RESEARCH ARTICLE Geospatial modelling techniques for rapid assessment of phytodiversity at landscape level in western Himalayas, Himachal Pradesh M. B. Chandrashekhar*, Sarnam Singh and P. S. Roy Indian Institute of Remote Sensing, 4, Kalidas Road, Dehradun 248 001, India The mountainous state of Himachal Pradesh is known for its vast natural wealth, including forests, alpine meadows, rivers and valleys endowed with a rich array of life forms. However, this biodiversity hot spot is under great peril owing to human-induced disturbance factors. The findings on spatial representation of the habitat at a scale which can be used for conservation planning and rehabilitation are lacking. In this study IRS-1D, LISS-III sensor data have been used to assess the vegetation coupling with RS and GIS techniques in the state. This communication presents an approach for rapid assessment of biodiversity at landscape level using satellite remote sensing, phytosociological data and knowledge base in geospatial model. The geospatial analyses at the landscape level reveal that most of the fertile valleys of the region are occupied by the human activities and possess very low bio-richness. Inverse relationship has been observed with disturbance at landscape level vis-à-vis phytodiversity or richness of the vegetation type/habitat. THERE is increasing awareness that biodiversity is not only intimately connected with long-term health and vigour of the biosphere, but is also a regulator of ecosystem functioning. The conventional species-level approach for biodiversity management has major limitations. The understanding of the priorities of biodiversity conservation and management has resulted in a policy shift from conservation of single species to habitats through interactive network of species at landscape level1,2. Microclimatic conditions are manifested in the form of four micro-endemic centres among 26 in India3. The Western Himalayan region has 12 out of the 71 genera endemic to the Himalaya. The Himalaya has more than five microendemic centres. The Himalayan ranges are among the youngest hills in the world, are active as well as fragile, and are facing threat from mankind in the name of development. Because of increased anthropogenic activities, as a result of population explosion and change in land use practices, the natural landscape has been modified resulting in fragmentation of forests with poor species composition4,5. The resulting landscape mosaic is a mixture of *For correspondence. (e-mail: chandrumb@iirs.gov.in) CURRENT SCIENCE, VOL. 84, NO. 5, 10 MARCH 2003 natural and human-managed patches that vary in size, shape and arrangement. Fragmentation of ecological units has been well-documented at landscape level using patch size, shape, abundance and forest matrix characteristics6–10. Ecosystem degradation and patch characteristics are associated with degree of spatial fragmentation11–13. Changes in landscape patterns through fragmentation or aggregation of natural habitats can alter patterns of abundance for single species and entire communities14. Understanding landscape spatial pattern is important since it contains all levels of biological hierarchy, from ecosystems to species and genes, which are targetted for biodiversity conservation. Fine resolution remote sensing technology is being widely used the world over for quick assessment of earth resources. Being cost-effective and repetitive in nature with synoptic coverage technology, it has endless application potentials. Coarse resolution sensor data (WiFS) were used by various workers to assess the vegetation, but such data have limitations to discriminate the vegetation types at a finer level15,16. In this communication, vegetation-type map derived from IRS-1D, LISS-III fine resolution sensor data has been analysed for discriminating various vegetation types of the region. Vegetationtype map is the prime input for landscape ecological analysis of the forest ecosystem. Geographic Information System (GIS) is used to derive landscape indices such as fragmentation, porosity, patchiness, patch density, interspersion and juxtaposition, which depict landscape characteristics. Himachal Pradesh (HP) in the western Himalayan range extends from the perpetual snow-covered mountains separating it from Jammu and Kashmir, and China in the north to Punjab Shiwalik ranges in the southwest and Uttaranchal in the southeast. Its hilly terrain system known for natural wealth, forests, meadows, rivers and steep valleys is enriched with rich cultural heritage. The majestic array of perpetual lofty snow peaks presents a breathtaking panoramic view. The state is known for its forests and their floral and faunal diversity. Among the 45,000 species of plants found in the country as many as 3295 species (7.3%) are reported in the state17. In HP, there are 30 wildlife sanctuaries, two national parks and three game reserves covering an area of 5940 km2 (ref. 17). 663 RESEARCH ARTICLE Study area HP has a mountainous landscape with altitudes ranging from 350 to 6975 m above mean sea level. It is located between lat 30°22′ 40″ N–33°12′ 20″ N and long 75° 45′ 55″ E–79° 04′ 20″ E (Figure 1). It has a deeply dissected topography, complex geological structure and a rich temperate flora in the high altitudes. Physiographically, the state is divided into five zones, viz. (i) wet sub-temperate zone, (ii) humid sub-temperate zone, (iii) dry temperatealpine high lands, (iv) humid subtropical zone, and (v) sub-humid subtropical zone. The average annual rainfall in the state is about 160 cm. The climate varies from warm and humid in the valley areas to freezing cold in the higher altitudes. tats and their spatial association. This extent of vegetation types is largely controlled by physiography, altitude and biogeography. The spatial organization of patches has been evaluated using landscape ecological principles22. Important landscape parameters used for landscape analysis are fragmentation, porosity, juxtaposition, patchiness and interspersion. Proximity of the habitat to a road or village determines human impact in the region. Required spatial information was integrated to determine disturbance regime. Spatial representation of biodiversity characterization at landscape level has been made by integrating important key attributes derived from the vegetation type-map, terrain complexity, disturbance regimes and phytosociological data. The methodology (geospatial model) used for biodiversity assessment at landscape level has been explained in Figure 2 (ref. 23). Material Vegetation cover-type mapping Digital data of ten IRS 1D, LISS-III satellite scenes have been used for land-cover and land-use classification of the area. Survey of India (SOI) maps on 1 : 250,000 scale have been suitably referred. Relevant literature on flora has also been consulted18–20. Analyses were carried out in Octane machine (IRIX OS) using ARC/INFO, ERDAS Imagine, Bio-CAP customized package21 (Figure 2). Method IRS-1D, LISS-III satellite remote sensing data were used for deriving vegetation cover-type map. The vegetation types thus derived were converted to represent the habi- Figure 1. 664 IRS-1D LISS-III data (March and November 2001) were used to prepare vegetation cover-type map. A total of ten scenes cover the whole state. Each scene was rectified with respect to 1 : 50,000 scale using ground-control points. A second-order transformation was used for rectification. Average root mean square error within one pixel was maintained while preparing the transformation model. Lambert Conformal Conic (LCC) projection was used during rectification of the image. Each rectified scene was subjected to radiometric correction before mosaicing it to a single image. Scene-wise classification has been carried out. Major vegetation types, deodar forest, mixed conifer forest, chir pine forest, temperate and alpine Location of the study area. CURRENT SCIENCE, VOL. 84, NO. 5, 10 MARCH 2003 RESEARCH ARTICLE meadows were extracted by supervised classification method. Training sites were selected and processed, and features with high classification accuracy were extracted. Using binary image, the remaining area was extracted and put to unsupervised classification. Both unsupervised and supervised classification approaches in isolation could not yield satisfactory results. Hence the approaches were combined in hierarchical pattern to extract the remaining classes. By using these approaches vegetation classes of moist deciduous, dry deciduous, oak, Betula/ Rhododendron, Sal, etc. were extracted. Certain refinements were still necessary taking into account the contextual information collected from ground truth to delineate intermixing tonal characterisitics of the temperate vegetation types such as Ephedra, Hippophae, juniper, etc. Finally independent classified scenes were merged together to generate vegetation map for the entire area. Classified vegetation types have been compared with Figure 2. classification scheme of Champion and Seth24 as represented in Table 1. Phytodiversity analysis Well-distributed samples are taken for information on species occurrence. Sampling intensity of 0.01% of natural vegetation has been found. Higher sample intensity is adopted (than recommended)21 in view of variability in the area. Stratified random sampling approach was followed and number of sample points was distributed to its probability proportional to size. Classified vegetation cover-type has been used for finding sample size. Nestedplot technique25 was used to optimize plot size and collect phytosociological data pertaining to trees, shrubs and herbs in a systematic way. Plot size was established on the basis of species area curve; however to keep the units Schematic representation of geospatial model for biological richness mapping at landscape level (adapted from ref. 23). CURRENT SCIENCE, VOL. 84, NO. 5, 10 MARCH 2003 665 RESEARCH ARTICLE uniform, quadrats were laid according to Roy et al.21. Sampling for trees was carried out using quadrats of 20 m × 20 m in size. Single quadrats each 10 m × 10 m in size were used for shrub and seedling layer within each 20 m × 20m size quadrat. For herbaceous layer, the quadrat size was restricted to 1 m × 1 m. Phytosociological analysis was carried out to determine species richness, economic and ecological importance. Detailed field inventory information is given in Table 2. Economic valuation of biodiversity. Economically important plants are the species which have social and economic value. Different economic values can be assigned to Table 1. Species richness (Shannon–Wiener Index). Species richness can be described as the number of species in a sample or habitat per unit area. The simplest measure of species diversity (H′ ) is based on the total number of Vegetation classes compared with classification of Champion and Seth Satellite-based vegetation cover type Alpine meadow Alpine scrub Betula/Rhododendron Chilgoza Chir pine Deodar Dry deciduous Ephedra Hippophae Juniper Blue pine Mixed conifer Moist deciduous Oak Riverine Sal Scrub Temperate broad-leaved Temperate grassland Temperate scrub different usage of plants. Importance value can be derived based on primary uses like forage, medicinal applications, human food, fuel wood, timber, charcoal, etc. and secondary direct benefits like production of oil, fibre, mat-making, ropes, etc. A scale of 0–10 points for each use was assigned based on the available literature18–20 as well as field information collected from local people to calculate the Total Importance Value (TIV)26 (Table 3). Vegetation type according to classification by Champion and Seth Alpine pastures (15/C3) Dry alpine scrub (16/C1), Deciduous subalpine scrub (14/1S2), Dwarf Rhododendron scrub (15/C2/E1) West Himalayan subalpine birch/fir forest (14/C1a) Neoza pine forest (13/C2a), Dry broadleaved and coniferous (13/C1) Himalayan subtropical pine forest (9/C1) Moist deodar forest (12/C1c) Northern dry mixed deciduous forest (5B/C2), Dry bamboo brakes (5/E9) Dry alpine scrub (16/C1) Hippophae–Myricaria scrub (13/1S1), Hippophae–Myricaria brakes (14/1S1) West Himalayan dry Juniper forest (13/C5), Dwarf juniper scrub (16/E1) Low level blue pine forest (12/2S1), West Himalayan high-level dry blue pine forest (13/C4) Upper west Himalayan temperate forest (12/C2), West Himalayan upper oak-fir forest (12/C2b) Khair-sissu forest (5/1S2) Ban oak forest (12/C1a), Moru oak forest (12/C1b), Oak scrub (12/DS1), Kharsu oak forest (12/C2a) Alder forest (12/1S1), Khair-sissu forest (5/1S2) Dry Siwalik sal forest (5B/C1a) Dry deciduous scrub (5/DS1), Subtropical Euphorbia scrub (9/C1/DS2), Subtropical Euphorbia scrub (DS2) Alder forest (12/1S1) Himalayan temperate pastures (12/DS3) Himalayan subtropical scrub (DS1), Dry temperate scrub (13C2/DS2) Table 2. Phytodiversity analysis of vegetation of Himachal Pradesh Number of species in each habit Vegetation type Alpine meadow Alpine scrub Betula/Rhododendron Chilgoza Chir pine Deodar Dry deciduous Ephedra Hippophae Juniper Blue pine Mixed conifer Moist deciduous Oak Riverine Sal Scrub Temperate broad-leaved Temperate grassland Temperate scrub 666 No. of plots Family Genus Trees Shrubs Herbs 145 204 11 8 39 31 25 10 19 9 19 52 36 38 9 6 17 36 59 62 43 41 54 17 23 66 61 9 35 28 63 88 80 75 30 24 53 86 30 75 98 114 104 34 31 144 155 10 68 55 137 235 221 199 73 58 94 225 98 200 0 0 18 8 35 42 27 – – 6 16 48 58 33 11 9 – 50 – 7 2 46 24 8 63 51 48 6 14 12 44 77 73 61 15 22 60 58 – 79 143 175 112 46 132 186 95 6 81 50 148 306 183 195 85 41 91 283 170 282 Climbers Epiphytes – – – – – 6 2 – – – 2 7 13 – – – 5 2 – 1 – – – – – 2 – – – – 3 2 2 9 – – – – – 1 Total 145 221 154 62 230 287 172 12 95 68 213 440 329 298 111 72 156 393 170 370 CURRENT SCIENCE, VOL. 84, NO. 5, 10 MARCH 2003 RESEARCH ARTICLE species and the total number of individuals in the sample or habitat. Greater the index value, higher the species richness (Table 3). The index represents the average degree of uncertainty in predicting the particular species an individual chosen at random from a sample will belong27. Survey of India (FSI). This may be a result of different classification methods followed in each case and also transformation of 1 : 50,000 scale assessment to 1 : 250,000 Table 3. Ecological importance. Ecological importance with respect to species uniqueness in terms of rare, endangered, threatened, endemic was considered for establishing ecosystem uniqueness. The species recorded during the field data collection were screened for their uniqueness with the help of the Red Data Book and other available literature and accordingly, proportional weights were given to them. The weights of the number of species of IUCN categories present in various forest types were added to derive a relational weight. The weights obtained for various forest types were fed as an input to simulate the biological richness index. Results and discussion Estimated forest cover of the state is 13880 km2, constituting 24.93% of the total geographical area of the region. This estimate was 1.4% more than that by Forest Figure 3. Vegetation type-wise Shannon–Wiener index (H′) and Total Importance Value (TIV) Vegetation type Alpine meadow Alpine scrub Blue pine Betula/Rhododendron Chilgoza Chir pine Deodar Dry deciduous Ephedra Hippophae Juniper Mixed conifer Moist deciduous Oak Riverine Sal Scrub Temperate broadleaved Temperate grassland Temperate scrub Species richness (H′) TIV 5.41 6.55 4.02 4.34 0.40 5.35 7.64 7.13 0.80 1.72 0.73 8.15 6.84 5.78 2.17 2.07 1.56 7.52 4.32 7.25 124 202 637 574 521 198 775 955 85 310 184 1023 1155 991 147 421 743 1045 287 956 Land-use/land-cover type map of Himachal Pradesh. CURRENT SCIENCE, VOL. 84, NO. 5, 10 MARCH 2003 667 RESEARCH ARTICLE scale by FSI28. Vegetation pattern of HP varies from dry scrub at lower altitudes to alpine meadows at higher altitudes. Between these two extremes distinct vegetation zones of dry deciduous forest, moist deciduous forest, pine, oak, deodar, mixed coniferous and temperate broadleaved forests are found. A total of 20 vegetation types were identified (Figure 3, Table 4). Distribution pattern of these forests follows regular altitudinal stratification, except where micro-climatic changes occur due to aspect, slope and edaphic changes breaking the continuity. Roy29 also observed this type of altitudinal control of vegetation in eastern Himalayas. Siwalik hills in HP is dominated by dry deciduous forest with inter-mixing of scrub vegetation. Moist deciduous forest in fragmented patches was observed up to an elevation of 1700 m usually on northern aspects, due to higher moisture availability. Drier vegetation of chir pine and scrub forest was found mostly in southern and southwestern slopes, but chir pine forest is restricted to an elevation of 800 to 1500 m. Dry scrub vegetation distribution was found to be related to edaphic factors, notably dry rocky ridges, and in places where biotic pressure is high in terms of grazing and browsing by domestic herds of cattle and goats. Gregarious formations of deodar (Cedrus deodar Roxb.) and oak (Quercus species) forests were observed in relatively higher altitudes Figure 4. 668 Table 4. Area under different land-cover/land-use classes in the region Land-use/land-cover class Area in km2 Alpine meadow Alpine scrub Betula/Rhododendron Chilgoza Chir pine Blue pine Deodar Dry deciduous Ephedra Hippophae Juniper Mixed conifer Moist deciduous Oak Riverine Sal Scrub Temperate broadleaved Temperate grassland Temperate scrub Orchard Agriculture Barren Water Settlement Snow Cloud Shadow 5346.21 2086.92 455.09 76.98 2005.52 2193.6 2153.35 26.78 81.97 258.43 208.41 3226.72 1573.62 879.38 24.66 306.97 2152.55 408.83 2154.36 321.64 542.77 7924.46 10097.78 386.43 14.45 6160.76 659.21 3945.54 Total 55673.42 Percentage 9.6 3.8 0.8 0.1 3.6 3.9 3.9 0.1 0.2 0.5 0.4 5.8 2.8 1.6 0.04 0.55 3.87 0.73 3.87 0.58 0.97 14.2 18.1 0.69 0.03 11.1 1.18 7.09 100 Biological richness map of Himachal Pradesh. CURRENT SCIENCE, VOL. 84, NO. 5, 10 MARCH 2003 RESEARCH ARTICLE between 1500 and 2500 m. Deodar and oak distributions of these two communities usually occupied the cooler aspects in most of the area. Grassland communities were found in dry and warmer southern slopes at similar elevations. These communities divide at the ridge line. Mixed coniferous forests with consociations of deodar, fir and spruce were delineated just above the oak deodar belt between 2500 and 3200 m altitude. Temperate broadleaved forest occupied the higher portion of the temperate belt, particularly in the outer ranges of the northern aspects. Degradation due to overgrazing has turned these forests into temperate grasslands having a variety of grasses and herbaceous flora. Forests of birch (Betula utilis D. Don), and rhododendron (Rhododendron species) with intermixing of fir (Abies pindrow Royle) forests were found in the subalpine zone. The uppermost portion of these forest forms the tree line in the eastern part of the state. On the other hand, kharsu oak (Quercus semicarpifolia Sm.) forests with stunted growth forms the tree line in most of the alpine zones at the average altitude of 3800 m. Alpine meadows bearing mostly grasses and mesophytic herbs, extended over vast tracks just above the timberline. Alpine scrubs, usually with stunted growth and often with xerophytic elements are found in the dry zones, chiefly on the northern and northeastern aspects broken by the meadows. Pure and relatively bigger patches of juniper scrub were demarcated and more frequently in very dry sites exposed to intense isolation. A more or less pure thicket of Hippophae scrub was observed along the river banks and riverbeds in the subalpine and alpine zones between 2500 m at the northern dry belts and 4200 m in the eastern part of the state. Similarly, fairly bigger patches of Ephedra scrub were delineated in arid cold deserts, usually in xerophytic formations. Mixed coniferous forest shows highest diversity (8.15) with total number of 440 species followed by deodar (7.64), temperate broadleaved (7.52), temperate scrub (7.25) and dry deciduous forest (7.13) with total of 287, 393, 370 and 172 species respectively. These are followed by moist deciduous, alpine scrub, oak, alpine meadow, chir pine, Betula/Rhododendron, temperate grassland and moist deciduous in decreasing order. Table 4 provides detailed information regarding biodiversity. The forests of HP are rich in their biodiversity especially in transitional zones between temperate and alpine (Figure 4). Alpine meadow and mixed conifer forest showed high degree of richness followed by alpine scrub, scrub, deodar and moist deciduous forest communities. The reason could be that these areas experience least disturbance because of very low population density and inaccessibility. It is also observed that disturbance affects biological richness. Relatively more biologically rich area was observed where the area under disturbance possesses very low. However, area under biological richness decreased with increase in disturbance (Figure 5). CURRENT SCIENCE, VOL. 84, NO. 5, 10 MARCH 2003 Figure 5. richness. Relationship between disturbance index and biological Conclusions The study has identified areas with various levels of biological richness. It is hoped that these results and maps will be useful in land-use zonaton and planning for sustainable use of natural resources. It may be said that only with this level of understanding of biodiversity can a long-term success of conservation policies be assured. Disturbance is one of the major factors for biodiversity loss. Biodiversity of forest patches depends on the existing environmental conditions. The analysis done in GIS domain using remotely sensed data and information from phytogeographical analysis revealed that majority of the area is highly disturbed, and therefore it is a matter of great concern for biodiversity conservation. The approach of this study is unique due to representation of the results in spatial form that may help in baseline study, planning of plant species inventory, and act as a prime input for species habitat evaluation, etc. 1. Orians, G. H., Ecol. Appl., 1993, 31, 206–208. 2. Edwards, P. J., Webb, N. R. and May, R. M., Large-Scale Ecology and Conservation Biology, Blackwell, Oxford, 1994, pp. 33–39. 3. Roy, P. S., Dutt, C. B. S., Kushwaha, S. P. S., Sarnam Singh and Hegde, V. S., ISPRS International Symposium on Resource and Environmental Monitoring, 3–6 December 2002, Hyderabad, 2002. 4. Noss, R. F. and Harris, L. D., Environ. Manage., 1986, 10, 299–309. 5. Roy, P. S. and Shrivastava, S., National Workshop on Himalayan Biodiversity 2000: Options for Development, 2–4 November 2000, G.B. Pant Institute of Himalayan Environment and Development, Almora, 2000. 6. Forman, R. T. T. and Godron, M., Landscape Ecology, John Wiley, New York, 1986. 7. Lehmkhul, J. F. and Reggiero, L. F., General Technical Report PNW-GTR-285, USDA Forest Service, Pacific North West Research Station, Portland, Oregon, 1991, pp. 35–46. 8. Ripple, W. J., Bradshaw, G. A. and Spies, T. A., Biol. Conserv., 1991, 57, 73–88. 9. Skole, D. and Tucker, C. J., Science, 1993, 260, 1905–1910. 10. Shirish, A. R. and Roy, P. S., Plant Ecol., 1997, 131, 129–141. 669 RESEARCH ARTICLE 11. Ludeke, A. K., Maggio, R. C. and Ried, L. M., J. Environ. Manage., 1990, 31, 247–259. 12. Mertens, B. and Lambin, E. F., Appl. Geogr., 1997, 17, 143– 167. 13. Roy, P. S., Tomar, S. and Jegannathan, C., NNRMS Bull., 1997, B21, 12–18. 14. Quinn, J. F. and Harrison, S., Oecologia, 1998, 75, 132–140. 15. Joshi, P. K., Sarnam Singh, Shefali Agrawal and Roy, P. S., Curr. Sci., 2001, 80, 941–947. 16. Joshi, P. K., Sarnam Singh, Shefali Agrawal and Roy, P. S., Curr. Sci., 2001, 81, 392–399. 17. Anon, Forests of Himachal Pradesh, Department of Forest Farming and Conservation, State Forest Department, Shimla, Himachal Pradesh Pub, New Delhi, 1993, p. 50. 18. Murti, S. K., Flora of Cold Deserts of Western Himalaya, Vol. I. Monocotyledons, BSI, Kolkata, 2001, p. 452. 19. Nyayar, M. P. and Sastry, A. R. K. (eds), Red Data Book of Indian Plants, BSI, Kolkata, 1987, vol. 1, p. 367. 20. The Useful Plants of India, Council of Scientific and Industrial Research, CSIR Publication, New Delhi, 1999, p. 918. 21. Roy, P. S. et al., Biodiversity Characterization at Landscape level 22. 23. 24. 25. 26. 27. 28. 29. using Satellite Remote Sensing and Geographic Information System, DOS-DBT Users Manual, Indian Institute of Remote Sensing (NRSA), Dept. of Space, Govt. of India, Dehradun, 1999. Turner, M. G. and Gardner, R. H. (eds), Quantitative methods in Landscape Ecology: The Analysis and Interpretation of Landcsape Heterogeneity, Ecological Studies Series, Springer-Verlag, New York, 1990. Roy, P. S. and Tomar, S., Biol. Conserv., 2000, 95, 95–109. Champion, H. G. and Seth, S. K., Revised Forest Types of India, Govt of India Publications, New Delhi, 1968. Kent, M. and Coker, P., Vegetation Description and Analysis: A Practical Approach, CRC Press and Belhaven Press, 1992. Belal, A. and Springuel, I., Nature Resour., 1996, 31, 33–39. Behera, M. D., Ph D thesis, Gurukul Kangri University, Hardwar, 2002. Anon., State of Forest Report, Forest Survey of India, Ministry of Environment and Forests, Govt of India, 1999. Roy, P. S., Geocarto Int., 1987, 1, 16–26. Received 27 August 2002; revised accepted 27 November 2002 FORM IV Particulars of Current Science—as per Form IV under the Rule 8 of the Registration of Newspapers (Central) 1956. 1. Place of Publication: Bangalore 4. Publishers’ Name, Nationality and Address: P. Balaram and S. Ramaseshan, Indian, Current Science Association, Bangalore 560 080 2. Periodicity of Publication: Fortnightly 5. Editors’ Name, Nationality and Address: P. Balaram and S. Ramaseshan, Indian, Current Science Association, Bangalore 560 080 3. Printers’ Name and Address: P. Balaram and S. Ramaseshan Current Science Association, Bangalore 560 080 6. Name and Address of the owner: Current Science Association, Bangalore 560 080 We, P. Balaram and S. Ramaseshan, hereby declare that the particulars given above are true to the best of our knowledge. Bangalore 1 March 2003 670 (Sd/-) P. Balaram and S. Ramaseshan Publishers, Current Science CURRENT SCIENCE, VOL. 84, NO. 5, 10 MARCH 2003