NUMERICAL TAXONOMY OF THE GENUS SENNA MILL. FROM BANGLADESH

This study examines the patterns of morphological variation and phenetic relationships among 11 species of Senna Mill. from Bangladesh using 32 vegetative and floral characters. The highest similarity is found between S. obtusifolia and S. tora, while the highest variation is observed between S. alata and S. hirsuta. UPGMA tree derived from cluster analysis reveals three major clusters, the first of which consists of two species (S. alata and S. auriculata), the second cluster comprises four species (S. hirsuta, S. obtusifolia, S. tora and S. occidentalis) and the third one is composed of five species (S. multiglandulosa, S. sophera, S. siamea, S. timoriensis and S. surattensis). A close relationship is also found between S. multiglandulosa and S. sophera, and between S. siamea and S. timoriensis. Results obtained from the present study are found congruent with cytological and anatomical studies showing the significance of numerical analysis for taxonomic relationship in the genus Senna. Introduction Numerical taxonomy, also termed as morphometrics deals with grouping by numerical methods of taxonomic units into taxa on the basis of their character state (Sneath and Sokal, 1973). Cluster analysis and principal component analysis are two techniques commonly used in numerical classification. Cluster analysis produces a hierarchical classification of entities (taxa) based on the similarity matrix. It thus provides a logical means of expressing the relationship existing between taxa. Numerical taxonomic studies are important for discovering and documenting new morphological character and character states, and many attempts have been made in this regard for understanding phenetic relationships in different groups of plants (Pinheiro and de Barros, 2007; Mulumba and Kakudidi, 2010; Deshmukh, 2011; Rahman and Rahman, 2012). The genus Senna Mill. (Caesalpiniaceae) is represented by 350 species and is distributed throughout the world (Marazzi et al., 2006). Irwin and Barneby (1982) reports that about 80% of the Senna species are found in the American continent, while most of the remaining members occur in tropical Africa, Madagascar and Australia, with only a few species in southeastern Asia and the Pacific Island. Senna are characterized by the presence of cylindrical or flattened, irregularly dehiscent pods and longest filaments without sigmoidally curved towards the base and seed surfaces usually with areole. Economically Senna species are very important since their bark and oil extract are used for flavouring purposes and in soaps, candy and perfumery (Hill, 1952). Several Senna species are reported to have medicinal properties as laxative, expectorant, antimalarial, relaxant and anti-inflammatory (Sadique and Chandra, 1987; Ajagbonna and Mojiminiyi, 2001; Tona and Mesia, 2001). Studies on the genus Senna are very much limited in Bangladesh. Baker (1879) described 18 species of Cassia s.l. of which 6 species now included in Senna are found in the area of Bangladesh. Prain (1903) listed 7 species of Senna from the then Bengal which falls under the Corresponding author. Email: dr_oliur@yahoo.com


Introduction
Numerical taxonomy, also termed as morphometrics deals with grouping by numerical methods of taxonomic units into taxa on the basis of their character state (Sneath and Sokal, 1973).Cluster analysis and principal component analysis are two techniques commonly used in numerical classification.Cluster analysis produces a hierarchical classification of entities (taxa) based on the similarity matrix.It thus provides a logical means of expressing the relationship existing between taxa.Numerical taxonomic studies are important for discovering and documenting new morphological character and character states, and many attempts have been made in this regard for understanding phenetic relationships in different groups of plants (Pinheiro and de Barros, 2007;Mulumba and Kakudidi, 2010;Deshmukh, 2011;Rahman and Rahman, 2012).
The genus Senna Mill.(Caesalpiniaceae) is represented by 350 species and is distributed throughout the world (Marazzi et al., 2006).Irwin and Barneby (1982) reports that about 80% of the Senna species are found in the American continent, while most of the remaining members occur in tropical Africa, Madagascar and Australia, with only a few species in southeastern Asia and the Pacific Island.Senna are characterized by the presence of cylindrical or flattened, irregularly dehiscent pods and longest filaments without sigmoidally curved towards the base and seed surfaces usually with areole.Economically Senna species are very important since their bark and oil extract are used for flavouring purposes and in soaps, candy and perfumery (Hill, 1952).Several Senna species are reported to have medicinal properties as laxative, expectorant, antimalarial, relaxant and anti-inflammatory (Sadique and Chandra, 1987;Ajagbonna and Mojiminiyi, 2001;Tona and Mesia, 2001).
Studies on the genus Senna are very much limited in Bangladesh.Baker (1879) described 18 species of Cassia s.l. of which 6 species now included in Senna are found in the area of Bangladesh.Prain (1903) listed 7 species of Senna from the then Bengal which falls under the territory of present Bangladesh.Although Khan et al. (1996) documented 6 species of the Senna from Bangladesh, recently Ahmed et al. (2008) reported 10 species of the genus from the country.The extensive field surveys through the present study revealed a total of 11 species of Senna are now found in Bangladesh.Despite few fragmentary studies are available on Senna, numerical approaches have never been tested in this genus to determine species relationships.Therefore, the present study aims at applying numerical method for examining morphological variation and inferring phenetic relationships among Senna species occurring in Bangladesh.

Plant materials:
Eleven species of Senna were used in the present study (Table 1).Both fresh materials collected from different areas of Bangladesh, and herbarium specimens housed at Dhaka University Salar Khan Herbarium (DUSH) and Bangladesh National Herbarium (DACB) were examined for numerical analysis.

Characters:
Thirty two characters were investigated and used in this analysis.Characters and character states were determined through examination of both living and herbarium specimens housed at DUSH and DACB.Both qualitative and quantitative characters were coded as binary-state.The characters and their binary states used for numerical taxonomic studies are listed in Table 2. Neither herbarium nor living specimens of Senna multiglandulosa were available; therefore character states for this species were determined from the relevant literature (Ahmed et al., 2008).

Statistical analysis:
The data matrix was scored using binary matrix.Dissimilarity matrix was prepared based on the data matrix.Cluster analysis was performed using UPGMA (unweighted pair group method with arithmetic mean) and a dendrogram was constructed to show the relationship among the species (Sneath and Sokal, 1973).All analyses were carried out using the program STATISTICA (Version 3.0).

Result and Discussion
Thirty two vegetative and reproductive characters have been identified for numerical analysis of Senna species (Table 2).The present study reveals that the lowest morphological variation is observed between S. obtusifolia and S. tora indicating that they are most closely related among all species studied.S. occidentalis is also very close to S. obtusifolia.The highest variation is found between S. alata and S. hirsuta (Table 3).A high variation has also been detected between S. alata and S. tora; and S. alata and S. obtusifolia.The numerical analysis presents the phenetic relationships among 11 Senna species.The UPGMA dendrogram based on cluster analysis reveals three clusters.The first cluster consists of two species, viz.S. alata and S. auriculata; the second one comprises four species, namely S. hirsuta, S. obtusifolia, S. tora and S. occidentalis; while the third cluster includes five species, viz., S. multiglandulosa, S. sophera, S. siamea, S. timoriensis and S. surattensis (Fig. 1).
In the first cluster S. alata is grouped with S. auriculata indicating that they are closely allied, and this is evidenced by the presence of their puberulent stem, persistent stipule and linear to oblong pod.A close association between S. hirsuta, S. obtusifolia, S. tora and S. occidentalis is evident in the second cluster.The common characters shared by these four species include linear cytological investigation where the somatic chromosome number 2n=28 was reported for these four species (Irwin and Turner, 1960;Bir and Kumari, 1980).In conclusion, our study shows the significance of numerical analysis for detecting variation and taxonomic relationships among Senna species available in Bangladesh as it is attested by previous studies based on cytological and anatomical characters.