Ann. Bot. Fennici 37: 197–206 Helsinki 27 September 2000 ISSN 0003-3847 © Finnish Zoological and Botanical Publishing Board 2000 Taxonomic status of Saussurea alpina subsp. esthonica (Asteraceae): phenetical analysis Anneli Narits, Malle Leht & Jaanus Paal Narits, A. & Paal, J., Institute of Botany and Ecology, University of Tartu, Lai Str. 40, Tartu 51 005, Estonia Leht, M., Institute of Zoology and Botany, Estonian Agricultural University, Riia Str. 181, Tartu 51 014, Estonia Received 13 April 1999, accepted 8 December 1999 Narits, A., Leht, M. & Paal, J. 2000: Taxonomic status of Saussurea alpina subsp. esthonica (Asteraceae): phenetical analysis. — Ann. Bot. Fennici 37: 197–206. Morphological variation of Saussurea esthonica Baer ex Rupr. and S. alpina (L.) DC. s. str. was studied with different multivariate methods. The most important characters supporting the clustering of the specimens into groups are the shoot hight and characters correlated to it, and hairiness of leaves, while the type of trichomes in both taxa is the same. Hence, these characters depend very much on ecological conditions and are, therefore, taxonomically not reliable enough. It appeared also that several specimens were morphologically intermediate between typical representatives of S. esthonica and S. alpina s.str. Therefore, we consider it more appropriate to treat these taxa as ecogeographical subspecies: S. alpina subsp. esthonica (Baer ex Rupr.) Kupff. and S. alpina subsp. alpina. Chromosome numbers established for S. alpina subsp. esthonica are 2n = 52, 54. Keywords: chromosome numbers, Estonia, Latvia, morphology, multivariate methods, Saussurea alpina, Saussurea esthonica INTRODUCTION The genus Saussurea comprises over 300 species distributed mostly in Asia (Bremer 1994) where, in the Himalayas and high mountains of China, its primary center of diversity lies (Lipschitz 1979). The genus is represented in Europe by three to nine or more species, depending on the species concept used. According to Hegi (1954), only three species occur in Europe: S. alpina (L.) DC., S. discolor (Willd.) DC. and S. pygmaea (Jacq.) Spreng. Lipschitz (1976) added to the list six species found only in the European part of Russia and in the Ukraine. The list becomes even longer, if some subspecies of S. alpina are recognized as species. Saussurea alpina s. lato is a polymorphic arctic-montane species with a disjunct distribution area in Europe and Asia (Hultén & Fries 1986). The only representative of the genus in Estonia 198 and Latvia is S. esthonica Baer ex Rupr., a taxon that has been defined either as a species (Lipschitz 1962, 1979, Üksip 1966, Kuusk 1978, Czerepanov 1995) or a subspecies of S. alpina (Kupffer 1902, Hegi 1954, Lipschitz 1976, Jalas 1980). The first data about S. esthonica was recorded by Ruprecht (1845), who mentioned that a new Saussurea taxon was found from central Estonia by K. E. von Baer in 1844 and described it as a new species. An improved description of S. esthonica was published by Meyer (1855). In 1902, Kupffer found that the differences between S. esthonica and typical arcto-alpine S. alpina were not large enough to justify recognition of two species, and recognized it as a subspecies, S. alpina ssp. esthonica (Baer ex Rupr.) Kupff. At that time the nearest known localities of Saussurea alpina s. lato were in Russia in the Arkhangelsk and Olonets provinces, and therefore the taxa seemed to be geographically well separated. Later, localities in the Leningrad and Novgorod districts were found, plants from there identified as S. alpina s. str. (Juzepchuk 1955, Sokolovskaya 1965). Hence, Lipschitz (1962) stressed that several specimens were morphologically intermediate between S. alpina s. str. and S. esthonica, and some were so close to S. esthonica that they could even be placed in that taxon. And since a race of S. alpina, S. subbendorffii Herd., which is morphologically very similar to S. esthonica, was found growing in East Siberia and Jakutia, Lipschitz has expressed the opinion that S. esthonica is not a local Estonian endemic. According to Laasimer (1965), S. esthonica originated from mires of the late glacial period and belongs in the group of plants reaching their present localites from the south. Ingelög et al. (1993) assume that S. esthonica is a neoendemic of the Baltic Sea region. Saussurea alpina s. lato shows variation especially in the length of the stem, in the shape, size, pubescence and margin of the leaves, in inflorescence type, and in the shape and pubescence of the involucre. The recorded chromosome numbers and ploidy levels of S. alpina s. lato are also quite variable: 2n = 26 (Rostovtseva 1983, Krogulevitsch & Rostovtseva 1984, Lid & Lid 1994), 2n = 36 (Ishikawa 1916, Lid & Lid 1994), 2n = 48 (Sokolovskaya & Strelkova 1948), 2n = 51– 52 (Favarger 1965), 2n = 52 (Löve & Löve 1948, Lavrenko et al. 1991, Lid & Lid 1994), 2n = 54 Narits et al. • ANN. BOT. FENNICI 37 (Sokolovskaya & Strelkova 1960, Morton 1977, Uotila & Pellinen 1985, Lid & Lid 1994), 2n = 72–76 (Rostovtseva 1979), 2n = 76? (Lid & Lid 1994). Chromosome numbers of S. esthonica have not been counted. The aims of the present study were to compare the morphological variation of Saussurea alpina s. str. and S. esthonica and to answer the following questions: 1. How well can these two taxa be distinguished from the statistical point of view? 2. What is the chromosome number of S. esthonica? 3. What is the taxonomic status of S. esthonica? 4. Is S. esthonica an endemic taxon? MATERIAL AND METHODS Morphological variation of Saussurea esthonica was studied from 187 plant specimens collected in 1996 from eight Estonian and two Latvian populations, and from 35 specimens (two populations) from the Herbarium of the University of Tartu (TU) (Table 1, Fig. 1). For studying S. alpina s. str., 43 specimens were collected in 1996 from two Norwegian populations, 79 were obtained from Krakow from the Herbarium of the Polish Academy of Sciences (KRAM), and 54 specimens from the Herbarium of the Norwegian University of Science and Technology (TRH). The material from KRAM originated mostly from the Tatra Mountains, but also included single specimens collected from Finland, Sweden, Switzerland and western Russia (Novgorod region); all the specimens from KRAM were treated conditionally as representatives of one population. The material from TRH originated mainly from the counties of Sør-Trøndelag and Nord-Trøndelag. Altogether, 222 specimens of S. esthonica and 176 specimens of S. alpina s. str. were studied using the 26 macromorphological characters most often considered in the diagnoses of S. alpina (Table 2). The characters were measured on herbarium material with a binocular microscope MBS-2 and a ruler. Micromorphological characters were studied with a scanning electron microscope (SEM) in the Centre for Materials Research of Tallinn Technical University. The epidermis from both sides of ANN. BOT. FENNICI 37 • Taxonomic status of S. alpina subsp. esthonica 199 Fig. 1. Distribution of Saussusrea alpina subsp. esthonica in Estonia and Latvia. Studied localities: Tammiku (1), Paraspõllu (2), Niitvälja (3), Halinga (4), Pärnu-Jaagupi (5), Varangu (6), Kalevi (7), Ambla (8), Hageri (9), Viirika (10) in Estonia and Pope (11), Tukums (12) in Latvia. Table 1. Material studied. Site types of Saussurea esthonica named according to Paal (1997). Site types of S. alpina s. str. according to Nordhagen (1943). ————————————————————————————————————————————————— No. Locality Site type Specimens ————————————————————————————————————————————————— S. esthonica 01. Estonia, Harjumaa county, Tammiku Spring fen 21 02. Estonia, Harjumaa county, Paraspõllu Spring fen 20 03. Estonia, Harjumaa county, Niitvälja Rich paludified grassland 20 04. Estonia, Pärnumaa county, Halinga Rich paludified grassland 20 05. Estonia, Pärnumaa county, Pärnu-Jaagupi Molinia site type 20 06. Estonia, Lääne-Virumaa county, Rich paludified grassland 21 07. Estonia, Jõgevamaa county, Kalevi Rich paludified grassland 20 08. Estonia, Järvamaa county, Ambla (TU) Rich paludified grassland 21 09. Estonia, Harjumaa county, Hageri (TU) Rich paludified grassland 11 10. Estonia, Raplamaa county, Viirika Spring fen 21 11. Latvia, Pope Rich paludified grassland 5 12. Latvia, Tuckums county, Tuckums Rich paludified grassland 20 S. alpina s. str. 13. Norway, Sør-Trøndelag county, Oppdal commune, Kongsvoll Rich, tall-herb birch forest, 30 (trees had been cut) 900 m 14. Norway, Oppland county, Dovre commune, Hjerkinn Minerotrophic, baserich 13 lawn mire, 1000–1200 m 15. Poland, Tatra mountains 1000–1200 m 82 16. Norway, Sør-Trøndelag and Nord-Trøndelag county 180–750 m 54 ————————————————————————————————————————————————— 200 one central cauline leaf, the epidermis of involucre, and the surface of the stem of one specimen from each population were examined on air dried herbarium material coated with gold. For chromosome counts, seeds from three plants from each of the six Estonian populations were collected. Seeds were stratified for three months and germinated at room temperature on moistened filter paper in Petri dishes. Chromosome counts were made from root tips pretreated for 3–4 hours at 15 °C with 0.002% 8-hydroxyquinoline, fixed in Farmer fixative, macerated in 1N HCl at 60 °C for 13–14 minutes, stained with 1% acetoorcein and squashed in acetic acid and glycerin (1:1). The material collected (incl. Narits et al. • ANN. BOT. FENNICI 37 voucher specimens) is preserved in the Herbarium of the Institute of Zoology and Botany (TAA). Data processing To avoid overweighting, characters strongly depending on ecological conditions in the (micro)habitat — shoot height, length of inflorescence and length of the lowest peduncle — were standardized by range before data processing, xi´= (xi – xmin)/(xmax – xmin), (1) where xmin and xmax are the minimum and maximum values of the character. Table 2. Characters used. ————————————————————————————————————————————————— No. Symbols Character Type ————————————————————————————————————————————————— 01 HSH/R Height of shoot (cm) standardised by range metric 02 NLSH Number of leaves on shoot count 03 NH Number of heads (incl. those not flowering yet) count 04 LI Length of involucre (cm) metric 05 WI Width of involucre (cm) metric 06 PB Spot of branching (distance between the topmost inflorescence to ordinary the peduncle of the lowest inflorescence): 1 (< 5 cm), 2 (5–10 cm), 3 (15–20cm), 4 (25–30cm) 07 NB Number of branches count 08 NHSH Number heads on the shoot count 09 LIF/R Length of inflorescence (cm) standardised by range metric 10 LLP/R Length of the lowest peduncle (cm) standardised by range metric 11 LFBL Length of the fifth basal leaf (cm) metric 12 WFL Width of the fifth leaf (cm) metric 13 BLL Base of lower leaves: 1: shortly narrowed, 2: slowly narrowed, ordinary 3: unnarrowed, 4: clasping 14 LUL Length of the topmost leaf (cm) metric 15 WUL Width of the topmost leaf (cm) metric 16 SC Stem colour: 1: green, 2: bright red, 3: dark red ordinary 17 IC Involucre colour: 1: bright, 2: edges dark, 3: dark ordinary 18 NTBL Number of teeth of the fifth basal leaf count 19 HUPB Hairiness of upper part of branch: 1: glabrous, 2: few hairs, ordinary 3: hairy, 4: very hairy 20 HLPB Hairiness of lower part of branch: 1: glabrous, 2: few hairs, ordinary 3: hairy, 4: very hairy 21 HI Hairiness of involucre: 1: glabrous, 2: few hairs, 3: hairy, ordinary 4: very hairy 22 HUSBL Hairiness of upper side of the fifth basal leaf: 1: glabrous, 2: few hairs, ordinary 3: hairy, 4: very hairy 23 HLSBL Hairiness of lower side of the fifth basal leaf: 1: glabrous, 2: few hairs, ordinary 3: hairy, 4: very hairy 24 LI/WI ratio of length and width of involucre ratio 25 LUL/WUL ratio of length and width of the topmost leaf ratio 26 LFBL/WFL ratio of length and width of the fifth basal leaf ratio ————————————————————————————————————————————————— ANN. BOT. FENNICI 37 • Cluster analysis was performed with the program package SYN-TAX 5.02 (Podani 1993). The method of incremental sum of squares (MISSQ) with the distance for mixed data (Podani 1994) was used. For the calculation of common statistics, Spearman rank correlation coefficients between characters were used, and for the evaluation of the importance of characters within clusters by classifactory discriminant analysis, the SAS program package (SAS Institute Inc. 1996) was utilised. Ordination was carried out with principal component analysis (CANOCO package, version 3.1, Ter Braak 1990; and CANODRAW package, version 3.0, Smilauer 1992) with default parameters settings. For the estimation of the adjacency of clusters, the distances of all specimens, or operational taxonomic units (OTUs), from all centroids (except the cluster to which the OTU belongs) were calculated according to the postulate that the jth cluster is interpreted as being adjacent to the ith cluster if the distance between at least one of the OTUs of the ith cluster and the centroid of the jth cluster is smaller than the distance to the centroids of all the other clusters (Paal & Kolodyazhnyi 1983, Paal 1994): min D(xi, mk) = D(xi, mj), k, k ≠ i (2) where xi is the vector of OTU xi, mk and mj are vectors of the kth and jth cluster’s centroid, D = distance (resemblance) function. To visualise the transitions between the two empirical taxa-clusters S. esthonica and S. alpina s. str., and to show the distribution of specimens located between the centroids of the clusters considered, the split window method (Parzen 1962) was applied, as in Paal et al. (1998). The density of the OTUs projection probability distribution on a straight line passing through the centroids of both clusters was calculated as p( x ) = 201 Taxonomic status of S. alpina subsp. esthonica 1 n 1  x – xi  ∑ Φ n i =1 h  h  parameter or window breadth, n is the number of OTUs in the cluster, and xi is the projection of the ith OTU on the line. The density of the normal distribution was regarded as the window function. The smoothing parameter h was calculated according to the formula 1   h = 2 s 0.05 +   n where s is the standard error of the projection. In order to measure the degree of distinctness of clusters, the α-criterion (Duda & Hart 1976) was used.  2 I2  –  1 –  πd I1  α= 8 21 – 2  nd  π d (5) I1 = ∑ xi – m (6) where where p(x) is the distribution density in the point x, Φ is the window function, h is the smoothing 2 x ∈X 2 I2 = ∑ ∑ xi – mi 2 (7) i =1 x ∈X I1 is the sum of square distances between the centroid of a united complex of two clusters, I2 is the sum of square distances between the sample plots and their cluster centroids after dividing the complex into two suboptimal parts, xi is vector of OUT xi, m is vector of the centroid of the united complex, mi is vector of the cluster X centroid, d is dimensionality of the united complex, d = min(q, n – 1), where q and n are the number of characters and specimens in the united complex, respectively. To acquire a better interpretation of the estimates, it is more convenient to use correponding probabilities called coefficients of indistincness (CI) instead of the direct values (Paal 1987, 1994): ∞ CI = (3) (4)  x2  100 exp  –  dx  2 2π α∫ (8) The last two analyses were made by the original SYNCONT 3.0 program composed by S. Kolodyazhnyi, J. Paal and A. Kink. 202 Narits et al. • ANN. BOT. FENNICI 37 Fig. 2. Dendrogram of Saussurea alpina subsp. esthonica and S. alpina s. str. cluster analysis. The hierarchical levels discussed in the text are marked with Roman numbers and letters. RESULTS Chromosome counts Chromosome numbers were counted in six Estonian spontaneous populations of Saussurea esthonica. Populations in Niitvälja, Viirika, Paraspõllu and Kalevi yielded 2n = 52, in the Pärnu-Jaagupi population the number was 2n = 54 and in the Tammiku population it could not be precisely determined 2n = 30(?). Clustering Cluster analysis grouped the data at a quite high level into two unequal clusters (Fig. 2). Cluster I includes the specimens estimated conventionally as belonging to Saussurea esthonica. The cluster branches further into two subclusters, Ia and Ib, consisting of representatives of different S. esthonica populations. Cluster II also branches further into two subclusters. Subcluster IIa comprises putative intermediate specimens identified as S. esthonica or S. alpina s. str., while in cluster IIb, specimens of S. alpina s. str. prevail. Also, the principal component analysis demonstrates that the whole sample consists of two groups corresponding to Saussurea alpina s. str. and S. esthonica, with several specimens on intermediate positions (Fig. 3). According to the first axis, the characters with the highest scores are the ratio of length and width of the topmost leaf (0.96), height of the shoot (0.64), the width of the topmost leaf (–0.59), the ratio of length and width of the fifth basal leaf (0.55), and the number of branches (0.53). Characters with the highest scores by the second axis are: the number of heads (0.83), the number of branches (0.61), the length of inflorescense standardised by range (0.55), the length of the lowest peduncle (0.45) and the ratio of the length and width of the topmost leaf (0,45). Conventionally estimated species-clusters of S. esthonica and Saussurea alpina s. str. were mutually distinct (with coefficient of indistincness CI = 0.0) (Fig. 4), while several pairs of population-clusters were not. Further, the indistinct populations were stepwise joined together until four distinct groups were constituted: cluster 1 includes all populations of S. esthonica, cluster 2 contains specimens from the Kongsvoll population (SørTrøndelag, Norway) and from TRH (Sør-Trøndelag and Nord-Trøndelag, Norway), cluster 3 is formed by specimens from the Hjerkinn population (Oppland, Norway) and cluster 4 by specimens from the Tatra mountains (Poland). ANN. BOT. FENNICI 37 • Taxonomic status of S. alpina subsp. esthonica 203 Characters SEM studies showed that hairs of specimens from both taxa, Saussurea esthonica and S. alpina s. str., as well as their stomata, are of the same type. Leaves and stems are covered with long, simple, smooth unicellular thin hairs and multicellular thick hairs, multicellular glandular trichomes also occured. Hairs on involucres are smooth, thin and curved. Differences between the two taxa appeared only in the density of hairs: plants of S. alpina are much more hairy. The upper surface of the leaves of S. esthonica is either glabrous or with single hairs, the lower surface of some S. esthonica specimens is also nearly glabrous while the surfaces of S. alpina are always covered with trichomes. According to the F-criterion of discriminant analysis, the most important characters for distinguishing the species are: height of the stem, hairiness of lower surface of the fifth leaf, spot of branchTable 3. Importance of characters according to classifactory discriminant analysis. F-criterion significance level for all character is below 0.001, except BLL (p = 0.012), SC (p = 0.013) and LUL (p = 0.099). Symbols as in Table 2. ———————————————————————— Characters F-criterion ———————————————————————— HSH/R 828.86 HLSBL 299.47 LUL/WUL 223.59 PB 214.83 LLP/R 148.94 LI/WI 141.61 HUSBL 139.25 LIF/R 127.56 HLPB 125.38 LFBL 122.11 NB 118.57 WUL 117.84 HUPB 108.01 NLSH 77.53 IC 50.47 LFBL/WFL 45.44 WI 41.09 NHSH 40.02 WFL 37.63 NLSH 32.70 LI 27.53 HI 14.48 NTBL 13.80 BLL 6.30 SC 6.21 LUL 2.74 ———————————————————————— Fig. 3. Classification polygons of conventionally estimated specimens of Saussurea esthonica and S. alpina s. str. superimposed on to PCA ordination. The first component counts for 48.6% of the variation and the second component for 17.1%. ing, number of branches, length of inflorescence and length of the lowest peduncle (Table 3). According to Spearman’s rank correlation coefficient most of the characters are well correlated with each other, uncorrelated are only the colour of the stem and the shape of the leaf base. Length of inflorescence is highly correlated with length of the lowest peduncle (r = 0.93), height of the shoot (r = 0.75), spot of branching (r = 0.81), number of branches (r = 0.79) and number of heads (r = 0.66). Number of branches is also strongly correlated with spot of branching (r = 0.64) and number of heads (r = 0.77). These characters constitute a quite dense correlation group, others stand more or less separately. Mean height of Saussurea alpina s. str. plants (26.6 cm) is smaller than the height of the smallest S. esthonica specimens (Table 4). The highest S. esthonica plants grow in Pärnu-Jaagupi (mean 74.0 cm), and the smallest in the Hageri population (44.9 cm). The most hairy plants belong to the Kongsvoll population (Norway) and the most glabrous ones to the Pärnu-Jaagupi population (Estonia). 204 Narits et al. • ANN. BOT. FENNICI 37 Fig. 4. Specimens projection probability distribution according to the split window method. The curve marked with dots portrays the empirically estimated Saussurea esthonica specimens, the curve shown with dash-and-dot line corresponds to the S. alpina s. str. specimens. The continuous curve above them represents the joint sample specimens projection probability distribution. E and A mark the centroids of S. esthonica and S. alpina s. str., respectively; the dotted lines perpendicular to the line through the cluster centroids delimits the transition zone. DISCUSSION According to Favarger (1965), the basic chromosome number in the genus Saussurea is x = 13. Thus, S. esthonica is a tetraploid (2n = 52, 54) with a tendency towards aneuploidy. Lipschitz (1976) states that the polymorphism of West-European Saussurea alpina s. lato is partly the result of hybridization with S. discolor. Therefore, S. esthonica could be an allotetraploid which has arisen from hybridization between diploids of S. alpina and S. discolor in the overlapping part of their distribution areas, while later, in the Post Glacial period, as Laasimer (1965) supposed, it spread northwards to Latvia and Estonia. However, taking into account that distances between S. discolor and S. esthonica areas are large, and that there exists a Siberian taxon S. stubendorffii morphologically very close to S. esthonica (Lipschitz even considered uniting them), which for geographic reasons cannot have arisen from the hybridization of S. alpina and S. discolor, we prefer the idea of the endemic origin of S. esthonica by the doubling of the chromosomal set of dip- Table 4. Mean ± standard deviation of quantitative characters. Symbols as in Table 2. ————————————————————————————————————————————————— No. Character Saussurea alpina Saussurea esthonica ————————————————————————————————————————————————— 01 HSH 26.63 ± 9.78 64.36 ± 11.91 02 NLSH 11.67 ± 2.94 15.44 ± 4.27 03 NH 6.96 ± 2.93 10.00 ± 5.41 04 LI 1.09 ± 0.09 1.17 ± 0.17 05 WI 0.88 ± 0.14 0.77 ± 0.18 07 NB 2.84 ± 1.35 4.89 ± 1.78 08 NHSH 2.85 ± 1.04 2.09 ± 0.91 09 LIF 2.47 ± 1.87 7.69 ± 5.16 10 LLP 1.58 ± 1.10 5.22 ± 3.61 11 LFBL 10.12 ± 2.56 14.13 ± 3.65 12 WFL 1.82 ± 0.57 1.44 ± 0.62 14 LUL 4.04 ± 1.29 3.87 ± 2.02 15 WUL 0.59 ± 0.32 0.26 ± 0.23 18 NTBL 13.0 ± 2.06 14.3 ± 3.89 ————————————————————————————————————————————————— ANN. BOT. FENNICI 37 • Taxonomic status of S. alpina subsp. esthonica loid S. discolor. This process may result in a population able to inhabit localities new in comparison with those of its diploid predecessors (Grant 1981). As the maximum elevation in the Baltic region is only 317 m, S. esthonica occurs here in conditions quite different from those of the arctoalpine habitats of typical S. alpina: on swampy meadows, wooded meadows, in shrubland and in mires (Kuusk 1978). Morphologically the specimens clustered clearly into two distinct clusters corresponding to empirically identified Saussurea esthonica and S. alpina s. str. MISSQ clustering and PCA revealed the existence of intermediates between typical S. esthonica and S. alpina s. str. specimens (Figs. 2 and 3). The most important characters in the conventional delimitation of Saussurea esthonica and S. alpina s. str. were the height of the stem (even despite of standardising this parameter by range) and the hairiness of lower side of the fifth basal leaf (Table 3). Both these characters depend greatly on ecological conditions — light, humidity, etc. Therefore, it is to be expected that S. esthonica, which grows in moist places among high grasses or shrubs, is taller than S. alpina s. str. growing in mountains in well lit and less humid sites. From the taxonomical point of view height is not considered a reliable character in delimiting taxa. In some cases even definite correlation between the height of the plant and its geographical origin occurred — specimens of more northern origin being considerebly lower than those from further south (Rousi 1971). Leaves of S. alpina are more hairy than leaves of S. esthonica, while their hairs are of the same type. Density of hairs is also a character that depends on habitat conditions: in dry and open windy places plants are more hairy than those growing in shady moist habitats (Rousi 1965). Thus, morphological data taken from specimens growing in different habitats should be used with great care (Øvstedal 1998). Therefore, since the most important characters supporting the clustering the material into groups are taxonomically not reliable enough and numerous specimens with intermediate characters exist, we cannot recognize these taxa as independent species. Hence, as they are geographically more or less separated and grow in different ecological conditions, we consider it more appropri- 205 ate to treat these taxa as eco-geographical subspecies: Saussurea alpina subsp. esthonica (Baer ex Rupr.) Kupff. and S. alpina (L.) DC subsp. alpina. For more detailed studies growth experiments would be useful — comparison of the morphology of taxa, which originally inhabit ecologically different sites, after cultivation in ecologically identical conditions, could help to decide about the reliability of characters: do they depend on the habitat conditions or not. ACKNOWLEDGEMENTS: The authors thank Dr. Eli Fremstad who kindly collected material from two Norwegian populations and supplied us with the material from TRH. We also thank Dr. Viesturs Òulcs from the Institute of Biology, Latvia, for his assistance in collecting material from Latvia. The authors are grateful to Prof. Urve Kallavus for helping with SEM studies, Ms. Silvia Sepp for helping with SAS and to Dr. Vello Jaaska and Ms. Silvia Sepp for critical reading of the manuscript. The English was revised by Mr. Ilmar Part. The study was supported by the Estonian Science Foundation (grants 2218, 3574 and 0552). REFERENCES Bremer, K. 1994: Asteraceae: cladistics and classification. — Timber Press, Oregon. 752 pp. Czerepanov, S. K. [Xerepanov, S. K.] 1995: Plantae vasculares Rossicae et civitatum collimitanearum. — Mir i Semia, S.Petropolis. 990 pp. [In Russian]. Duda, R. & Hart, P. [Duda, R. & Hart, P.] 1976: [Patterns classification and scene analysis.] — Mir, Moscow. 511 pp. [In Russian.] Favarger, C. 1965: Notes de caryologie Alpine IV. — Bull. Soc. Neuchat. Sci. Nat. 88: 5–60. Fedorov, A. A. (ed.) [Fedorov, A. A.] 1969: Chromosome numbers of flowering plants. — Nauka, Leningrad. 926 pp. [In Russian.] Grant, V. 1981: Plant speciation. 2 ed. — Columbia Univ. Press, New York. 563 pp. Hegi, G. 1954: Saussurea DC. — In: Hegi, G. (ed.), Illustrierte Flora von Mitteleuropa 6(2): 835–840. Carl Hauser Verlag, München. Hultén, E. & Fries, M. 1986: Atlas of North European vascular plants. Vol. 2. — Koeltz Sci. Books, Königstein. 1172 pp. Ingelög, T., Andersson, R. & Tjernberg, M. 1993: Red data book of Baltic region. Part 1. — Swedish Threatened Species Unit, Uppsala. 95 pp. Jalas, J. 1980: Saussurea DC. — Läätteen suku. — In: Jalas, J. (ed.) Suuri kasvikirja 3: 792–794. Otava, Helsinki. Juzepchuk, S. [ìzepxuk, S.] 1955: List of the Herbarium of Flora URSS 18: 77–80. Editio Acad. Scient. URSS, Moscow, Leningrad. [In Russian.] 206 Krogulevich, R. & Rostovtseva, T. [Krogulevix, R. & Rostovceva, T.] 1984: Chromosome numbers of flowering plants of Siberia and Far East. — Nauka, Novosibirsk. 286 pp. [In Russian.] Kupffer, K. R. 1902: Saussurea alpina DC. subsp. esthonica (Baer ex Rupr. pr. sp.) Kupffer. — Korrespondenzbl. Naturf.-Ver. Riga 45: 94–103. Kuusk, V. 1978: Perekond soojumikas — Saussurea DC. — In: Kask, M. (ed.), Eesti NSV Floora 6: 377–380. Valgus, Tallinn. Laasimer, L. 1965: Eesti NSV taimkate. — Valgus, Tallinn. 397 pp. Lavrenko, A., Serditov, N. & Ulle, Z. [Lavrenko, A., Serditov, N. & Ulle, Z.] 1991: Chromosome numbers in some species of vascular plants from PechoroIlychky Reservation (Komi ASSR). — Bot. Zh. 776: 473–479. [In Russian]. Lid, J. & Lid, D. T. 1994: Saussurea DC. — In: Elven, R. (ed.), Norsk flora. 6th ed.: 628. Det Norske Samlaget, Oslo. Lipschitz, S. [Lipwitc, S.] 1979: Genus Saussurea DC. (Asteraceae). — Nauka, Leningrad. 283 pp. [In Russian.] Lipschitz, S. 1976: Saussurea DC. — In: Tutin, T. G., Heywood, V. H., Burges, N. A., Moore, D. M., Valentine, H., Walters, S. M. & Webb, D. A. (eds.), Flora Europea 4: 216–217. Cambridge Univ. Press, Cambridge. Lipschitz, S. [Lipwitc, S.] 1962: Genus Saussurea DC. — In: Schischkin, B. K. & Bobrov, E. G. [Wiwkin, B. K. & Bobrov, E. G.] (eds.), Flora URSS 27: 361–537. Editio Acad. Scient., Moscow & Leningrad. [In Russian]. Löve, A. & Löve, D. 1948: Chromosome numbers of northern plant species. — Ingólfsprent, Reykjavik. 131 pp. Meyer, C. A. 1855: Kleine Beiträge zur nähern Kenntniss der Flora Russland’s. — Mem. l’Acad. Imp. Sci. SaintPetersbourg 7: 5–6. Morton, J. K. 1977: A cytological study of the Compositae (excluding Hieracium and Taraxacum) of the British Isles — Watsonia 11: 211–223. Orlova, N. [Orlova, N.] 1966: Genus Saussurea DC. — In: Poyarkova, A. I. [Poärkova, A. I.] (ed.), Flora of Murmansk region 5: 240–241. Nauka, Moscow & Leningrad. [In Russian]. Øvstedal, O. 1998: Variation within some Nordic Saxifraga species (Saxifragaceae). — Nordic J. Bot.18: 171–181. Paal, J. 1987: Taxonomic continuum, some problems and methods for its quantitative analysis. — In: Laasimer, L. & Kull, T. (eds.), The plant cover of Estonian SSR. Flora, vegetation and ecology: 108–122. Valgus, Tallinn. Paal, J. 1994: Moss synusiae in South Estonian forests. — Folia Geobot. Phytotax. 29: 497–509. Paal, J. 1997: Eesti taimkatte kasvukohatüüpide klassifikat- Narits et al. • ANN. BOT. FENNICI 37 sioon. — Keskkonnaministeeriumi & ÜRO Keskkonnaprogramm, Tallinn. 297 pp. Paal, J. L. & Kolodyazhnyi, S. F. [Paal, J. L. & Kolodäånyj, S. F.] 1983: Quantitative methods of analysis transitions between the syntaxa of vegetation. — Bot. Zh. 68: 1467–1474. [In Russian]. Paal, J., Kullmann, B. & Huijser, H. A. 1998: Multivariate analysis of the Scutellinia umbrorum complex (Pezizales, Ascomycetes) from five ecotopes in Netherlands. — Persoonia 16: 491–521. Parzen, E. 1962: An estimation of a probability density function and mode. — Ann. Math. Stat. 33: 1065–1076. Podani, J. 1993: SYNTAX-pc — computer program for multivariate data analysis in ecology and systematics. Version 5.0 User’s Guide. — Scientia Publ., Budapest. 104 pp. Podani, J. 1994: Multivariate data analysis in ecology and systematics. A methodological guide to the SYN-TAX 5.0 package. — Ecol. Comput. Ser. 6. SPB Acad. Publ., The Hague. 317 pp. Rostovtseva, T. S. [Rostovceva, T. S.] 1979: Chromosome numbers of some species of the family Asteraceae Dumort. — Bot. Zh. 64: 582–589. [In Russian]. Rostovtseva, T. S. [Rostovceva, T. S.] 1983: Chromosome numbers of some species of family Asteraceae Dumort. II. — Bot. Zh. 68: 660-664. [In Russian]. Rousi, A. 1965: Biosystematic studies on the species aggregate Potentilla anserina L. — Ann. Bot. Fennci 2: 47–112. Rousi, A. 1971: The genus Hippophae L. A taxonomic study. — Ann. Bot. Fennici 8: 177–227. SAS Institute Inc. 1996. SAS/STAT user’s guide. Software: changes and enhancement through release 6.11. — SAS Inst. Inc., WC. 1104 pp. Òmilauer, P. 1992: CanoDraw 3.0. User’s guide. — Scientia Publ., Budapest. 118 pp. Sokolovskaya, A. [Sokolovskaä, A.] 1965: Genus Saussurea DC. — In: Flora of Leningrad region: 264–265. Editio Leningrad Univ., Leningrad. [In Russian]. Sokolovskaya, A. & Strelkova, O. [Sokolovskäa, A. & Strelkova, O.] 1960: Geographical distribution of polyploid plants in Eurasian arctic region. — Bot. Zh. 45: 369–381. [In Russian]. Uotila, P. & Pellinen, K. 1985: Chromosome numbers in vascular plants from Finland. — Acta Bot. Fennica 130: 1–37. Takhtajan, A. [Tahtadåan, A.] 1990: Chromosome numbers of flowering plants of Flora URSS. — Nauka, Leningrad. 507 pp. [In Russian]. Ter Braak, C. J. F. 1990: Updated notes: CANOCO version 3.1. — Wageningen. 26 pp. Üksip, A. 1966. Perekond soojumikas Saussurea DC. — In: Kask, M. & Vaga, A. (eds.), Eesti taimede määraja: 668. Valgus, Tallinn.