 6: 277–283.        [ Links ]

Das, M. K., and L. Rahman. 1984. Estimates of genotypic and phenotypic variability, heritability and genetic advance in common wheat. Bangladesh J. Agric. Res. 9(1): 15–18.        [ Links ]

Gomez, K. A. 1972. Techniques for field experiments with rice. Int. Rice Res. Inst., Los Baños, Laguna, Philippines. 46 p.        [ Links ]

Huan, T. T. N., T. Q. Khuong, P. S. Tan, and H. Hiraoka. 1999. Path–coefficient analysis of direct–seeded rice yield and yield components as affected by seeding rates. OmonRice. 7: 104–110.        [ Links ]

Jongkaewwattana, S., and S. Geng. 2001. Inter–relationships amongst grain characteristics, grain filling parameters and rice (Oryza sativa L.) milling quality. J. Agron. Crop Sci. 187: 223–229.        [ Links ]

Kumar, G. S., M. Mahadevappa, and M. Rudraradhya. 1998. Studies on genetic variability, correlation and path analysis in rice during winter across the locations. Karnataka J. Agric. Sci. 11(1): 73–77.        [ Links ]

Ladha, J. K., G. J. D. Kirk, J. Bennett, S. Peng, C. K. Reddy,P. M. Reddy, and U. Singh. 1998. Opportunities for increased nitrogen use efficiency from improved lowland rice germplasm. Field Crops Res. 56: 41–71.        [ Links ]

Laksanlamai, V., and S. Ilangantileke. 1993. Comparison of aroma compound 2–acetyle–1–pyrroline in leaves from pandan. Cereal Chem. 70: 381–384.        [ Links ]

Manuel, W. W., and S. Palanisamy. 1989. Heterosis and correlation in rice. Oryza 26(3): 238–242.        [ Links ]

Miller, B. C., J. E. Hill, and S. R. Roberts. 1991. Plant population effects on growth and yield in water seeded rice. Agron. J. 83: 291–297.        [ Links ]

Murthy, P. S. S., and K. S. Murthy. 1982. Influence of source and sink on spikelet sterility in rice. Madras Agric. J. 10: 104–108.        [ Links ]

Rao, S. P. 1991. Influence of source and sink on the production of high density grain and yield in rice. Indian J. Plant Physiol. XXXIV(4): 339–348.        [ Links ]

Rao, S. P. , B. Venkateswarlu, and T. L. Acharyulu. 1985. Screening technique for differentiating the degree of spikelet filling in rice. Plant and Soil 88: 289–293.        [ Links ]

Singh, B. D. 2000. Biometrical techniques in plant breeding. In: Plant Breeding –Principles and Methods. Kalyani Publishers, Rajinder Nagar, Ludhiana 141008, India. pp: 107–131.        [ Links ]

Singh, R. K., U. S. Singh, and G. S. Khush. 2000. Aromatic Rices. Oxford and IBH publishing Co. Pvt. Ltd. New Delhi. 292 p.        [ Links ]

Shahidullah, S. M., M. M. Hanafi, M. Ashrafuzzaman, M. Razi Ismail, and A. Khair. 2009. Genetic diversity in grain quality and nutrition of aromatic rices. Afr. J. Biotechnol. 8(7): 1238–1246.        [ Links ]

Vange, T., A. A. Ojo, and L. L. Bello. 1999. Genetic variability, stability and correlation studies in lowland rice (Oryza sativa L.) genotypes. Indian J. Agric. Sci. 69(1): 30–33.        [ Links ]

Venkateswarlu B., B. S. Vergara, F. T. Parao, and R. M. Visperas. 1986. Enhancing grain yield potentials in rice by increasing the number of high density grains. Philipp. J. Crop Sci. 11: 145–152.        [ Links ]

Vlek, P. L. G., C. W. Hong, and L. J. Y. Youngdahl. 1979. An analysis of N nutrition of yield and yield components for the improvement of rice fertilization in Korea. Agron. J. 71: 829–833.        [ Links ]

Wang, G., M. S. Kang, and O. Moreno. 1999. Genetic analyses of grain–filling rate and duration in maize. Field Crops Res. 61: 211–222.        [ Links ]

Wells, B. R., and W. Faw. 1978. Short–statured rice response to seeding and N rates. Agron. J. 70: 477–480.        [ Links ]

Yajima, I., T. Yanai, and M. Nakamura. 1978. Volatile flavour components of cooked rice. Agric. Biol. Chem. 42: 1229–1233.        [ Links ]

Yamamoto Y., T. Yoshida, T. Enomoto, and Yoshikawa G. 1991. Characteristics for the efficiency of spikelet production and the ripening in high–yielding japonica–indica hybrid and semidwarf indica rice varieties. Japan. J. Crop Sci. 60: 365–372.        [ Links ]

Yang, W. , S. Peng, Maribel L. Dionision–See, Rebecca C. Laza, and Romeo M. Visperas. 2008. Grain filling duration, a crucial determinant of genotypic variation of grain yield in field–grown tropical irrigated rice. Field Crops Res. 105: 221–227.        [ Links ]

Yoshida, S., J. H. Cock, and F. T. Parao. 1972. Physiological aspects of high yields. In: Rice Breeding. International Rice Research Institute, Los Baños, Philippines. pp: 455–469.        [ Links ]

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License