The Purple Saxifrage, Saxifraga oppositifolia, in Svalbard: two taxa or one? ANNE K. BRYSTING, TOVE M. GABRIELSEN, OVE S B R L I B d m N , ODDMUND YTREHORN and CHRISTIAN B R O C H M A " Brysting, A. K., Gabrielsen, T. M., S ~ r l i b r l t e n , O., Ytrehorn, 0. & Brochmann, C. 1996: The Purple Saxifrage, Saxifraga oppositifolia, in Svalbard: two taxa or one? Polar Research 15(2), 93-105. Several studies have demonstrated high levels of genetic (DNA), ecophysiological, ecological, and morphological variation within the species Purple Saxifrage, Saxifraga oppositifolia, in Svalbard. It has recently been proposed that S. oppositifolia is represented by two conspicuously different subspecies in this archipelago: ssp. reptans, a late-flowering, prostrate ecotype of snow-protected, damp habitats, and ssp. pulvinata, an early-flowering, cushion-like ecotype of dry, wind-exposed heaths and ridges. It has also been suggested that the subspecies may be differentiated at the tetraploid and diploid levels, respectively, which would promote reproductive isolation. These hypotheses are tested by examining variation in morphology, ecology, and pollen size and stainability in 150 plants of S. oppositifolia growing in 50 vegetation samples at four sites in the Kongsfjorden area. Although analyses of the various data sets demonstrated the large variation within the species, the material could not be separated into distinct groups. The morphological variation was continuous along local ecological gradients. The pollen grains were fully stainable and the pollen diameter data showed a unimodal distribution, suggesting that the plants analysed represent only one ploidal level. These results reject a hypothesis that the morphologically intermediate plants are hybrids between two taxa at different ploidal levels. Thus, the conspicuous variation in S. oppositifolia in Svalbard probably results from local, in situ ecoclinal differentiation. Although this variation clearly is wifhout taxonomic significance, it is important in the broader context of arctic conservation biology and the potential impact of global warming on arctic vegetation. Anne K. Brysting, rove M. Gabrielsen, Oddmund Ytrehorn and Christian Brochmann, Botanical Garden and Museum, University of Oslo, Trondheimsveien 238, N-0562 Oslo, Norway; Ove S@rlibriten, Department of Biology, Division of Zoology, University of Oslo, P.O. Box 1050 Blindern, N-0316 Oslo, Norway. Introduction It has been shown recently that, although fewer in number than in more southern areas, many arctic plants possess extraordinary high levels of varia- tion. High levels of intraspecific variation in arctic plants have been documented in their ecological traits (McGraw & Antonovics 1983; McGraw 1995a, b), ecophysiology and morphology (Craw- ford et al. 1993), and molecular genetics (Broch- mann et al. 1992a, b, c; Haraldsen & Wesenberg 1993; Nordal & Iversen 1993; Gabrielsen et al. 1995; Hagen et al. 1995; Tollefsrud et al. 1995; Bauert 1996; Brysting et al. in press). For polyploid species, which constitute the majority of arctic floras, use of a wide species concept is often necessary because many such species are composed of races of different hybrid origins that bridge the morphological discontinuities origin- ally found between their diploid progenitors (Brochmann 1992b, 1995a, b). This type of evolution is probably associated with the turbulent climatic history of arctic areas, with repeated glaciations/deglaciations, massive plant migra- tions, and frequent hybridisation. In diploid arctic species, high levels of variation may be caused by differential selection along gradients in snow cover and duration of the growth season. Such factors vary considerably over short distances in the treeless arctic and alpine areas, and this is probably the main reason for the ecotypic differentiation observed in the arctic diploid Dryas octopetala (McGraw & Antonovics 1983; McGraw 1995b). The high level of intraspecific variation in arctic plants has important implications for arctic conservation biology. It is widely accepted that genetic variation and ecotypic differentiation is necessary for the long-term persistence of a species, which is of particular relevance today with regard to potential global warming (Craw- ford et al. 1993; McGraw 1995b). It is paradoxi- 94 A. K . Brysting et al. cal, however, that traditional conservation biology focuses on taxon numbers and rare taxa rather than on the genetic diversity occurring within widely defined, widespread taxa, which probably is a better approach to conserving arctic bio- diversity. The Purple Saxifrage, Saxifraga oppositifolia L., is a variable species with wide circumpolar distribution, extending also into the higher mountains southwards in Europe, Asia, and North America (Hultbn & Fries 1986). It is a long-lived perennial with opposite, small leaves and con- spicuous, pink or purple flowers appearing soon after snow-melt. Although this species, in likeness with most other arctic species, is self-compatible, it strongly depends on insect pollination for seed set (Stenstrom & Molau 1992). The plants grow in a wide range of habitats in the Arctic, including moist snow beds, heaths, dry ridges, and barren, recently unglaciated areas. The broad morphological variation within the species has been noted by many scientists, and various attempts have been made to split it into several taxa (see e.g., Hayek 1905). The most recent taxonomic treatment of the intraspecific variation in S. oppositifolia is that of Webb & Gornall (1989), who closely follow Engler & Irmscher (1919). Webb & Gornall (1989) recog- nise one widely distributed and morphologically variable subspecies, ssp. oppositifolia (the only subspecies they recognise in the European Arctic), four subspecies endemic to the central and southern European mountains, and two additional subspecies occurring in the North American Arctic (and probably across the Beringia into eastern Asia (Hultbn & Fries 1986). The species also varies in chromosome number. Most populations investigated are diploid (2n = 26), but several tetraploid (2n = 52) popula- tions have been reported from arctic areas in Europe, Asia, and North America (Hedberg 1967; Johnson & Packer 1968; Lave & Liive 1975). Two chromosome counts, both tetraploid, have been published for S. oppositifolia in Svalbard, one from the @fjorden area (Flovik 1940) and one from the Ny Alesund area (Chapman 1995). All studies from the rest of northern Europe have reported a diploid number (e.g., Skovsted 1934; E v e & Lave 1956). To what extent the chromosomal variation in S. oppositifolia is correlated with the morphological differentiation of the species is unknown. Recently, there has been particular focus on the variation within S. oppositifolia in Svalbard. This focus has been accentuated by the publication of a colo;r photograph of a cushion-formed plant from Ny Alesund on the front cover of the journal Molecular Ecology (Abbott et al. 1995), and by the recognition of two ecologically and morpho- logically divergent subspecies, one cushion- formed and one prostrate, in the new edition of Rbnning’s Svalbards flora (Rbnning 1996). This variation in growth form within S. oppositifolia in Svalbard was already reported by Anderson & Hesselman (1900). Recent studies have demon- strated that there not only is large morphological and ecological variation in S. oppositifolia in the archipelago, but that the species also shows large variation in DNA markers (Abbott et al. 1995; Gabrielsen et al. 1995, Gabrielsen 1996) and ecophysiological traits (Crawford et al. 1993, 1994; Crawford & Abbott 1994). The different growth forms observed in S. oppositifolia in Svalbard have also been suggested to be correlated with other characteristics (Craw- ford et al. 1993; Rbnning 1996; Elven & Elvebakk 1996). The prostrate plants grow in snow- protected, damp habitats and have high photo- synthetic rates, distant leaves with marginal cilia, and open flowers with narrow, non-overlapping petals, whereas the cushion-formed plants grow on dry, wind-exposed ridges and have low photosynthetic rates, dense, overlapping leaves entirely or mostly without cilia, and urn-shaped flowers with broad, overlapping petals. In the new edition of Rbnning’s Svalbards flora ( R ~ n n i n g 1996), the prostrate plants are referred to as S. oppositifolia ssp. reptans (Anderss. & Hesselm.) Rbnning, and the cushion-formed plants are referred to as S. oppositifolia ssp. pulvinata (Anderss. & Hesselm.) Rbnning. Although only tetraploid plants of S. oppositifolia have been recorded in Svalbard, it has been suggested that both diploid and tetraploid plants are present and that the difference in ploidal level may be correlated with growth form (Crawford et al. 1993; cf. also Liive & E v e 1951, 1975). The aim of the present study was to determine if it is reasonable to divide S. oppositifolia in Svalbard into two ecologically, morphologically, and possibly also chromosomally, distinct taxa, or if the variation merely is ecoclinal, representing continuous morphological variation along local ecological gradients. Four different data sets were produced to answer this question: (1) Ecological The Purple Saxifrage 95 parameters were measured or estimated to detect possible habitat differences correlated with growth form differences; (2) macro-morphologi- cal characters were measured in field-collected plants to detect any possible characters correlated with ecological traits and/or growth form differ- ences; (3) pollen size was measured to reveal possible reproductive barriers owing to ploidal differences between growth forms (see Broch- mann 1992a); and (4) pollen stainability was estimated to reveal if there was lowered fertility in morphologically intermediate plants, if such plants (potential hybrids) were detected. Materials and methods Ecological data A total of 50 vegetation samples, each of 1 mz, were analysed at four sites in the Kongsfjorden area where the postulated subspecies grow in close geographic proximity: (1) B r ~ g g e r h a l v ~ y a , Stuphallet, UTM VH 28-29 68-69, 12 samples (nos. 22-33, cf. Table 2); (2) BrBggerhalvoya, NW of Bayelva, UTM VH 3265, 16 samples (nos. 4-19); (3) BrGggerhalvBya, S of Ny h e s u n d , UTM VH 34-35 63, 14 samples (nos. 1-3,20, 21, 42-50); and (4) Blomstrandoya, London, UTM VH 3768, 8 samples (nos. 34-41). The vegetation samples were subjectively selected so that the samples at each site covered the range of variation observed in the vegetation where S. oppositifolia occurred, and so that at least one plant of S. oppositifolia occurred within each sample. In each sample, the percentage cover of each vascular plant species (nomenclature follows Elven 1994) and the total cover of the vegetation, vascular plants, cryptogams, bryophytes, lichens, stones, and bare ground were recorded. In addition, the humidity conditions (1 = dry, 2 = in- termediate, 3 = damp, 4 = moist) and degree of wind exposure (1 = protected, 2 = intermediate, 3 = exposed) were subjectively estimated from the topography and a general impression of the sample site. The aspect and slope (angle of inclination) were measured for each sample. The vegetation samples and the surrounding areas were classified into five vegetation types following Brattbakk (1981): (1) Dryas octopetala heaths and ridges, (2) Luzula arcuata-lichen heaths, (3) S. oppositifolia-lichen heaths, (4) Talus slope vegetation, and (5) Deschampsia alpina snowbeds. Morphological and pollen stainability data A total of 150 plants of S. oppositifolia occurred within the vegetation samples. In the field, each plant was tentatively classified after growth form as typically prostrate or typically cushion-formed. It was, however, difficult to classify all plants into these two groups, and several plants were tentatively grouped as “intermediates”. Thirteen macroscopic morphological characters were scored for each plant. Seven characters were quantitative, three characters were qualitative, and Table 1 . Macroscopic characters used in morphometric analyses of Saxifraga oppositifolia. Five measurements were made of each character for each plant, and the median was used in further analyses. Characters included in the PCO analysis are indicated. Measurements in mm. Character Note PCO Internode length Leaf width Leaf length Leaf index No. of leaf cilia No. of leaf cilia pr. Petal width Petal length Petal index Petal apex Overlap of petals Flower shape Flower diameter mm the five longest internodes selected in each plant maximum width leaf widthileaf length number of cilia along one side of the leaf number of ciliafleaf length maximum width petal width/petal length 1 - rounded, 2 - intermediate, 3 - acute 1 - non-overlapping, 2 - tangent, 3 - overlapping 1 - flattened, 2 - intermediate, 3 - urn-shaped from petal apex to petal apex 96 A. K. Brysting et al. Table 2. Floristic composition, frequency, and percentage cover of vascular plants in 50 vegetation samples including Saxifraga oppositifolia. Vegetation type no. &talus slope vegetation, no. 5-Deschampsia alpina snowbed. Two species groups in Saxifraga were scored collectively because of the early developmental stage of the plants (S. cernua/hyperborea/rivuluris and S. nivalis/ tenuis). 1 Dryas octopetala ridgeheath 2 Luzula arcuata - lichen heath Speciedsample no. 10 11 12 13 17 18 25 26 28 30 33 38 freq. 1 2 3 42 43 44 45 46 47 48 49 50 freq. Saxifiaga oppositifolia Bistorta vivipara Cardamine bellidifolia Carex rupestris Cerastium arcticum Cerastium regelii Cochlearia groenlandica Deschampsia alpina Draba lactea Dryas octopetala Equisetum scirpoides Juncus biglumis Lwula arctica Lwula arcuata Oxyria digyna Pecicularis hirsuta Poa alpina Sagina nivalu Salk polaris Saxifraga cern.lhyp.lriv. Saxijiaga cespitosa Saxifraga nivalisltenuis Silene acaulis Silene uralensis 2 4 2 3 4 2 1 0 5 0 8 1 0 1 5 1 0 1 . 0 5 0 5 2 1 0 5 4 3 2 5 5 5 4 5 1 . 0 1 1 1 1 - 1 1 7 2 2 1 1 0 . 9 - - - - I - - - - - 1 - 0 . 2 0.2 - - - - - - - - - 1 - 1 0 . 2 1 1 - - _ - _ - - _ - - _ _ 0.2 - _ - - _ _ _ - _ _ - _ 0.0 1 - 0 . 5 1 - - 1 1 1 1 - 1 1 1 - 0 . 7 _ _ 0.2 _ _ _ _ _ _ 1 - _ _ _ _ 0.1 - _ 0 . 0 - - - 1 - - 1 - - 1 - - 0 . 3 _ _ 0 . 1 - - - 1 - - 1 - - - 1 - 0 . 3 _ _ 0.2 1 - - 1 - 1 - - - - - - 0 . 3 4 5 - 0.6 _ _ - - _ - - _ _ _ _ _ 0.0 _ _ 0.0 _ _ _ _ _ - _ - _ _ _ _ 0.0 _ _ 0.2 _ _ - - _ - - - _ _ - _ 0.0 - - o . o - - - 1 _ - _ - _ I - - 0.2 _ _ 0.2 5 3 1 1 5 3 1 2 7 3 3 - 0 . 9 _ _ 0.0 1 1 1 1 1 1 1 - - - 1 1 0 . 8 _ _ 0 . 2 - - - - 1 - - 1 1 1 1 - 0 . 4 - 1 0 . 2 - 1 1 - 1 0 . 1 - - - 1 - - 1 - 1 1 - - O o . 3 1 1 0 . 8 1 - 6 1 4 1 5 - 2 5 1 5 3 5 5 0 . 8 - 1 0 . 3 1 1 1 3 1 - 1 - - 1 1 - 0 . 7 _ _ 0.3 2 1 1 1 1 1 1 - - 1 1 - 0 3 _ _ 0.1 - - 1 - 1 - _ - - I - - 0.3 _ _ 0.1 - _ - - - - _ - - - - - 0.0 _ _ 0.3 _ _ - - _ - - - - _ - - 0.0 Additional species (occurring in 1-3 samples only): Draba alpina, D. corymbosa, D. oxycarpa, D. subcapitatu, D. micropetala, Equisetum arvense, E . variegatum, Minuartia biflora, Papaver dahlianum, Saxifraga aizoides, S. hieracifolia, Stellaria longipes, Carew misandra, Phippsia sp. three characters were ratios between quantitative characters (Table 1). When possible, five mea- surements were made of each character for each plant, and the median of these five measurements was used in further analyses. The leaf characters were measured on the five longest shoots, on a leaf positioned 1 cm below the shoot apex. The flower characters were measured on the five best developed flowers on each plant. Pollen size and stainability were measured in 36 plants, 12 plants from each of the three tentative growth form groups (typically prostrate, typically cushion-formed, or "intermediate"). Pollen grains were stained in cotton blue in lactophenol (Radford et al. 1974). The largest diameter of ten subjectively selected pollen grains of each plant was measured using a light microscope. Pollen fertility was estimated by counting the percentage of well-stained and regularly shaped pollen grains in a sample of 200 grains per plant. Herbarium vouchers of 20 representative speci- mens of S. oppositifolia, selected among the 150 plants analysed, have been deposited at the Botanical Museum, University of Oslo. Multivariate analyses and statistics Descriptive statistics and correlations were per- formed using SPSS for Windows (Norusis 1993). The ecological data were analysed using de- trended correspondence analysis (DCA; cf. Hill & Gauch 1980) based on floristic composition of the vegetation samples. The CANOCO program (Braak 1987) was used for these computations with standard parameter choice except that rare species were downweighted according to Eilertsen et al. (1990). Principal coordinate analysis (PCO; Gower 1966, 1967) and nonmetric multidimen- sional scaling (NMDS; Kruskal 1964) were performed using NTSYS-pc (Rohlf 1990) based on seven morphological characters, excluding qualitative multistate characters and characters with overlapping information (Table 1). The quantitative, continuous morphological characters were In-transformed before analyses to eliminate The Purple Saxifrage 97 Table 2. Continued 3 Saxifraga oppsififoolia - lichen heath 4 5 Total 4 5 6 7 8 9 14 15 16 19 20 21 22 24 29 31 32 34 35 36 37 39 40 41 freq. 23 27 freq. 5 5 2 6 3 8 50 7 20 5 10 3 3 10 25 5 10 25 2 8 10 15 30 15 1.0 10 30 0.1 1 1 1 2 1 - 1 1 1 - 1 5 1 1 1 5 2 1 1 - 1 - 1 2 - 1 0 . 8 3 1 0 . 1 the effect of size (cf. Brochmann 1987). All morphological characters were subsequently stan- dardised by “ranging” (i.e., each character was scaled between 0 and 1, Gower 1971), and morphological dissimilarities between plants were calculated as average Manhattan distances (Rohlf 1990). Kendall’s z correlations (Kendall 1938) were calculated for the two first PCO axes, DCA axis 1, the morphological characters, and the ecological parameters. Ecology Saxifraga oppositifolia showed a wide ecological amplitude in the Kongsfjorden area. The species occurred in a wide range of habitats varying from dry, wind-exposed, almost unvegetated ridges and slopes, via snow-protected, damp, densely vege- tated heaths, to moist, densely vegetated snow- beds. A total of 38 vascular plant species was recorded in the 50 vegetation samples analysed (Table 2). Most samples were classified into three vegetation types, i.e., Dryas octopetala heaths and ridges (12 samples), Luzula arcuata-lichen heaths (12 samples), and S. oppositifolia-lichen heaths (24 samples; Tables 2 and 3). One sample was classified as talus slope vegetation, and one as Deschampsia alpina snowbed. The percentage cover of S. oppositifolia in each vegetation sample varied between 2% and 50%, and the cover of the species varied considerably within as well as between the main vegetation types (Table 2). The most conspicuously cushion-forming plants of S. oppositifolia were found at the driest, uppermost Dryas ridges. The Dryas octopetala vegetation type was, on average, the driest and most wind-exposed one among the main vegeta- tion types (mean humidity 1.8; mean wind exposure 2.7), but the degree of humidity and wind exposure varied within this vegetation type along a gradient from ridges to slopes and heaths. The total vegetation cover (mean 52%) as well as the cover of bryophytes and lichens were, on \D 0 3 p R T ab le 3 . E co lo gi ca l p ar am et er s sc or ed in 5 0 ve ge ta tio n sa m pl es w ith S ax ifr ag a op po si tif oo lia , c la ss if ie d in to v eg et at io n ty pe s fo llo w in g B ra ttb ak k (1 98 1) . b 0 > H um id ity W in d ex po su re Sl op e (a ng le o f T ot al v eg et at io n V as cu la r pl an t N o. o f 1 - d ry , 4 - m oi st 1 - p ro te ct ed , 3 - e xp os ed in cl in at io n) co ve r (% ) co ve r (% ) V eg et at io n ty pe sa m pl es m ea n sd m in m ax m ea n sd rn in m ax m ea n sd m in rn ax m ea n sd rn in m ax m ea n sd m in m ax D ry as o ct op et al a ri dg eh ea th 12 1. 8 0. 87 1 3 2. 7 0. 49 2 3 14 16 .2 0 60 52 35 .5 5 95 26 25 .8 3 70 L uz ul a ar cu at a - li ch en h ea th 12 2. 3 0. 49 2 3 2. 0 0. 00 2 2 1 1. 8 0 5 82 23 .0 35 10 0 22 17 .3 4 60 Sa xi fr ag a op po si tif ol ia - li ch en h ea th 24 2. 8 0. 82 2 4 1. 8 0. 51 1 3 9 10 .2 0 40 83 20 .9 15 10 0 21 15 .7 3 55 T al us s lo pe v eg et at io n 1 3. 0 - - - 2. 0 - - - o - - - 5 0 - - - z - - - D es ch am ps ia a lp in a sn ow be d 1 4. 0 - - - 2. 0 - - - o - - - 9 5 - - - 4 5 - - - V eg et at io n ty pe ~ ~ C ry pt og am B ry op hy te L ic he n St on e B ar e gr ou nd N o. o f co ve r (% ) co ve r (% ) co ve r (% ) co ve r (% ) co ve r (a ) sa m pl es m ea n sd rn in rn ax m ea n sd rn in m ax m ea n sd rn in m ax m ea n sd m in m ax m ea n sd mi n m ax ~ D ty as o ct op et al a ri dg eh ea th 12 31 28 .5 1 85 21 25 .9 1 75 11 13 .5 1 45 36 35 .9 1 95 14 25 .8 0 92 L uz ul a ar cu at a - lic he n he at h 12 75 21 .7 34 98 60 20 .8 33 95 15 13 .9 1 50 13 16 .6 0 40 5 9. 1 0 30 Sa xi fr ag a op po si tif ol ia - li ch en h ea th 24 71 21 .5 10 99 44 22 .1 10 98 27 20 .6 1 90 10 13 .6 0 50 8 16 .1 0 70 T al us s lo pe v eg et at io n 1 z - - - 5 - - - 2 0 - - - 2 5 - - - z - - - D es ch am ps ia a lp in a sn ow be d 1 5 0 - - - 3 5 - - - 1 5 - - - 5 - - - o - - - The Purple Saxifrage 99 Table 4. Descriptive statistics of macroscopic morphological characters analysed in Saxifraga oppositifolia, based on the median of five measurements per plant. See Table 1 for explanation of characters. Character No. of plants Mean S D Min Max lnternode length (mm) Leaf width (mm) Leaf length (mm) Leaf index No. of leaf cilia No. of leaf cilia pr. mm Petal width (mm) Petal length (mm) Petal index Petal apex (1 - rounded, 3 - acute) Overlap of petals (1 - non-overlapping, 3 - overlapping) Flower shape (1 - flattened, 3 - urn-shaped) Flower diameter (mm) average, lowest in the Dryas vegetation type. The uppermost ridges were almost unvegetated (cover 5%), whereas the most humid Dryas heaths were densely vegetated (cover up to 95%). The vascular plant species most frequently observed in the Dryas vegetation type, except for the only dominating species Dryas octopetala and S. oppositijolia, were Bistorta vivipara, Cerastium arcticum, and Salk polaris (Table 2). The most prostrate, creeping plants of S . oppositijolia were found in the most humid parts of the lichen heaths. The two lichen heath vegetation types were fairly similar except for quantitative differences in the occurrences of some species, in particular S . oppositijolia, Luzula arcuata, Bistorta vivipara, and Oxyria digyna, and in relative importance of bryophytes versus lichens (Tables 2 and 3). The lichen heaths were, on the average, densely vegetated (mean cover 82-83%), but had about the same cover of vascular plants as the Dryas vegetation type 146 146 146 146 145 145 132 132 132 131 132 132 132 2.8 1.9 3.3 0.57 8.6 2.6 4.2 6.8 0.60 1.6 2.0 1.9 11.3 1.66 0.26 0.45 0.06 2.0 0.58 1.15 1.07 0.11 0.5 0.6 0.6 2.29 0.3 8.1 1.3 2.7 2.1 4.5 0.44 0.73 4 15 1.3 4.2 1.5 8.6 4.6 10.1 0.33 0.85 1 3 1 3 1 3 7 18 (means 21-22% and 26%, respectively). The cryptogam layer covered, on the average, 71-75% in the lichen heaths, versus 31% in the Dryas vegetation. The DCA analysis based on the floristic composition of the 50 vegetation samples did not result in an interpretable pattern, probably caused by the species-poor samples with con- siderable overlap in species content, and this analysis is therefore not shown. Among the ecological parameters, only cryptogam cover and bryophyte cover were significantly (p < 0.05) correlated with DCA axis 1. Because no other parameters (e.g., humidity or total vegetation cover) were significantly correlated with DCA axis 1, these two parameters did not per se provide any meaningful ecological interpretation of the DCA axis. DCA axis 1 may nevertheless represent a humidity gradient. This might have been verified by a more accurate measure of humidity. Several ecological parameters were, however, Fig. 1 . Variation in internode length in Saxifraga oppositifolia from the Kongsfjorden area in Svalbard. n = 150 plants. 100 A . K. Brysting et al. 0.1 c.l v) 3 0.0 - - -- - - o cushion-formed A intermediate -0.1 -. A A - prostrate A PCO axis 1 significantly correlated with individual morpho- logical traits of S . oppositifolia as well as the PCO axes (see below). Macromorphology The plants analysed showed conspicuous variation in most individual morphological characters (Table 4). However, all quantitative characters showed a unimodal distribution, most distinctly so, and it was thus not possible to separate the material into two or more groups based on any individual characters. Internode length, the character most closely associated with the growth form of the plants, varied from 0.3 mm (i.e., 3.3 leaves per mm of the stem) in the most cushion-formed plants to 8.1 mm (i.e., 0.1 leaf per mm of the stem) in the most prostrate plants, with an average of 2.8 mm for the 146 plants measured for this character. As for all morphological characters, the internode length showed a unimodal distribution (Fig. 1). The leaves varied considerably in size and shape. The leaf length varied from 2.1 mm to 4.5 mm (mean 3.3 mm), the leaf width varied from 1.3 mm to 2.7 mm (mean 1.9 mm), and the leaf shape varied from narrowly to widely elliptic (mean leaf index 0.57). Leaf ciliation varied less than expected. All plants had ciliated leaves, varying from 1.3 to 4.2 cilia per mm of the leaf margin. The variation in floral characters was particu- larly conspicuous. The petals varied nearly six- fold in width, from very narrow (1.5 mm) to very broad (8.6 mm), and more than two-fold in length (4.6-10.1 mm). The petals also varied consider- Fig. 2. PCO analysis (axes 1 and 2) of 146 plants of Saxifiaga oppositifolia from the Kongsfjorden area in Svalbard based on seven morphological characters. The classification into growth forms (cushion-formed, intermediate, or prostrate) was not included prior to the analysis, but subsequently assigned to the plants subjected to the PCO analysis. ably in shape, from very long, narrow, and non- overlapping (petal index 0.33) to nearly as broad as long and overlapping (petal index 0.85), and with an apex varying from rounded to acute. The shape of open flowers varied from flattened to distinctly urn-shaped. Nor did the PCO analysis result in any distinct grouping of the plants, although the analysis revealed that there was a mainly one-dimensional, simple structure in the total morphological data set (Fig. 2). The first PCO axis had a high eigenvalue (1.40) and accounted for 78.3% of the total variance in the morphological data, and the first two PCO axes accounted for 95% of the variance. The result of the NMDS analysis (stress 0.22; not shown) was largely similar to that of the PCO analysis. Most of the morphological characters were signifiantly correlated with PCO axis 1 (see also below; cf. Table 5). Internode length was most strongly correlated, and negatively so (7) = -0.73, p < 0.001), indicating that the most prostrate plants (i.e., those with long internodes) had low values along PCO axis 1 and the most cushion- formed plants (i.e., those with short internodes) had high values along PCO axis 1. The tentative field classification of the plants after growth form as typically prostrate, “intermediate”, or typically cushion-formed was superimposed on the PCO diagram (Fig. 2; note that this character was not included prior to the PCO analysis). Most of the plants classified as typically prostrate had low values (-0.3 to 0) along PCO axis 1, whereas most of the plants classified as typically cushion- formed had higher values (0 to 0.2), indicating that this subjective growth-form classification The Purple Saxifrage 101 agreed with the exact measurements of internode lengths. Notably, neither the growth-form classi- fication nor the internode analysis resulted in distinct groups in the material; rather, the scores were distributed continuously along the two PCO axes. B 8. Pollen size and stainability The pollen diameter in individual plants varied from 20.0 to 32.2 pm, and the mean pollen diameter for the 36 plants analysed was 25.0 pm (sd 2.6). The pollen diameter showed a distinct unimodal distribution (Fig. 3). All plants except one had more than 80% stainable pollen (mean 92%, sd 8%, min 60%, max 95%; n = 36 plants). Correlations among ecological traits, morphological traits, and PCO axes Several of the individual traits were significantly correlated (Table 5). Among the ecological traits, humidity, vegetation cover, vascular plant cover, cryptogam cover, and bryophyte cover were positively correlated (p < 0.01). These traits were negatively correlated (p < 0.01) with stone cover, bare ground cover, and wind exposure. Among the morphological characters, leaf width, flower diameter, petal length, petal width, and petal index were positively correlated (p < 0.01). Internode length was positively correlated (p < 0.05) with humidity, growth form, flower diameter, petal length, petal width, petal index, leaf length, and leaf width. The number of leaf cilia per mm, petal apex, and wind exposure were negatively correlated with internode length. Growth form was positively correlated (p < 0.05) with flower diameter, leaf length, leaf width, leaf index, vascular plant cover, lichen cover, and humidity, and negatively correlated 0, < 0.05) with number of leaf cilia per mm, petal apex, and wind exposure. Internode length and number of leaf cilia per mm were most strongly correlated with PCO axis 1 (Table 5 ) . Internode length was negatively correlated, and the number of leaf cilia per mm was positively correlated with PCO axis 1. Floral characters and leaf width were negatively corre- lated with PCO axis 1. Except for internode length, similar, but weaker correlations were found for PCO axis 2. The significant correlations found for PCO axis 2 were leaf width (T) = -0.13, 0.01 < p < 0.05), number of leaf cilia pr. mm 102 A . K. Bysting et al. Fig. 3. Variation in pollen diameter in Saxifraga oppositifolia from the Kongsfjorden area in Svalbard. n = 36 plants (360 grains). (z) = 0.24, p < 0.001), petal width (z) = -0.32, p < 0.001), petal index (t) = -0.28, p < 0.001), and flower diameter (z) = -0.25, p < 0.001). The median pollen diameter was positively correlated with internode length (z) = 0.53, p < 0.001) and growth form (z)=O.53, p < 0.001), and negatively correlated with PCO axis 1 (t) = -0.58, p < 0.001). Thus, the prostrate plants had larger pollen grains, longer internodes, fewer leaf cilia, larger leaves, larger flowers, larger, broader, and more rounded petals and occurred in wetter and less windy habitats than the cushion-formed plants. Discussion Neither the individual morphological characters nor the PCO analysis indicate that it is possible to divide the plant material into two distinct groups, as would be expected if the two growth forms of s. oppositifolia deserve status as separate taxa. Most characters vary continuously and show unimodal distributions. The growth form was not included as a character prior to the PCO analysis, but when the plants subsequently were labeled according to the tentative growth form classification, the three groups, prostrate, “inter- mediates” and cushion-formed, showed a contin- uous, overlapping distribution along PCO axis 1. In particular, the unimodal distribution of inter- node length (Fig. l), a character which could be quantified exactly and which is closely associated with growth form, and the strong correlation between this character and PCO axis 1 (Table 5 ) strongly support the hypothesis that S. oppositifo- lia is represented by a single, but variable, taxon in Svalbard. Several morphological traits besides internode length have been suggested to be diagnostic characters distinguishing between the two pro- posed subspecies of s. oppositifolia (Crawford et al. 1993; Elven & Elvebakk 1996; R ~ n n i n g 1996): The prostrate growth form has been suggested to be correlated with occurrence of marginal leaf cilia, open flowers, and narrow, non-overlapping petals, whereas the cushion-like growth form has been suggested to be correlated with absence of leaf cilia, urn-shaped flowers, and broad, over- lapping petals. Our results demonstrate, however, (1) that the number of cilia is negatively correlated with internode length (i.e., prostrate plants are more sparsely ciliated than cushion- formed plants), (2) that petal width is positively correlated with internode length (Lee, prostrate plants have broader petals than cushion-formed plants), and (3) that neither flower shape nor overlap of petals are correlated with internode length. We have thus shown that none of the cilia and floral characters separate the two growth forms in the way proposed by Crawford et al. (1993), Elven & Elvebakk (1996), and R ~ n n i n g (1996). It is well known that the pollen size may be useful to discriminate between different ploidal levels in plants, in particular between closely related populations at low ploidal levels (Ehren- dorfer 1964; Ugborogho 1973; Brochmann 1992a). The stainable (and thus probably fully fertile) pollen grains and the unimodal distribution of pollen diameter observed in this study suggest that the plants analysed of S. oppositifolia represent a single ploidal level (diploid or The Purple Saxifrage 103 In conclusion, the present study clearly demon- strates the high levels of morphological and ecological variation within S. oppositifolia in Svalbard, but gives no support for maintaining the two growth forms as the separate subspecies reptans and pulvinata as proposed by Ronning (1996). It seems rather most reasonable to refer all of the populations of S. oppositifolia occurring in Svalbard to the widespread and variable, typical subspecies oppositifolia, in agreement with Webb & Gornall (1989), who note that this subspecies is “usually fairly lax in growth (except in very exposed situations)” (p. 106). The differentiation within S. oppositifolia in Svalbard is clearly of no taxonomic significance, but probably caused by local ecoclinal evolution occurring independently in different sites and geographic areas. The close geographic proximity of the extreme forms, a considerable overlap in flowering time, and the existence of many, fully fertile plants that cannot easily be classified as either prostrate or cushion- formed suggest that the plants at each site represent single panmictic populations rather than different taxa evolved by sympatric sub-specia- tion. Genetic diversity and ecological versatility within species seem to be common phenomena in many arctic plants, without necessarily being of taxonomic significance (McGraw 1995; Murray 1995). Such variation is, however, of significance in a broader context of arctic conservation biology and the potential impact of global warming on arctic vegetation. Achawledgements - This study was carried out as a project during the graduate student course in Arctic Biology (BIO 390) arranged in 1995 by the Department of Biology, University of Oslo, in Ny hesund, Svalbard. We thank the Norwegian Polar Institute !or logistic support and use of their research facilities in Ny Alesund. The study was partly supported by grant 101183/720 to CB from the Terrestrial Ecological Research Program in Svalbard, the Norwegian Research Council. tetraploid). Because cushion-formed, “intermedi- ate”, and prostrate plants were equally well represented among the investigated plants, the growth forms are thus probably not reproductively isolated by genetic barriers. Notably, the plants with intermediate morphology had fully stainable pollen. This result would not have been expected if the intermediates were triploid hybrids between a diploid and a tetraploid taxon. However, without chromosome counts it is not possible to determine if the plants analysed are diploids or tetraploids. The significant correlations obtained between some of the ecological parameters and internode length as well as growth form indicate that the cushion-formed plants are most common on dry, wind-exposed ridges, whereas the prostrate plants usually are found in less exposed and more damp habitats. This is in agreement with the observa- tions by Crawford et al. (1993), RBnning (1996), and Elven & Elvebakk (1996). The PCO and correlation analyses demonstrate, however, that the morphological variation in S. oppositifolia is continuous along local ecological gradients. Plants in the driest and most wind-exposed habitats have shorter internodes, smaller and more densely ciliated leaves, and smaller flowers than plants growing in successively more damp habitats. These characters probably all reflect response to drought. Similar, genetically based clinal variation along local humidity gradients has recently been demonstrated in the genus Franke- nia, suggested to be caused by one or a few genetic factors that control organ size and are subjected to differential selection with respect to drought tolerance (Brochmann et al. 1995). It is likely that the observed morphological and ecological variation in S. oppositifolia is geneti- cally based, but this cannot be determined with certainty without molecular genetic data, manip- ulation of environmental factors, or reciprocal transplant experiments. However, we observed that prostrate and cushion-formed plants occa- sionally grow in close proximity within the same, apparently homogeneous habitat. This observation may suggest that the growth form is genetically determined, and that the different genotypes sometimes are able to disperse and establish in suboptimal habitats. Notably, similar conspicuous differences in morphology and photosynthetic capacity that are genetically based have been demonstrated along similar environmental gradi- ents for Dryas octopetala in Alaska (McGraw & Antonovics 1983; McGraw 1995a). References Abbott, R. J., Chapman, H. M., Crawford, R. M. M. & Forbes, D. G . 1995: Molecular diversity and derivations of popula- tions of Silene acaulis and Saxifiaga oppositifolia from the high Arctic and more southerly latitudes. Molec. E d . 4, 199-207. Andersson, G . & Hesselman, H. 1900: Bidrag ti1 kannedomen om Spetsbergens och Beeren Eilands karlvaxtflora. Bih. K. Svenska Vet.-Akad. Handl. 26, Ill, I, 1-88. Bauert, M. R. 1996: Genetic diversity and ecotypic differentia- tion in arctic and alpine populations of Polygonurn viviparum. Arct. Alp. Res. 28, l W 1 9 5 . 104 A . K . Brysting et al. Braak, C. J. F. ter 1987: CANOCO- F O R T M p r o g r a m for canonical communiry ordination by (partial)(detrended)(ca- nonical) correspondence analysis, principal components analysis and redundancy analysis (version 2.1). TNO Inst. appl. Comp. Sci., Stat. Dept. Wageningen, Wageningen. Brattbakk, I. 1981: Br@ggerhalv@ya, Svalbard. Vegetasjonskart 1:10.000. K. Norske Vitensk. Selsk. Mus. Bot. Avd., Trondheim. Evaluation of some methods for hybrid ed by hybridization in Argyranthemum (Asteraceae). Nord. J . Bot. 7, 609-630. Brochmann, C. 1992a: Pollen and seed morphology of Nordic Draba (Brassicaceae): phylogenetic and ecological implications. Nord. J . Bot. 12, 657473. Brochmann, C. 1992b: Polyploid evolution in arctic-alpine Draba (Brassicaceae). Sommerf: Suppl. 4 , 1-37. Brochmann, C. 1995a: Arktisk biodiversitet er underestimert: ny innsikt fra evolusjons- og molekylrergenetiske studier. Univ. Trondheim Vitensk.mus. Rapp. Bot. Ser. 1995 3, 18-20. Brochmann, C. 1995b Conservation biology and polyploid evolution in the Arctic: insights from molecular and reproductive studies. Abstract, Vl International Symposium IOPB, “Variation and Evolution in Arctic and Alpine Plants”, Tromso, Norway. Brochmann, C., Lobin, W., Sunding, P. & Stabbetorp, 0. 1995: Parallel ecoclinal evolution and taxonomy of Frankenia (Frankeniaceae) in the Cape Verde Islands, W Africa. Nord. J. Bot. 15, 603423. Brochmann, C., Soltis, D. E. & Soltis, P. S. 1992a: Electrophoretic relationships and phylogeny of Nordic polyploids in Draba (Brassicaceae). Plant Syst. Evol. 182, 35-70. Brochmann, C., Soltis, P. S. & Soltis, D. E. 1992b: Multiple origins of the octoploid Scandinavian endemic Draba cacuminum: electrophoretic and morphological evidence. Nord. J . Bot. 12, 257-272. Brochmann, C., Soltis, P. S. & Soltis, D. E. 1 9 9 2 ~ Recurrent formation and polyphyly of Nordic polyploids in Draba (Brassicaceae). Amer. J . Bot. 79, 673-688. Brysting, A. K., Elven, R. & Nordal, I. In press: The hypothesis of hybrid origin of Poa jemtlandica supported by morpho- metric and isozyme data. Nord. J. Bot. Chapman, H. M. 1995: P. 10 in Stace, C. A. (ed.): IOPB chromosome data 9. IOPB Newsletter 24. Crawford, R. M. M. & Abbott, R. J. 1994: Pre-adaptation of arctic plants to climate change. Botanica Acta 107, 271-278. Crawford, R. M. M., Chapman, H. M., Abbott, R. J. & Balfour, J . 1993: Potential impact of climatic warming on arctic vegetation. Flora 188, 367-381. Crawford, R. M. M., Chapman, H. M. & Hodge, H. 1994: Anoxia tolerance in high arctic vegetation. Arct. Alp. Res. 26, 308-312. Ehrendorfer, F. 1964. Notizen zur Cytotaxonomie und Evolu- tion der Gattung Artemisia. &err. Bot. Zeitschr. 111, 84-142. Eilertsen, O., 0kland, R. H., Okland, T. & Pedersen, 0. 1990: Data manipulation and gradient length estimation in DCA ordination. J . Veg. Sci. 1 , 261-270. Elven, R. 1994. 6. utg. av Lid & Lid, Norskflora. Det Norske Samlaget, Oslo. Elven, R. & Elvebakk, A. 1 9 9 6 Part 1. Vascular plants. Pp. 9-53 in Elvebakk, A. & Prestrud, P. (eds.): A catalogue of Svalbard plants, fungi, algae, and cyanobacteria. Norsk Polarinst. Skr. 198. Engler, H. G. A. & Itmscher, E. 1919: Saxifragaceae: Saxifraga. Das Pflanzenreich 69, 449-709. Flovik, K. 1940: Chromosome numbers and polyploidy within the flora of Spitzbergen. Hereditas 26, 4 3 W O . Gabrielsen, T. M. 1996: RAPD variation and phylogeography of Nordic populations of Saxifiaga oppositifolia. Cand. scient. thesis, Univ. of Oslo. Gabrielsen, T. M., Bachmann, K., Jakobsen, K. S. & Brochmann, C. 1995: RAPD variation in an arctic-alpine, mainly diploid outbreeder: Saxifraga oppositifolia. Abstract, VI International Symposium IOPB, ‘ ‘Variation and Evolu- tion in Arctic and Alpine Plants”, Tromsg, Norway. Gower, J. C. 1966: Some distance properties of latent root and vector methods used i n multivariate analysis. Biometrika 53, Gower, 1. C. 1967: Multivariate analysis and multidimensional geometry. Statistician 17, 13-28. Gower, J. C. 1971: A general coefficient of similarity and some of its properties. Biometrics 27, 857-874. Hagen, A,, Schjoll, O., Brochmann, C., Elven, R., Nordal, I. & Borgen, L. 1995: Genetic variation in the polyploids Cerastium alpinum and C . arcticum (Caryophyllaceae): morphology, isozymes and RAPDs. Abstract, V1 Interna- tional Symposium IOPB, ‘ ‘Variation and Evolution in Arctic and Alpine Plants”, Tromse, Norway. Haraldsen, K. B. & Wesenberg, J. 1993: Population genetic analyses of an amphi-Atlantic species: Lychnis alpina (Caryophyllaceae). Nord. J. Bot. 13, 377-387. Hayek, A. 1905: Monographische Studien iiber die Gattung Saxifraga, 1. Die Sektion Porphyrion Tausch. Denkschr. kaiserl. Akad. Wissensch., Wien, 77, 61 1-709. Hedberg, 0. 1967: Chromosome numbers of vascular plants from arctic and subarctic North America. Arkiv Bot. 11, 6 , Hill, M. 0. & Gauch, H. G. 1980: Detrended correspondence analysis: an improved ordination technique. Vegetatio 42, 47-58. Hulth, E. & Fries, M. 1986: Atlas of North European vascular plants north of the Tropic of Cancer. Koeltz, Konigstein. Johnson, A. W. & Packer, J. G. 1968: Chromosome numbers in the flora of Ogotoruk Creek, N. W. Alaska. Bot. Not. 121, 403-456. Kendall, M. G. 1938: A new measure of rank correlation. Biometrika 30, 81-94. Kruskal, J. B. 1964: Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis. Psychometrika 29, 1-27. Liive, A. & E v e , D. 1951: Studies on the origin of the Icelandic flora. 11. Saxifragaceae. Svensk Bot. Tidskr. 45, 368-399. Liive, A. & Liive, D. 1956: Cytotaxonomical conspectus of the Icelandic flora. Acta Horli Gotob. 20, 65-291. E v e , A. & E v e , D. 1975: Cytotaxonomical atlas of the arctic flora. Cramer, Vaduz. McGraw, J. B. 1995a: Patterns and causes of genetic diversity in arctic plants. Pp. 33-43 in Chapin 111, F. S. & Korner, C. (eds.): Arctic and Alpine Biodiversity: Patterns, Causes and Ecosystem Consequences. Springer, Berlin. 325-338. 309-326. The Purple Saxifiage 105 McGraw, J. B. 1995b: Prevalence and significance of ecotypic variation in arctic and alpine plants. Abstract, VI Interna- tional Symposium IOPB, ‘ ‘Variation and Evolution in Arctic and Alpine Plants”, Tromso, Norway. McGraw, J. B. & Antonovics, J. 1983: Experimental ecology of Dryas octopetala ecotypes. 1. Ecotypic differentiation and life cycle stages of selection. J . Ecol. 71, 879-897. Murray, D. F. 1995: Causes of arctic plant diversity: origin and evolution. Pp. 21-32 in Chapin 111, F. S. & Komer, C. (eds.): Arctic and Alpine Biodiversity: Patterns, Causes and Ecosystem Consequences. Springer, Berlin. Nordal, I. & Iversen, A. P. 1993: Mictic and monomorphic versus parthenogenetic and polymorphic-a comparison of two Scandinavian mountain grasses. Opera Bot. 121, 19-27. Norusis, M. J . 1993: SPSS for Windows. Base systems user’s guide. Release 6.0. SPSS Inc., Chicago. Radford, A. E., Dickison, W. C., Massey, J. R. & Bell, C. R. 1974: Vascularplant systematics. Harper & Row, New York. Rohlf, F. J. 1990: NTSYS-pc. Numerical taxonomy and multivariate analysis system. Version 1.60. Exeter Software, New York. Ronning, 0. I. 1996: Svalbards Flora. Ed. 3. Norsk Polarinstitutt, Oslo. Skovsted, A. 1934: Cytological studies in the tribe Saxifrageae. Dansk Bot. Arkiv 8(5): 1-52. Stenstrom, M. & Molau, U. 1992: Reproductive ecology of Saxifraga oppositifolia: phenology, mating system, and reproductive success. Arct. Alp. Res. 24, 337-343. Tollefsrud, M. M., Bachmann, K., Jakobsen, K.S. & Broch- mann, C. 1995: RAPD variation in an arctic-alpine, polyploid inbreeder: Saxifraga cespitosa. Abstract, VI International Symposium IOPB, ”Variation and Evolution in Arctic and Alpine Plants”, Tromse, Norway. Ugborogho, R. E. 1973: North American Cerastium arvense L. I . Cytology. Cytologia 38, 559-566. Webb, D. A. & Gomall, R. J. 1989: A manual of Saxifrages and their cultivation. Timber Press, Portland.