 |
| About Bioline | All Journals | Testimonials | Membership | News |
|
||||||
|
||||||
Vol.5. No.2, pp. 127-133, 1997 Multivariate analysis of genetic diversity in kenaf, Hibiscus cannabinus (L.)
B. A. OGUNBODEDE Institute of Agricultural Research and Training, Obafemi Awolowo University, P.M.B. 5029, Ibadan, Nigeria
(Recieved 5 June, 1996; accepted 31 December, 1996)
Code Number: CS97019
Sizes of Files:
Text: 24.5K
Graphics: Line drawings (gif) - 9.9K
ABSTRACT Fifty-four (54) accessions of kenaf of diverse eco-geographical origins were evaluated in an 8 x 8 lattice design in three environments in south-western Nigeria. Two multivariate techniques - the coefficient of racial likeness (CRL) and principal component analysis (PCA) were used to assess the extent of genetic divergence among the accessions. The CRL distances for the 1431 possible pairs of accessions were each less than 2.0. This does not necessarily suggest lack of genetic diversity among the accessions. The first three principal axess accounted for 67.17% of the total variation among the accessions. From a two-dimensional ordination of the first two principal axes, six clusters can be identified. Clustering was closely related to average CRL values and there was no relationship between clustering and eco-geographical distributions. The analysis of CRL values showed that butt-diameter and fiber yield each accounted for 13.0% of the variation detected in the accessions. Plant height and retting percentage each contributed 11.3% while core weight, number of leaves per plant, core percentage and fresh plant weight contributed 12.5, 10.3, 10.2 and 10.0%, respectively. The implications of these findings in kenaf improvement programmes are discussed.
Key Words: Coefficient of racial likeness. genetic diversity, principal component analysis
RESUME
Cinquante-quatre (54) echantillons de kenaf provenant de differents milieux geographiques ont ete evalues dans un quadrat de 8 x 8 placettes dans trois environnements du sud-ouest du Nigeria. Deux techniques multivariees - le coefficient de similitude raciale (CRL) et l'analyse de la composante principale (PCA) ont ete utilisees pour evaluer l'importance de la divergence entre les echantillons. Les distances CRL de 1431 paires d'echantillons possibles se sont sevelees toutes inferieures a 2.0. Cela ne signifie pas necessairement un manque de diversite genetique entre les echantillons. Les premiers trois principaux axes representaient 67.17% de la variation totale entre les echantillons. A partir d'une ordination a deux dimensions de deux premiers principaux axes, on peut identifier six groupes. Le groupage etait etroitement lie aux valeurs moyennes CRL. Il n'y avait donc pas de lien entre le groupage et les distributions eco-geographiques. L'analyse des valeurs CRL a montre que le diametre des souches et la production totale de fibre representaient 13.0% de la variation detectee parmi les echantillons. La hauteur de la plante et le pourcentage total de l'evaluation ont contribue chacun pour 11.3%. Par contre, le poids de la partie centrale, le nombre de feuilles par plante, le pourcentage de la partie centrale et la poids frais de la plante ont contribue pour 12.5, 10.3, 10.2 et 10.0% respectivement. Les implications de ces resultats dans les programmes d'amelioration du kenaf sont discutes.
Mots Cles: Coefficient de similitude raciale, divernsite genetique, analyse de la composante principale
INTRODUCTION
Development of improved varieties of crop plants necessarily involves the incorporation of specific gene complexes governing desired traits e.g. pest resistance, abiotic stress tolerance, adaptability to the environment, product quality, etc. The superiority of segregants resulting from selection for these critical sometimes mono- but often oligo-genic traits depends to a large extent on the genetic diversity of the initial population. Thus, the wealth of any germplasm collection is measured in terms of the genetic variability of the crop species it contains.
Many techniques have been employed by researchers to determine the extent of variability in a germplasm collection. The coefficient of racial likeness (CRL) proposed by Pearson (1926) and principal component analysis (PCA) have been used to measure genetic divergence among genotypes. The coefficient of racial likeness technique standardises the difference between pairs of observations by dividing them with the within-variety standard error before combining them in Euclidean numerical fashion (Ariyo, 1990). Principal component analysis, a common ordinational numerical technique, reduces the dimensions of multivariate data by removing intercorrelations among attribute-variables (characters on which units are to be compared), and enables multidimensional relationships to be plotted on two or three principal axes (Hayman, 1967). Principal component analysis chooses independent or orthogonal axes, which are minimally correlated and represents linear combinations of the original characters (Clifford and Stephenson, 1975). The relatively discriminating power of axes and their associated characters are measured by eigen-values and factor scores, respectively. Principal component analysis was used as a descriptive method to show patterns of covariation of characters among cultivars. An intecorrelation matrix among characters with 1.0's in the diagonal of the correlation matrix, is partitioned into components to account for a maximum amount of variance of the characters. To facilitate the interpretation of the results, the components can be rotated mathematically so that a given character tends to show its greatest contribution on a given component (Rhodes and Martins, 1972). Detailed accounts of PCA are described by Hayman (1967). Principal component analysis reliably represents large distances between major groups, but not between closely spaced units within the groups (Akoroda, 1983). This technique has been used to partition observed variation in many crops including cocoa (Theobroma cacao, Falowo, 1982) and maize(Zea mays L., Meyer et al., 1991).
The objectives of this study were to determine the extent of genetic variability among fifty-four (54) kenaf accessions and to compare the relative effectiveness of the two techniques.
MATERIALS AND METHODS
The fifty-four accessions of kenaf used in this study consisted of 13 accessions from Australia, 3 from Cuba, 2 from Guatemala, 3 from Taiwan 3 from the United States of America and the rest collected from different parts of Nigeria (Table 1). The exotic lines were imported over three decades ago and have continuously been cultivated for most of the period. Following land preparation, these kenaf accessions were evaluated in 8 x 8 incomplete lattice design with three replications in two locations (Ibadan and Ilora) in 1989 and in only one location (Ibadan) in 1990. Plot size was 4m x 4m with a spacing of 25 cm between and 10cm within rows resulting in a population of 400,000 plants per hectare. NPK fertilizer (25-10-0) at the rate of 200 kg ha^-1, i.e., 50 kg N and 20 kg P205 ha^-1 were applied about three weeks after plantinting. The plants were rainfed. Total precipitation during the growing season were 992.6 and 733.4mm for Ibadan (rainforest ecology) in 1989 and 1990, respectively, and 662.0mm for Ilora (derived savanna ecology) in 1989. Data were collected on nine characters at 25% flowering (about 75 days after planting). The characters included plant height, butt (stem) diameter, number of leaves per plant, number of nodes and fiber yield. These characters were chosen since previous studies (Ogunbodede, 1990) indicated that many of the characters were correlated with fiber yield in kenaf. The level of genetic diversity among the accessions was determined by using the technique of coefficient of racial likeness (CRL) following the procedure outlined by Pearson (1926):
N
CRL2ij = (1/N) Sigma[Xik - Xjk) 2Sek)^2
k = 1
where:N = the number of characters Se = the standard error associated with the variety means for the kth character Xij and Xjk = the means of the kth character for variety i and j respectively.
TABLE 1. List of Kenaf accessions evaluated -------------------------------------- Entry Entry -------------------------------------- AU-41 AU-72 AU-75 Local line 1870 S72-15-2 AU-36 GR25/63 1892-10 CD-98-13 Cuba 1805 AC-299 A60-282 Tiannung 1801 BG 58-7 Everglades 41 AU-15 Tiannung No.1 S.69-J-11 AMC-108 AU-71 J69 163 AU-24 A60-284 J-6987 S72-78-18-3 S75-1-117 3431-29 HC-583 A63-511 AU-156 Balaglades No. 7 Local line Local line 1866 3789-50 AU-151 BG 61-53 Tiannung No. 2 Guatemala - 45 AU-64 AU-25 Bomo Local Cuba 2032 Hall Malavia Local line 1869 AU-191 S72-78-18-10 25-A5M AU-159 S72-45-9 Guatemala 48 AC-313 Cuba 108 AC-1882 A63-512 -------------------------------------- The coefficient of racial likeness (CRL) is a standardised distance and should have a value of 2.0 for a pair of varieties which differ at the 5% level of significance on all the characters considered. To analyse the variation among the accessions, a principal component analysis was performed with a computer programme at the Statistical Department of the Ministry of Agriculture, Cairo, Egypt. Data on the nine characters were standardised and transformed and the programme produced eigen-values of the principal component axis. Based on the first two principal axes, average component analysis was used to cluster components 1, 2, 3 and 4 using PRINCOMP of the SAS computer programme at the Statistical Unit, International Institute of Tropical Agriculture (IITA), Ibadan, Nigeria. The projections (cultivar scores) were calculated from normalised vectors and standardised characters. Varimax rotation was used to redistribute the principal component results so that the solution approached orthogonal simple structure (Rhodes and Martin, 1972). Cluster analysis is generally employed to sort individuals into distinct groups. The technique used in taxonomy usually begins with a matrix of distances or correlations between taxon means, then the individuals are assorted into distinct groups. Detailed accounts of cluster analysis are given by Sokal and Sneath (1963).
Since CRL2 is the square of the standardised distance between two genotypes (similar to the D2 described by Trocher (Rao, 1952), the relative contribution of each character to the total CRL2 value between each pair of genotypes was determined following the procedure outlined by Bhatt (1970).
RESULTS AND DISCUSSION
The PCA yielded eigen-values of each principal component axes of ordination of accessions with the first nine axis accounting for 100% variation among the accessions while two of these (with eigen-values greater than 2.0) accounted for 54.4% (Table 2). The first three principal axes accounted for 67.2% of the total variation among the nine characters describing 54 accessions. Based on the first two principal axes, a two-dimensional ordination of the accessions was drawn (Fig. 1). Grouping the accessions such that within group distance is less than between groups distance, six groups/clusters were identified and described (Table 3). Group I has one accession, group II has four accessions while groups III to VI have 16, 17, 10 and 6 accessions, respectively. Group I is comparable to other groups in most parameters except in butt diameter, retting percentage and fibre yield (Table 2). Group VI is superior to other groups only in respect of number of leaves per plant, number of nodes per plant and core weight. Majority of the accessions fell into group III or IV. The distinguishing features of these two groups are number of leaves per plant, number of nodes per plant, core weight and fibre yield (Table 2). The CRL distance for the 1431 possible pairs of accessions ranged from 0.047 (for GR25/63 and Local 1866) to 1.325 (for A44.2 and GR25/63). The 66 CRL distances for all possible combinations in 12 out the 54 accessions (which include the most outstanding varieties Cuba 108, Tiannung No. 1, Guatemala 45 and 48) are presented in Table 4. The values ranged from 0.17 (for Cuba 108 and Tiannung No.1) to 1.14 (for Cuba 108 and Guatemala 45). Even though all the values are less than 2.0 and some authors including Ariyo (1987, 1990) suggest that appreciable progress cannot be expected from hybridisation between the accessions, alternative explanations can be advanced for low CRL values. This could be due to soil heterogeneity. Furthermore, phenotypic similarity of accessions for the traits measured does not necessarily impy genotypic similarity. For quantitative characters controlled by many genes, two accessions could have similar phenotypes but be very distinct genetically. Some of these accessions were introduced from Guatemala, Cuba, and United States of America many decases ago. It appears unusual, therefore, that cultivars from such diverse origin could be genetically similar. Table 2. Eigen-values and percent of total variation accounted for by the first three principal component axes of the ordination of kenaf varieties ------------------------------------------------------------------------- Principal Component Eigen-value % of total variation Cummulative axis accounted for percent ------------------------------------------------------------------------- I 2.79 31.01 31.01 II 2.12 23.53 54.54 III 1.14 12.63 67.17 ------------------------------------------------------------------------- Table 3. Characteristic patterns of six clusters of kenaf accessions showing mean values with standard errors ------------------------------------------------------------------------- Clusters I II III No. of accessions 1 4 16 -------------------------------------------------------------------------- Traits 1. Plant height (m) 2.06 1.99+/-0.03 1.94+/-0.01 2. Butt diameter (cm) 1.79 1.14+/-0.03 1.15+/-0.02 3. Leaves/plant 65.48 56.65+/-3.19 61.35+/-1.22 4. No. of nodes 40.93 34.78+/-0.38 36.95+/-0.50 5. Fresh plant weight (kg) 2.16 1.66+/-0.02 1.71+/-0.05 6. Core weight (g) 703.11 606.81+/-41.36 675.96+/-36.86 7. Core % 15.36 19.32+/-1.89 18.00+/-0.63 8. Retting 7.47 6.81+/-0.63 6.17+/-0.42 9. Fibre yield/plant (g) 109.33 87.28+/-7.58 80.63+/-2.19 ------------------------------------------------------------------------- Clusters IV V VI No. of accessions 17 10 6 -------------------------------------------------------------------------- Traits 1. Plant height (m) 2.02+/-0.02 2.11+/-0.02 20.1+/-0.04 2. Butt diameter (cm) 1.18+/-0.07 1.19+/-0.04 1.18+/-0.04 3. Leaves/plant 73.18+/-1.85 72.84+/-2.11 80.84+/-2.23 4. No. of nodes 42.28+/-0.80 41.09+/-1.13 46.20+/-1.20 5. Fresh plant weight (kg) 1.97+/-0.07 2.35+/-0.07 2.20+/-0.18 6. Core weight (g) 720.18+/-29.61 629.18+/-52..69 793.71+/-59.02 7. Core % 15.53+/-1.04 10.51+/-0.38 15.43+/-1.56 8. Retting 5.80+/-0.33 4.82+/-0.30 4.87+/-0.40 9. Fibre yield/plant (g) 94.17+/-3.66 88.09+/-4.28 84.03+/-3.44 --------------------------------------------------------------------------- TABLE 4. Coefficient of racial of likeness (CRL) distances between pairs of twelve kenaf accessions --------------------------------------------------------------------------- Genotype Cuba 108 Local 1870 S69-J-I Guat 48 AU-75 AU-72 --------------------------------------------------------------------------- Tian-1 0.45 0.35 0.17 0.89 0.76 0.46 Cuba 108 - 0.35 0.45 0.83 0.76 0.54 Local 1870 - 0.37 0.80 0.67 0.58 S69-J-I - 0.84 0.73 0.51 Guat 48 - 0.96 0.82 AU-75 - 0.46 AU-72 - J6787 HC583 AC299 AU151886 Guat 45 --------------------------------------------------------- Genotype J6987 AC299 AU151886 Guat45 HC583 --------------------------------------------------------- Tian-1 0.62 1.04 0.88 0.73 1.04 Cuba 108 0.72 1.11 0.91 0.82 1.14 Local 1870 0.67 1.01 0.83 0.72 1.04 S69-J-I 0.63 1.00 0.85 0.71 1.00 Guat 48 0.81 0.97 0.78 0.78 0.84 AU-75 0.46 0.66 0.49 0.55 0.73 AU-72 0.24 0.78 0.64 0.47 0.78 J6987 - 0.62 0.44 0.34 0.60 HC583 - 0.56 0.37 0.41 AC299 - 0.46 0.50 AU151886 - 0.48 Guat 45 - --------------------------------------------------------- The clustering pattern of the Kenaf accessions did not indicate any relationship between genetic diversity and eco-geographical distribution. Cluster II contains entries from Nigeria, Cuba and the USA while Cluster VI contains introductions from Australia, Nigeria and Guatemala (Fig. 1). Dudley and Davies (1966), Bhatt (1970), Chedda and Fatokun (1982) and Dasgupta and Das (1984) working with cultivars of alfalfa (Medicago sativa), wheat (Triticum aestivum), okra (Abelmoschus esculentus) and black gram (Vigna mungo), respectively, also noted that cultivars clustered in different groups irrespective of their countries of origin. Ariyo (1987) also working on okra arrived at similar conclusions. In this study inter-cluster distances were less than 1.0 in all cases (Fig. 1). The clustering of the accessions was closely related to the CRL values. Guatemala 48, reasonably separated from all other accessions, has a mean CRL value of 0.84 (Table 3). Mean CRL values for accessions in cluster II range from 0.67 (for Tiannung No. 1) to 0.78 (for S69-54) with an average of 0.74. Average CRL values for accessions in cluster VI on the other hand, range from 0.45 (for HC 583) to 0.58 (for AU72) with an average of 0.51. The analysis of CRL values showed that butt diameter and fibre yield each accounted for 13.0% of the little variation detected in the accessions. Plant height and retting percentage each contributed 11.3% while core weight, number of leaves per plant, core percentage and fresh plant weight contributed 12.5, 10.3, 10.2 and 10.0%, respectively. The least contribution to observed variation (8.4%) was found in number of nodes per plant. Since butt diameter and plant height, two easily assessed parameters also ranked high in their relative contributions to observed variation, these two character will be reliable in distinguishing among kenaf varieties. Previous findings (Ogunbodede, B.A.1990; unpubl.) also affirmed that these two characters were significantly correlated with fibre yield and are therefore reliable selection criteria in kenaf improvement programmes.
ACKNOWLEDGEMENT
The author is grateful to Dr. Ahmed Abd El Halim, Director, Central Statistical Laboratory, Agricultural Research Centre, Cairo, Egypt for assistance rendered in PCA computer analysis, Dr. S. Nokoe of the Computer Unit, International Institute of Tropical Agriculture, Ibadan for assistance in cluster analysis, and to the Director, Institute of Agricultural Research and Training, Obafemi Awolowo University, Ibadan, Nigeria, for permission to publish this work.
REFERENCES
Akoroda, M.O. 1983. Principal component analysis and metroglyph of variation among Nigerian yellow yams. Euphytica 32:565- 573.
Ariyo, O. J. 1987. Multivariate analysis and choice of parents for hybridisation in okra (Abelmoschus esculentus (L.) Moench. Theoretical and Applied Genetics 74:361- 363.
Ariyo, O.J. 1990. Measurement and classification of genetic diversity in okra (Abelmoschus esculentus). Annals of Applied Biology 116:335-344.
Bhatt, G.M. 1970. Multivariate analysis approach to selection of parents for hybridization aiming at yield improvement in self-pollinated crops. Australian Journal of Agricultural Research 21:1-7.
Chedda, H. R. and Fatokun, C.A. 1982. Numerical analysis of variation patterns in okra (Abelmoschus esculentus L. Moench). Botanical Gazette Chicago 43:253-261.
Clifford, H.T. and Stephenson, W. 1975. An Introduction to Numerical Classification. Academy Press, London, New York, xii, 229pp.
Dasgupta, T. and Das, P.K. 1984. Multivariate analysis and selection of parents for hybridization in black gram. Philippine Journal of Agriculture 67:86-92.
Duddley, J.W. and Davies, R.L. 1966. Preliminary groupings of plant introductions of alfalfa (Medicago sativa L.) heterosis studies. Crop Science 6:597-601.
Falowo, D.O. 1982. An application of principal component analysis: response of F3 amazon cocoa seedlings to insecticide treatments. Nigeria Agricultural Journal 17/18:178 - 188.
Hayman, H.H. 1967. Modern Factor Analysis. 2nd edition. University of Chicago Press, Chicago.
Meyer, S.J., Hubbard, K.G. and Wilhite, D.A. 1991. The relationship of climatic indices and variables to corn (maize) yields; a principal component analysis. Agricultural and Forest Metereology 55:59-84.
Pearson, K. 1926. On the coefficient of racial likeness. Biometrika 18:105-117.
Rao, C.R. 1952. Advanced Statistical Methods in Biometrical Research. John Wiley and Sons, New York.
Rhodes, A.M. and Martin, F.W. 1972. Multivariate studies of variations in yams (Dioscorea alata L.) Journal of American Society of Horticultural Science 97:685-688.
Sokal, R.R. and Sneath, P.H.A. 1963. Principles of Numerical Taxonomy. W.H. Freeman and Co., San Francisco, 359 pp. Copyright 1997 The African Crop Science Society
The following images related to this document are available:Line drawing images[cs97019b.gif] [cs97019a.gif] |
| |||||||||
