INTRODUCTION The turmeric of commerce is dried rhizome of Curcuma longa L. (syn. Curcuma domestica Val.) belonging to the family Zingiberaceae and traces its origin to tropical rain forests of South East Asia. Turmeric, the Indian saffron, is mainly valued for its colouring constituent, ‘curcumin’, which is indispensable in food industry, confectionery, pharmaceuticals and cosmetics (Zachariah and Babu, 1992). In recognition of the importance of turmeric, emphasis on evolving new varieties with high yield and quality has been in the forefront of research. Germplasm characterization is an important link between conservation and utilization of plant genetic resources. For any breeding programme, genetic diversity is raw material to the breeder since genetic variation determines the potential for selection and is useful in resolving phylogenetic relationships. However, existence of multiple local names and lack of authentic identity of materials have narrowed genetic resource base. Conventional taxonomic techniques in conjunction with molecular biology tools may help resolve taxonomic confusion prevailing in the genus. Though a few studies on morphological and anatomical characterization of Curcuma species have been attempted, not much has been Multivariate based marker analysis in turmeric (Curcuma longa L.) K. R. Vijayalatha and N. Chezhiyan Horticultural College and Research Institute Tamil Nadu Agricultural University, Coimbatore –641003, India. E-mail: viji_k_r@yahoo.com ABSTRACT Genetic diversity of 30 accessions of turmeric was assessed at the molecular level and compared to morphological traits for degree of divergence. The pattern of clustering of quantitative data based on D2, K means and UPGMA revealed discrepancy among them. The cluster profile, based on quantitative data and RAPD markers exhibited considerable levels of congruence between them. Accessions studied for degree of divergence by RAPD profiles revealed 68.50% polymorphism for 21 primers. The highest number of fragments (10) was obtained with primer OPG 19 while OPC 18 was completely monomorphic. Primers OPB 08, OPC 20, OPE 09 and OPG 19 detected a high level of polymorphism (> 90%). Discrepancy observed at both morphological and molecular levels in the accessions emphasizes the need for specific morphological and molecular markers for discriminating these accessions. Key words : Genetic divergence, turmeric, marker, polymorphism, primer done on molecular characterization (Sasikumar, 2005). Recently, use of molecular markers has assumed great significance in germplasm characterization and assessment of genetic diversity. Universal acceptance of PCR- based markers has accelerated the use of molecular markers, more specifically DNA based markers, that are valid in assessing genetic diversity as these are genetically stable and detectable during all stages. Hence, an effort was made to understand quantitative relationship and genetic relation using multivariate statistical tools and RAPD markers. MATERIAL AND METHODS Two hundred and twenty three accessions of turmeric drawn from different states of India are maintained in the germplasm pool at Department of Spices and Plantation Crops, Horticultural College and Research Institute, Tamil Nadu Agricultural University, Coimbatore and were used as the experimental material. Plant material : The number of accessions in the germplasm was too high to assess the divergence at a molecular level. The D2 statistic was found to be a useful tool in grouping the accessions phenotypically. Therefore, all the 223 accessions were subjected to Mahalonobis D2 statistic to cull out variable parameters. The accessions clustered into two groups, exhibiting a wide variability for eight parameters of J. Hortl. Sci. Vol. 3 (2): 107-111, 2008 108 J. Hortl. Sci. Vol. 3 (2): 107-111, 2008 importance, among the 22 quantitative traits evaluated. Further D2 statistic was performed based on eight parameters, to group the accessions into five clusters. Thirty accessions were selected to represent each cluster, based on growth and yield performance. The accessions were raised in randomized block design and replicated twice. Biometrical observations were recorded on five randomly tagged plants. Details of accessions are furnished in Table 1. DNA extraction Molecular profiling of selected accessions was done using RAPD marker. DNA from all the 30 accessions was extracted using the protocol described by McCouch et al (1988) from leaves frozen at –20°C. RAPD analysis Genomic DNA from 30 accessions was amplified using a set of 20 arbitrary oligonucleotide decamer primers (Operon Technologies, Alameda, Calif., USA) (Table 1). Amplification reactions were in volumes of 20 ml containing 10 mm Tris HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl 2 , 0.001 per cent gelatin, dNTPs each at 0.1 mM, 0.2 mM primer, 25-30 ng of genomic DNA and 0.5 unit of Taq DNA polymerase (Bangalore Genei Pvt. Ltd., Bangalore). Amplifications were performed in 0.2 ml thin-walled PCR tubes in a PTC 100 thermal cycler (MJ Research Inc.) programmed for 35 cycles. After initial denaturation for 2 min at 92°C, each cycle consisted of 1 min at 92°C, 1 min at 36°C and 2 min at 72°C. These 35 cycles were followed by 7 min. final extension at 72°C. PCR amplified products (15 ml) were subjected to electrophoresis in 1.5% agarose gel in 1 x TBE buffer at 120 V for 3.5 h using Apelex electrophoresis unit. Statistical analysis Amplified DNA fragments detected upon electrophoretic separation in each genotype were scored for presence (1) or absence (0) of clear and unambiguous bands. A data matrix comprising ‘1’ and ‘0’ was formed and this data matrix was subjected to further analysis. Two different sets of data gathered (quantitative and RAPD marker) were subjected to cluster analysis. The binary data generated from RAPD marker data set was subjected to Sequential Agglomerative Hierarchical Non-Overlapping (SAHN) clustering. UPGMA (Unweighted Pair Group Method with Arithmetic Average) dendrogram was constructed based on Jaccard’s similarity coefficient matrix. The quantitative data were also subjected to cluster analysis using D2 statistic (Mahalonobis, 1936) and K means (Sneath and Sokal, 1973) based on Euclidean distance and UPGMA method based on squared Euclidean distance. RESULTS AND DISCUSSION The D2 statistic identified eight variable parameters and clustered them into five groups. Thirty accessions were selected from the five clusters (Table 1). On ranking the characters, relative contribution of yield (65.65%), followed by girth of secondary rhizomes (10.46%), weight of primary and secondary rhizomes (8.57% and 7.77% respectively) and days to maturity (6.63%) were the highest contributors to total divergence, while plant height, number of leaves and number of tillers contributed the least to divergence. Hence, seventeen accessions of cluster 1, ten accessions of cluster 5 and single genotype from cluster 2, 3 and 4 were shortlisted as the experimental material and were compared at the morphological and molecular level (Table 2). Table 1. Accessions selected by D2 statistic S. No. Accession number Accession 1 2 BS - 2 2 15 BS - 16 3 22 BS -23 4 47 BS - 48 5 52 BS - 53 6 75 BS - 76 7 88 BS - 89 8 89 BS - 90 9 114 BS - 115 10 120 Kanthi 11 121 Shoba 12 122 VK 5 13 130 Sudarshana 14 131 Suguna 15 132 Suvarna 16 133 Roma 17 134 Kalimpong 18 146 Alleppey 19 148 PTS 12 20 151 Prabha 21 152 Prathibha 22 156 PTS 2 23 169 Erode local 24 175 Erode local 25 184 Erode local 26 189 CO 1 27 194 JTS 2 28 195 Rajendrasonia 29 198 PTS 55 30 209 Erode local Vijayalatha and Chezhiyan 109 Multivariate analysis The accessions were subjected to different statistical tools : D2 statistic, K means and UPGMA, based on 22 quantitative traits. Using D2 statistic The accessions were grouped into five clusters based on D2 statistic for 22 quantitative traits. The clustering pattern revealed that except for BS 2 and BS 76, accessions from Bhavanisagar clustered together. Among accessions from Erode, ER 169 was found to be distinct. Accessions having the same geographical origin, viz., Sudarshana and Suguna from Andhra Pradesh, VK 5, Prabha and Prathibha from Kerala, clustered together, while, PTS 55 was distinct among the accessions from Orissa (Table 3). Using K means A cluster tree was constructed with 22 quantitative traits using K means clustering, based on Euclidean distance. The K means grouped the accessions into five clusters (Table 3). The pattern of clustering revealed that the accessions from Bhavanisagar clustered together, except for BS 76, BS 89 and BS 115. Among the accessions from Erode, ER 209 and ER 175 clustered together, while, ER 184 clustered with CO 1. Accessions from the same geographical origin, Sudarshana and Suguna, VK 5, Prabha and Shoba were similar, while, Prathibha and Suvarna were distinct from other Kerala accessions. Among accessions from Orissa, PTS 2 was distinct from the others and clustered with Bhavanisagar genoytpes. The genotype Rajendrasonia was distinct from rest of the accessions. Using UPGMA A dendrogram was constructed with 22 quantitative traits using UPGMA, on the basis of squared Euclidean distance of standardized data. The pattern of clustering indicated two distinct groups among accessions. All the accessions, except Kalimpong, formed a single cluster. The remaining 29 accessions formed two subgroups, one with 17 accessions and the other with 12 accessions. Table 3. Cluster composition of turmeric genotypes based on D2 and K means cluster analysis Cluster No. Cluster members based on D2 statistic Cluster members based on K means 1 BS 2, BS 76, Kalimpong BS 2, BS 16, BS 23, BS 53, BS 89, BS 90, Prathibha, PTS 2, JTS 2, ER 209 2 CO1, JTS 2 BS 76, Shoba, VK 5, Prabha, ER 175, ER 184, CO1, Salem local 3 Bs 16, BS 23, BS 53, BS 89, BS 90, BS 115, Sudarshana, Suguna, Rajendrasonia Kanthi, Shoba, Roma, Salem local 4 VK 5, Sudarshana, Suguna, Suvarna, Alleppey, BS 115, Kanthi, Allepey PTS 12, Prabha, Prathibha, PTS 2, ER 169 5 ER 175, ER 184, Rajendrasonia, PTS 55, ER 209 Suvarna, Roma, Kalimpong, PTS 12, ER 169, PTS 55 Table 2. Cluster mean values of eight characters for turmeric accessions Cluster / character 1 2 3 4 5 % contribution Number of accessions 143 2 2 2 74 Plant height (cm) 37.36 34.07 32.61 32.46 34.06 0.08 Number of leaves 15.03 12.04 11.38 11.75 12.84 0.22 Number of tillers 3.68 1.67 1.75 1.71 1.74 0.60 Days to maturity 248.71 233.25 242.75 251.50 235.38 0.63 Weight of primary rhizomes (cm) 100.89 92.50 127.08 87.09 106.51 8.57 Weight of secondary rhizomes (cm) 67.44 71.67 75.00 61.25 71.38 7.77 Girth of secondary rhizomes (cm) 5.11 4.42 4.46 4.21 4.19 10.46 Yield (kg/plot) 3.77 2.96 2.78 2.70 3.65 65.65 Fig 1. Dendrogram of thirty accessions using UPGMA based on squared Euclidean distances Molecular markers in turmeric J. Hortl. Sci. Vol. 3 (2): 107-111, 2008 110 Accessions from Bhavanisagar and Erode locations exhibited similarity except for BS 90 and ER 209, which were distinct. Among the accessions, BS 76 and BS 115 from Bhavanisagar and PTS 12 and PTS 55 from Orissa, exhibited higher level of homogenity between them. Accessions Sudarshana and Suguna from Andhra Pradesh, and, Prabha and Prathibha from Kerala were very distinct though these were from the same geographical area (Fig 1). However, accessions from diverse geographical origin like PTS 2 and Roma (Orissa), and, VK 5 and Suvarna (Kerala) got clustered together with accessions from Bhavanisagar (Tamil Nadu). RAPD polymorphism RAPD profiles for the thirty accessions were generated to make genetic diversity analysis. Of the seventy- decamer primers used for RAPD analysis, 21 primers yielded scorable, unambiguous markers while 14 primers failed to amplify. PCR amplification of template DNA produced a total of 89 markers of which 61 markers were found to be polymorphic (68.5%) and the rest were monomorphic (Table 4). The number of markers produced per primer varied from two (OPB 11, OPB 14, OPC 01, OPC 16, OPC 18, OPE 03 and OPE 04) to ten (OPG 19). A higher number of polymorphic markers (10) were obtained with primer OPG 19 while primer OPC 18 produced monomorphic markers. Based on the level of polymorphism detected by individual primers, four primers (OPB 08, OPC 20, OPE 09 and OPG 19) revealed over 90% polymorphism (Table 5). Cluster analysis was performed based on Jaccard’s similarity coefficient and a dendrogram was constructed (Fig 2) involving all the accessions. The dendrogram based on RAPD profiles reflected considerable level of genetic considering the geographical distribution. Accessions were classified into two major groups. The first group was further sub-grouped into three (1A, 1B, 1C) clusters. This group consisted of accessions from Bhavanisagar, viz., BS 16, BS 23, BS 53, BS 89, BS 90 and BS 115, with higher level of similarity among those from the same geographical origin (Shamina et al, 1198). Other accessions grouped along with accessions of Bhavanisagar include PTS 12, Prabha, Roma, JTS 2 and Rajendrasonia. Accessions ER 175 and ER 184 were grouped with CO1. Some accessions having the same geographical origin and higher level of similarity (based on quantitative data) were found in different clusters. Accessions Prabha, Table 4. Level of polymorphic loci detected by RAPD analysis Number of primers 21 Total number of markers produced 89 Range of markers 2 – 10 Average number of markers 4 Number of monomorphic markers 27 Number of polymorphic markers 61 Per cent polymorphism 68.50 Fig 2. Dendrogram of thirty accessions using UPGMA based on Jaccard similarity coefficient for RAPD markers Table 5. Details of random primers used for RAPD analysis Random 5 ’ to 3’ GC Total Poly- Poly- Primer sequence content markers morphic morphism (%) markers ( % ) OPB - 08 GTCCACACGG 70 4 4 100.00 OPB - 09 TGGGGGACTC 70 7 6 85.71 OPB - 11 GTAGACCCGT 80 2 1 50.00 OPB - 14 TCGGCTCTGG 70 2 1 50.00 OPB - 15 GGAGGGTGTT 60 5 3 60.00 OPC - 01 TTCGAGCCAG 70 2 1 50.00 OPC - 16 CACACTCCAG 60 2 1 50.00 OPC - 18 TGAGTGGGTG 60 2 0 0.00 OPC - 20 ACTTCGCCAC 60 3 3 100.00 OPE - 03 CCAGATGCAC 60 2 1 50.00 OPE - 09 CTTCACCCGA 60 4 4 100.00 OPF - 03 CCTGATCACC 60 5 3 60.00 OPF - 04 GGTGATCAGG 60 2 1 50.00 OPF - 09 CCAAGCTTCC 60 9 8 88.89 OPF - 13 GGCTGCAGAA 60 9 5 55.55 OPG - 10 AGGGCCGTCT 70 6 2 33.33 OPG - 13 CTCTCCGCCA 70 3 1 33.33 OPG - 14 GGATGAGACC 60X 3 2 66.67 OPG - 17 ACGACCGACA 60 4 3 75.00 OPG - 19 GTCAGGGCAA 60 10 9 90.00 OPG - 20 TCTCCCTCAG 60 3 2 66.67 Total 89 61 68.50 Average marker / primer 4.23 2.90 Vijayalatha and Chezhiyan J. Hortl. Sci. Vol. 3 (2): 107-111, 2008 111 Pathibha, VK 5 and Kanthi from Kerala were found to be in different clusters. However, accessions of diverse geographical origin such as PTS 12 (Orissa), Prabha (Kerala), Suvarna (Andhra Pradesh) and VK 5 (also from Kerala) formed separate clusters. Genetic diversity assessed using 22 quantitative traits among the 30 accessions was consistent with not all the methods used. The accessions exhibited considerable level of congruence between them. The probable reason could be variation inherent in the nature of algorithms involved. The inconsistency may be due to efficiency of the algorithm to eliminate bias in the quantitative data, which is invariably under the influence of environment. Discrepancy among the three tools studied shows that morphological traits alone are not sufficient for discriminating the accessions, as, turmeric lacks defined morphological descriptors, coupled with cultivar-specific characters. This meager morphological diversity exhibited among accessions may be due to their related pedigree, which make discrimination of accessions rather cumbersome. This also reinforces the belief that accessions collected from the same geographical area do not have any impact on genetic diversity. In comparing clustering pattern of quantitative traits with RAPD markers, genetic dissimilarity was noticed. However, clustering of Bhavanisagar accessions showed that Sudarshana and Suguna, ER 175 and ER 184 were similar both at the morphological and molecular level. Credibility of genetic diversity analysis involving 22 quantitative traits and 89 RAPD markers could not be established beyond doubt, since; pedigree details of the accessions involved were not available. Though the accessions are known for their geographical origin, the latter could not be used as a criterion to establish relationship between accessions, since, accessions from the same geographical origin had clustered separately. Molecular profiling of Curcuma accessions had some similarity with morphological characterization, though, there were incongruities of the accessions of unidentical morphology falling in the same group, or vice versa (Syamkumar and Sasikumar, 2007). Although RAPD markers are widely used for genetic analysis, the inherent problems in the RAPD technique do not make these markers a reliable system for reproduction of results. However, the technical simplicity and high throughput of this technique (Williams et al, 1990) helps to engage RAPD markers for large-scale survey. Marker systems such as SSR may help establish differences among accessions with greater accuracy compared to RAPD markers. Further, use of molecular markers should be combined with reliable morphological descriptors of qualitative nature and pedigree details of the genotype, to strengthen the diversity analysis and, in turn, aid germplasm management. ACKNOWLEDGEMENT The authors thank Dr. M. Maheswaran, Associate Professor, CPMB, TNAU, Coimbatore, for extending facilities to carry out the research and Mr. R. Venkatachalam, Assistant Professor (Hort), for interpretation of results. REFERENCES Mahalonobis, P.C. 1936. On the generalise distance in statistics. Proc. Natl. Instt. Sci., 2:49-55 McCouch, S.R., Kochert, G. Yu, Z.H. Wang Z.Y., Khush, G.S.. Coffman W.R and Tanksley. S.D. 1988. Molecular mapping of rice chromosomes. Theor. Appl. Genet., 76:815-829 Sasikumar, B. 2005. Genetic resources of Curcuma: diversity, characterization and utilization. Plant Gen. Res. C&U 3: 230-251 Shamina, A., Zachariah, T.J., Sasikumar, B and George, J.K. 1998. Biochemical variation in turmeric (Curcuma longa L.) accessions based on isozyme polymorhism. J. Hortl. Sci. and Biotech., 73:479-483 Sneath, P.H.A. and Sokal, R.R. 1973. Principles of Numerical Taxonomy. San Francisco, W.H. Freeman Syamkumar, S and Sasikumar. B. 2007. Molecular marker based genetic diversity analysis of Curcuma species from India. Sci. Horti., 112:235-241 Williams, J.G.K., Kubelik, A.R., Livek, K.J., Rafalski, J and Tingey. S.V. 1990. DNA polymorphisms amplified by arbitary primers are useful as genetic markers. Nucl. Acid Res., 18:6531-6535 Zachariah, T.J. and Babu. K.N. 1992. Effect of storage of fresh turmeric, oleoresin and curcumin content. J. Spices and Aromatic Crops, 1:55-58 (MS Received 15 October 2007, Revised 16 June 2008) Molecular markers in turmeric J. Hortl. Sci. Vol. 3 (2): 107-111, 2008