Available online at www.scholarsresearchlibrary.

com

Scholars Research Library

Study of yield and yield components Cultivars Promising Grain Sorghum using cluster analysis and Factor analysis Ahad Jahangiri Ajirlou1, Shahrooz Aghaei1, and Shamsali Darvishi1*
1

Department of Agriculture, Parsabad, Moghan, Islamic Azad University, Parsabad Branch, Iran

____________________________________________________________________________________________ ABSTRACT 20 varieties of sorghum were evaluated in a randomized complete block design with four replications in 2007 in the region of the Moghan. Analysis of variance of traits showed that there is a significant difference among the varieties except for stamen leaf area. Using cluster analysis of Ward's method based on standardized data and all traits the Iranian varieties and BISC-11 from India were ranked superior clusters. This grouping was confirmed by the detection function. In factor analysis, five factors were determined which explained 86.24% of the total variation. The first main factor explained 33.890 % of the total variation was introduced as a performance factor. Keywords: Promising Varieties, Factor Analysis, Cluster Analysis, Sorghum, Yield ____________________________________________________________________________________________ INTRODUCTION Sorghum is one of the most important cereal crops in arid and semi-arid. The plant in the world after wheat, rice, maize and barley is in the fifth place [1]. This plant is used for providing protein for many people in Asia and Africa [2], malt production of non-alcoholic drinks, flour production and animal feed [3]. Cultivation of sorghum in the world in 2007 was nearly 47 million hectares of which 90% of the cultivated area is dedicated for Grain sorghum varieties. Therefore, sorghum is considered as the world's primary cereal. India, with about 9 million hectares under cultivation in first place and the USA with 3 million hectares under cultivation have the greatest production in the world [4]. The success of the breeders depends on the choice of appropriate materials and diversity. Breeding those traits which have high heritability is more important. It is notable that the evaluation and application of the results have a significant role in Agricultural Sciences [5]. The purpose of principal component analysis is to reduce the volume of data. In this method examining the correlation between variable we are able to realize the relationship between the traits. In component analysis, the correlation between lots of dependent variables is expressed by a few independent components. The role of each trait is determined in variation of each trait. In addition, the principal component analysis is used for genotype classification [6]. Several reports of increasing genetic distance between genotypes of a species increases the likelihood of heterosis in cross-linkage programs. The genotype classification based on genetic distance, is effective in corrective programs when such traits simultaneously are examined. Therefore, in order to determine genetic variation standard, genotype classifications and genetic distance among them the Cluster analysis is done [7]. MATERIALS AND METHODS The study examined 20 varieties of sorghum, of which the first 17 variety were Iranian, two Indian and one Thai. The land under the test immediately plowed after wheat harvest in May 2007. The amount and type of fertilizer according to soil samples were added to the land. Other operations for Seedbed preparation were performed. Stacks were constructed within 60 cm of each other and planting was done. Each cultivar in each plot was planted in four

201 Scholars Research Library

Shamsali Darvishi et al

Annals of Biological Research, 2013, 4 (8):201-204

_____________________________________________________________________________
rows of seven meters in length. After emergence, the distance between they were narrowed to six centimeter apart in the phase of 6-4 leaves. Therefore, on each seven meters line remained 118 plants. Plants at a height of about 40 cm were given 150 kg of Serk nitrogen fertilizer on the banks of the stacks in the form of ribbon. Then permeable irrigation was done. The irrigation period determined once every 7-10 days. In order to classify varieties based on the entire traits and standardized data of Ward's method cluster analysis using squared Euclidean distance based on the mean standardized data were carried out. To determine the appropriate location for cutting the dendrogram, discriminant function analysis was used. To obtain more information on the relationship between various characteristics and profound understanding of data structures, factor analysis was used. Data analysis was performed using SPSS-16 software. RESULTS AND DISCUSSION Cluster Analysis Cluster analysis of all varieties of seeding sorghum based on all traits was performed using Ward's method with standardized data. Based on the results of the discriminant function analysis on various sections of the cut, the maximum difference amongst the group was observed in two clusters (Table 1 and Figure 1). The first cluster includes varieties KGS-29, KGS-3, KGS-11, KGS-17, KGS-24, KGS-15, KGS-31, KGS-9, KGS-12, KGS-19, KGS-32, BICS-11, KGS-23, KGS-33, KGS-25, KGS-5, KGS-27 and KGS-36 and the second cluster included genotypes ICSV-274 and UT-378B, respectively. Variety classification of the experiments showed the good nature of the classification in geographical distribution due to exposure of foreign varieties groups in similar groups. To demonstrate the effectiveness of each trait in distinction of each cluster, the mean of each cluster, and the average deviation from total mean and total mean were calculated for each of the traits (Table 2). In the first cluster, the value of yield, harvest index, plant biomass, seed weight, number of grains per panicle, number of branches per panicle, panicle length and stem diameter was greater than the mean of the whole group. In the second cluster, number of tillers, plant height and stamen leaf area had a value greater than the mean of the entire group.

Figure 1. The dendrogram of sorghum cultivars resulting from cluster analysis using Ward Method based on standardized data of all traits TABLE 1. Discriminant function analysis to determine the cut-off point of dendrogram resulting from cluster analysis based on all standardized traits
Number of groups 2 3 Eigenvalues 140.220 5.910 Percent of variance 94.6 5.4 Canonical Correlation 0.995 0.925 Wilkes Lambda 0.001 0.145 Probability 0.000 0.05

202 Scholars Research Library

Shamsali Darvishi et al

Annals of Biological Research, 2013, 4 (8):201-204

_____________________________________________________________________________
Table 2. Groups average and their deviation percentage based on all traits of 20 cultivars of sorghum.
Group Average Group 1 Performance Harvest index Plant biomass Seed weight Number of grains per panicle Branches per panicle Panicle Length Days to emergence Number of tillers Stem diameter Plant height Stamen leaf area 5.15 51.44 77.62 29.19 1431 84.85 27.97 65.61 0.26 7.241 124.9 146.3 5.94 5.56 4.9 3.43 8.2 0.46 0.61 -0.06 -49.0 0.85 -4.6 -0.85 2.765 24.305 37541 28.8 346.75 80.95 25.25 66.0 2.77 6.605 185 158.35 Deviation of total mean Group Average 2 -53.47 050.12 -44.0 -1.2 -73.78 -4.15 -9.17 0.53 443.0 -8.0 41.32 25.65 4.868 48.729 73.99 29.15 1322.85 84.46 27.87 65.65 0.51 7.177 130.9 147.505 Deviation of total mean total mean

FACTOR ANALYSIS In factor analysis, five factors which accounted for a total of 86.244% of the total variation were selected (Table 3). In this analysis, factors with eigenvalues greater than one were selected. Validity of the factor selection was confirmed by Scree graph (Figure 1). In this study, the first main factor which explained 33.890% of the variation had high correlation with traits such as yield, harvest index, biomass, plant height, number of tillers, and number of grains per panicle. Therefore the first factor can be introduced as a performance factor. It is noteworthy that this factor had a high negative correlation with Plant height and Number of tillers. The second factor accounted for 16.232% variation, had a high positive correlation with stem diameter and number of days to panicle emergence. Therefore the second factor can be introduced as an effective factor for growth. The third factor explained 13.999% along with the second and first factor with 64.120% of total variance having a high positive correlation with grain weight and biomass. And other traits in selection of cultivars through this factor were less important factors. The fourth factor with 13.612 percent variation (along with the first three, 77.732 percent variation) accounted for which had a high positive correlation with panicle length and panicle branches. And other traits in selection of cultivars through this factor were less important factors. The fifth and the first four factors explained 8.512% and 86.244% of the total variance, respectively. This factor had high positive correlation with stamen leaf area. Other traits were less important factor in selecting varieties. Table 3 - Factor Coefficients, Eigenvalues and Cumulative Changes Principal Factors with Varimax Rotation.
Traits Stamen leaf area Plant height Stem diameter Number of tillers Days to panicle emergence Panicle Length Branches per panicle Number of grains per panicle Seed weight Biomass Harvest index Performance Eigenvalues Cumulative variation Factor 1 -0.049 -0.885 0.360 -0.847 0.031 0.233 -0.030 0.900 -0.196 0.749 0.899 0.790 4.745 33.890 Factor 2 -0.111 -0.102 0.708 -0.249 0.903 0.115 -0.053 -0.187 0.216 -0.046 -0.064 0.009 2.272 50.122 Eigenvalues Vectors Factor 3 0.021 -0.237 0.427 -0.076 -0.064 -0.036 -0.019 -0.155 0.819 0.619 0.026 0.109 1.960 64.120 Factor 4 0.112 -0.058 -0.033 -0.027 0.185 0.923 0.957 0.158 -0.138 -0.100 0.112 -0.034 1.906 77.732 Factor 5 0.933 0.235 -0.109 -0.132 0.111 0.026 0.096 -0.135 0.111 -0.065 -0.108 0.327 1.192 88.244

203 Scholars Research Library

Shamsali Darvishi et al

Annals of Biological Research, 2013, 4 (8):201-204

_____________________________________________________________________________

Scree Plot

Eigenvalue

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Component Number

Figure 1 - Scree graph obtained from the Factor analysis CONCLUSION There was a significant difference between experimented cultivars for most of studied traits indicating the difference between seeding sorghum varieties. So this provides selection, breeding and introduction of high yielding cultivars in different regions. In cluster analysis using Ward's method with standardized data varieties classified in two clusters. Varieties of cluster one were superior in terms of yield. Variety classification of the experiments showed the good nature of the classification in geographical distribution due to exposure of foreign varieties groups in similar groups. In factor analysis the first factor was introduced as the performance factor with 33.890% of the total variation. REFERENCES [1] Almodares, A., Taheri, R., and Safavi, V. 2008. Sorghum. Isfahan Jahad. Daneshgahi Press First Edition. 263pp. [ 2] Belton, D. S., and Taylor, J. R. N. 2004. Trends in Food Science and Technology. 15: 94-98. [3] Defoor, D. J., Cole, N. A., Galgean, M. L., and Jones, O. R. 2001. Journal of Animal Science. 79: 19-25. [4] Anonymous, Food and Agricultural Organization. 2007. Crops production, sorghum harvesting area, Retrieved November, 15, 2009, from http: //www.fao.org/crops production. [5] Borojevic, S., 1990. crop Sci. 17:145-152. [6] Jafari, AS., Nosrati Nygjh, M.. Heidari Sharif Abad, H., 2003. Quarterly Journal of genetics and plant breeding for forestry Iran (11): 63-103, published by Research Institute of Forests and Rangelands, Tehran. [7] Farshadfar, AS., 1998. Application of Quantitative Genetics in Plant Breeding (Volume I), publications, Tagh Bostan, Razi University.

204 Scholars Research Library