Classification of dry grassland vegetation in ... - Wiley Online Library

Journal of Vegetation Science 11: 585-596, 2000 © IAVS; Opulus Press Uppsala. Printed in Sweden

- Classification of dry grassland vegetation in Denmark -

585

Classification of dry grassland vegetation in Denmark Bruun, Hans Henrik1 & Ejrnæs, Rasmus1,2 1

Botanical Institute, University of Copenhagen, Øster Farimagsgade 2 D, DK-1353 Copenhagen K, Denmark; Corresponding author; Fax +4535322321; E-mail [email protected]; 2Present address: Department of Landscape Ecology, National Environmental Research Institute, Grenåvej 14, DK-8410 Rønde, Denmark; E-mail [email protected]

Abstract. A hierarchic classification of Danish semi-natural grassland vegetation on well-drained soils is presented. TWINSPAN was used for clustering of 614 samples of grassland vegetation showing floristic gradients and turnover in species composition in more dimensions. The optimal hierarchic level of clustering was determined by indicator species analysis. The classification was interpreted in terms of variables relating to abiotic environment and vegetation structure and to major ecoclines previously identified by gradient analysis. The 12 final clusters were compared to syntaxa of formal phytosociology and to communities in the British Vegetation Classification. Criteria for achieving floristically homogeneous clusters without sacrificing the ecological interpretability and validity of the clusters in time and along geographical gradients are discussed.

Keywords: Acidic; Calcareous; Continuum; Europe; FestucoBrometea; Koelerio-Corynephoretea; Nardo-Callunetea; Plant community type; Trifolio-Geranietea sanguinei.

Nomenclature: Tutin et al. (1964-1980).

designations of vegetation types, e.g. dune, heath, and forest, are abstractions implying differences in vegetation structure, edaphic and climatic conditions, and not necessarily a certain composition of species. An approach with narrow clusters demands the erection of equal clusters for vegetation apparently stable for centuries as well as for successional stages only existing for a few years, whereas an approach based solely on measured environmental variables demands an a priori selection of variables, which may not be the most relevant. Vegetation classification in Denmark has been statistical and ecological rather than pure floristical, following the tradition of Warming (1895) and Raunkiær (1909). Danish vegetation has only rarely been described according to the Central European tradition (Lawesson & Krienke 1996), neither have multivariate methods often been used (Lawesson 1998). The aim of the present study was to classify Danish grassland vegetation on well drained soils based on a compromise between cluster homogeneity and reflection of environmental conditions. High priority was given to consistency of clusters in time (succession) and space (within Denmark).

Introduction Faced with the diverse array of approaches to the classification of vegetation (see Whittaker 1978), a fundamental question arises: why classify vegetation? Shimwell (1971) stated that classification is an inherent and intuitive act of the human race. There are, however, strong differences among approaches, ranging from those who believe to identify superorganismal communities, to those who classify for practical reasons. Parallel contrasts exist with respect to the methods applied; from ideal true cluster seeking (e.g. Noy-Meir & Whittaker 1977) to pragmatic continuum segmentation. A related fundamental question of vegetation classification is whether to optimize within-cluster floristic homogeneity or maximum reflection of the environment. The Braun-Blanquet approach (Westhoff & van der Maarel 1978) seems to favour very narrow communities (associations) defined mainly in terms of floristic composition, whereas the majority of vernacular

Material and Methods Floristic and environmental data The floristic material treated here was also subjected to gradient analysis (Ejrnæs & Bruun 2000). The data comprise a compilation of all published and unpublished data satisfying two criteria: 1. Sampling following the Raunkiær method (Raunkiær 1909; 10 subsamples of 0.1 m2 within a homogeneous stand, species values are frequency in subplots); 2. Predominantly herbaceous vegetation from old (never cultivated, or if so, then centuries ago) and unfertilized sites. The studies selected are Böcher (1945, 1946, 1947, 1957, unpubl.), Böcher et al. (1946, unpubl.), Mortensen (1953, 1955), Fredskild (1954), Ejrnæs (unpubl.), Ejrnæs & Bruun (1995) and Wind & Ballegaard (1996). The number of species per sample plot, the number of species per

586

Bruun, H.H. & Ejrnæs, R.

subplot (mean of 10), and the same measures for annual species only, are used in the description of clusters. Topsoil pHH O (ca. 0 – 10 cm) was measured. Inclina2 tion and aspect of slope was used to calculate the potential incident radiation on March 1 (range 0 – 2). A number of variables were calibrated by weighted averaging (Jongman et al. 1987: 83) of autecological information from Ellenberg et al. (1992) and three categories from FIBS (Grime et al. 1996): lightc (from Ellenberg’s L-value), temperaturec (from Ellenberg’s T-value), continentalityc (from Ellenberg’s K-value), waterc (from Ellenberg’s F-value), Nitrogenc (from Ellenberg’s Nvalue), competitionc (from Grime’s C-value), stressc (Grime’s S-value) and disturbancec (Grime’s R-value). In all cases the superscript c denotes calibrated variables. Refer to Ejrnæs & Bruun (2000) for a thorough description of data compilation, and a full list of explanatory variables. Outliers were identified by an iterative process until DCA-axis stability (see Ejrnæs & Bruun (2000) for details). The six most outlying samples were omitted prior to classification to avoid distortions resulting in exceptions overruling general patterns. The resulting data set consisted of 387 species in 614 samples. Classification and Indicator Species Analysis Two-Way-INdicator-SPecies-ANalysis (TWINSPAN; Hill 1979) was chosen for the classification procedure mainly for two reasons: 1. It is based on ordination, i.e. Correspondence Analysis, which should result in a classification based on floristic composition, and yet interpretable in terms of environmental conditions. 2. Due to the noise-reducing effect of ordination, the divisive algorithm should be less noise sensitive than agglomerative ones (Gauch 1982). This was desirable since a major part of the floristic variation in the data could be considered noise (see Ejrnæs & Bruun 2000 for a discussion). TWINSPAN in the version of Oksanen & Minchin (1997) with superstrict convergence criteria was run by means of the package PC-ORD 3.0 (McCune & Mefford 1997). The maximum possible number of cut levels was used, resulting in a somewhat coarser abundance scale than in the original data but still an acceptable resolution (0.1-0.2, 0.3-0.4, 0.5-0.6, 0.7-0.8 and 0.9-1.0). The maximum number of indicators per division was set to 15, as the data set was expected to be sufficiently large and the samples sufficiently species-rich. Altering these options did not markedly affect the output. Divisive clustering was allowed to proceed to level 5 (25 end-clusters of potentially 32). In order to select the appropriate stop level this preliminary tree was evaluated by Indicator Species Analysis (McCune & Mefford

1997), as proposed by Dufrêne & Legendre (1997). For each species in a data set Dufrêne & Legendre (1997) computed an indicator value for a given cluster of samples by multiplying its relative abundance (average abundance of the species in the cluster over the average abundance of the species in all samples) with its relative frequency (fraction of samples in the cluster where the species is present), and they tested the significance of the indicator value by a permutation test. Dufrêne & Legendre (1997) argued that the sum of significant indicator values over all species will increase with cluster homogeneity (many small clusters), and with species fidelity to clusters (a few large clusters). The sum of significant indicator values should reach a maximum somewhere between 1 and N clusters. In a dichotomous tree, the number of clusters is doubled in every level with respect to the previous level. The non-hierarchic kmeans-clustering method applied by Dufrêne & Legendre (1997) can provide a series of clusterings with the number of clusters increasing by one. Although we used the method to determine the optimal hierarchical level rather than the optimal number of clusters, the method proved useful. In the case of our dichotomous TWINSPAN tree, the sum of significant indicator values (1000 Monte Carlo permutations) was maximal at level 2 (four clusters, results not shown). We decided to make two changes in the tree. First, there were no significant indicators for the subdivision of cluster 4 (sand grassland), and we decided not to divide this cluster further. Second, since we did not want to treat slightly different successional stages as different communities, the subdivision of cluster 7 (within xeric grassland) was also avoided. These refinements yielded a tree similar to that shown in Fig. 1, except that all ‘new’ clusters in level 4 were further subdivided to level 5. In a new species indicator analysis the sum of significant indicator values was maximal at level 4 in the tree (Fig. 1), and it was decided not to proceed to level 5. Naming of clusters The naming of clusters constituted a specific challenge. We decided to combine the names of two indicator species of high constancy that, preferably individually but alternatively in combination, characterize and distinguish one community from the others. The first species was chosen to indicate the relationship of the cluster to the major Danish grassland types appearing after the first two TWINSPAN divisions: Artemisia campestris indicates xeric grassland, Dactylis glomerata indicates calcareous grassland, Rumex acetosa indicates acidic grassland and Carex arenaria indicates sand grassland.


587

Results A structure with 12 final clusters emerged from the clustering procedure (Fig. 1). The first dichotomy divided the most important coenocline (DCA1; 5.19 SD units) which runs from open vegetation on south-facing slopes with calcareous soil to humid leached acidic grasslands (Ejrnæs & Bruun 2000). The two dichotomies at level 2 both divided the second most important coenocline (DCA2; 4.71 SD units) which mainly reflects productivity – from relatively fertile grasslands on clay, limestone and occasionally diluvial sand with good water supply, to infertile dry grasslands on marine or eolian sand (Ejrnæs & Bruun 2000). The division of cluster 1 also followed DCA1 to some extent. See App. 1 for the significant indicator species for the divisions.

Fig. 1. TWINSPAN tree with 12 end-clusters. The number of samples is shown next to each node, and the sum of significant indicator values used to assess the optimal level is shown to the left at the four levels. The similar sum for level 5 (not shown) is 3011.

Main types

Cluster 3. Xeric grasslands (181 samples) This is a relatively open vegetation – often rich in annual species – occurring on drought-prone soil. The soil pH ranges from approximately 5 to well above 8 (mean 7.1). Apart from the species indicative of both clusters 3 and 4, e.g. Dactylis glomerata, Galium verum, Avenula pratensis, character species for this cluster are Artemisia campestris, Bromus hordeaceus, Arenaria serpyllifolia, Sedum acre, Convolvulus arvensis and Phleum phleoides.

Four main clusters emerged from the first two divisions, and they correspond to the poles of the two main ecoclines and are centred towards the four corners of the almost rhombic configuration of sample points in the primary coenoplane. At this level the sum of significant indicator values is close to its maximum (93%), which means that these clusters are relatively homogeneous and well suited as main grassland vegetation types. See Fig. 2 for the optima and variation of the main types and the end-clusters with respect to the primary coenoplane.

•

• •

• •

4

• • •

•

•

•

3 • •

• • • •

• •

•

•

•

•

• •

1

• •• • •• • •

•

•

Fig. 2. Contour-plot of the 12 grassland clusters projected onto a scatter-plot of the sample plots on the first two DCA axes. Contour-lines indicate the 0.3 probability-line of each cluster. The contour-lines are extrapolated from the predicted probabilities derived from a multinomial log-linear regression of cluster-membership as a function of DCA1 (reflecting oceanity) and DCA2 (reflecting fertility).

•

•

15 17 18 19 20 21

• •

•

• • •

• • • •• •• • • • • •• • • • • • • • • • • • • • • • • • •• • • • • •• • • • • •• •• • • • • • • •• •• • • • • • • • • •• • • • • • • • • • • • • •• • • • • • • • • • •• • •• • • • • •• •• • •• •• • • •• • • • • • • • • • • • • • • • • •• • • • • • • •• • • • • •• • • •• • • • • • • • • • • • •• • •• • • • • • •• • •• • • •• • • • • • • • •• •• • • • • • • • • • • •• • • • • • • • • •• •• • • • • • • • •• • •• • • • • • • • • • • • • • • •• • • • • • • •• • • • • •• • • •• • • • •• • • • • • • • • • •• • •• • • • • •• • • • • •• • • • • • • •• • • • • • • • • • • • • • • •• • • • • • • • •• •• • • • •• • • •• • • • • • • • • • • • • • • • • • • • • • • • • • • •• • • • • • • • • • • • • • • • • • •• • • • • • • •• •• •• • • • • • • • • • • • • • • • • • • • • • • • • • • • • •• • • • • • • • • • • •• • • •• • • • • • • • • • • • • • • •• • • • • • • • • • • • • • • •• • • • • • • •• • • • • •

2

••

•

•

•

• •

•

•

•

• • • •

• •

•

22 23 24 25 26 27

• • •

•

• •

• •

•

• • •

•

• •

•

•

• • •

•

• • • • •

•

• •

• • • •

•

• •

• • •

•

•

• •

•

•

• •

• •

• •

• • •

•

•

• •

•

•

• •

0

•

0

1

2 3 DCA-1 [Oceanity]

4

5

588


Cluster 4. Calcareous grasslands (181 samples) This is a closed species-rich vegetation on limestone soil or till containing calcite CaCO3 (pH ranging from below 6 to 8.5 with a mean of 7.3). It occurs less often on south-sloping terrain, and then the soil has a significant content of clay. Besides Dactylis glomerata, Galium verum and Avenula pratensis, all of which are also indicative of cluster 3, the most prominent indicator species are Primula veris, Cirsium acaule, Carex flacca, Viola hirta and Leucanthemum vulgare. Cluster 5. Acidic grasslands (160 samples) This is a closed vegetation on north-facing slopes and other sites which are not exposed to drought, and where the soil is acid and often leached: pH from below 4 up to 7 (mean 5.2). Beside the species indicative of both cluster 5 and 6 – Luzula campestris, Agrostis capillaris and Anthoxanthum odoratum, the strongest indicators are Rumex acetosa, Danthonia decumbens, Veronica officinalis and Potentilla erecta. Cluster 6. Sand-grasslands (48 samples) This cluster was not divided further and it is described below as cluster 27 Carex arenaria-Corynephorus canescens grassland.

End-clusters The frequency-dominant species for each end-cluster appear from Table 1a. These are species with a frequency > 0.95 in more than 50% of the samples in the cluster. These species need not be dominant in terms of biomass, as even species with a small stature occurring scattered in a species-dense vegetation can reach high frequency counts. Significant indicator species are shown in App. 1 along with their indicator values and constancy in the end-clusters. The number of species per plot and the number of species per subplot are shown in Fig. 3. For each end-cluster the local environment is described below by the most significant explanatory variables (see also Fig. 3 and Table 1b). Cluster 15. Artemisia campestris-Poa compressa grassland (103 samples) The vegetation is relatively species-rich, and rich in annuals. The local environment is characterised by very high pH, a very high irradiation, and unstable soil. Drought is the major limit to productivity, and ceased grazing leads only to a very slow succession towards scrub, often dominated by Prunus spinosa. The community occurs predominantly in the precipitation-poor regions of Denmark: along the coasts of the Great Belt and the eastern Baltic Sea.

26. Plantago maritima -Solidago virgaurea grassland

69,0

93,3

27. Carex arenaria -Corynephorus canescens grassland

25. Carex arenaria -Festuca rubra grassland

59,8

24. Rumex acetosa -Agrostis capillaris grassland

51,6

23. Rumex acetosa -Deschampsia flexuosa grassland

22. Dactylis glomerata -Koeleria pyramidata grassland

20. Dactylis glomerata -Arrhenatherum elatius grassland

19. Dactylis glomerata -Lolium perenne grassland

18. Artemisia campestris-Carex arenaria grassland

21. Dactylis glomerata -Ranunculus bulbosus grassland

a. Agrostis capillaris Avenula pratensis Deschampsia flexuosa Festuca ovina Festuca rubra Galium verum Plantago maritima

17. Artemisia campestris-Trifolium striatum grassland

15. Artemisia campestris-Poa compressa grassland

Table 1. a. Dominant species in the 12 end-clusters. Percentage of samples in a cluster where the species occur with a frequency > 0.95 (only percentages above 50 shown). b. Percentages of crude soil types within the end-clusters.

66,7 51,0

86,5 59,3 76,9

8,0

51,9 53,3

Number of samples

103

51

27

13

50

31

87

37

123

29

15

48

b. Clay Sandy clay Diluvial sand Eolian sand Alluvial sand or shingle Limestone

11 15 72 1 1 1

4 2 83 0 8 2

0 5 27 15 30 14

54 15 15 0 0 15

22 13 51 0 2 11

33 27 33 0 0 7

10 1 38 0 0 51

0 0 97 0 3 0

9 11 68 1 11 0

0 0 17 59 7 17

0 7 0 7 0 87

0 0 19 23 46 11


589

Fig. 3. Species richness and local environment in the 12 endclusters. Number of species per plot, number of species per subplot, percentage of annuals in the number of species per subplot, soil pH, incident radiation on March 1, and angle of slope. The black boxes are the 3. quartiles below and above the median (white bar), the thin horizontal bars are the 99% confidence limits and the vertical bars each represent an outlier.

Cluster 17. Artemisia campestris-Trifolium striatum grassland (51 samples) The vegetation is rich in species, particularly annuals. A moderate soil pH (6.8 ± 0.4) combined with high irradiation indicates non-calcareous but unleached soils. The community occurs predominantly in the precipitation-poor regions of Denmark, but with a wider range than cluster 15. Cluster 18. Artemisia campestris-Carex arenaria grassland (27 samples) The vegetation is relatively species-rich and moderately rich in annuals. The community mainly occurs on

south-facing slopes of post-glacial marine sand and shingle deposits. The substrate may be decalcified, but never acidified. In Northwest Jylland it mainly occurs on dry fixed dunes. The vegetation shares some typical species with cluster 27 Carex arenaria-Corynephorus canescens grassland. A typical feature of the cluster is the large proportion of continentally distributed species. This continental element in the flora of calcareous coastal dune grassland, e.g. Artemisia campestris, Geranium sanguineum and Hippophae rhamnoides, is unique in Europe (Böcher 1954).

590


Cluster 19. Dactylis glomerata-Lolium perenne grassland (13 samples) Species richness is variable but may be very high, despite the prevalence of potentially dominant species. The soil is clayey, either moraine clay or Eocene (plastic) clay, the pH is very high, and moderate irradiation and often seeping subsoil water make sites less susceptible to drought, and potentially productive. We have inadequate information regarding grazing in the old data sets, but we hypothesise that grazing controls the competitive species and allows for the co-occurrence of less competitive species. Despite the occurrence of species typical of improved pastures, the studied areas have not been fertilised and contain numerous species indicative of absence of fertilization and cultivation. Cluster 20. Dactylis glomerata-Arrhenatherum elatius grassland (50 samples) The vegetation is not species poor, but has a low number of species per subplot. This is probably a reflection of the larger plant modular size. The vegetation is supposedly ungrazed or lightly grazed, but the steepness of the slopes makes scrub encroachment take place at a very slow rate. The soil pH is high, and irradiation moderate. This community occurs all over eastern Denmark.

case of the cluster. The remaining part of cluster 22 and the entire cluster 21, which also shares the diagnostic species of the Mesobromion, cannot easily be assigned to Solidagini-Helictotrichetum. Cluster 23 Rumex acetosa-Deschampsia flexuosa grassland (37 samples) The vegetation is species-poor. The soil is acid, and often podzolized. This type occurs predominantly on level terrain to north-facing slopes with low irradiation. It can be seen as a transition to the dwarf-shrub heath, though chamaephytes play a subordinate role. The apparent concentration of the cluster on Mols peninsula, eastern Jylland, is probably a sampling artefact. This grassland type is underrepresented in Böcher’s samples; it is known to be common, especially in regions with high precipitation. Cluster 24. Rumex acetosa-Agrostis capillaris grassland (123 samples) The only significant indicator species for this the largest cluster is Stellaria graminea. This apparent paradox arises as the cluster represents the middle of the primary coenoplane – or ‘typical’ grassland vegetation. The vegetation is species-rich – and occurs mainly on level to north-facing ground. It is related to Rumex acetosa-Deschampsia flexuosa grassland but the soil is less acidified and the number of species higher.

Cluster 21. Dactylis glomerata-Ranunculus bulbosus grassland (31 samples) The vegetation is very species-rich. The soil pH is high. The soil is sandy till deposits, of which some contain calcite and some are decalcified with pH just below 7. Irradiation is relatively low, the sites often being north-facing slopes. The cluster is somewhat transitional to acidic grassland. This community is mainly found in eastern and northern Jutland.

Cluster 25. Carex arenaria-Festuca rubra grassland (29 samples) The vegetation is relatively species-poor. The substrate is eolian sand, often previously calcareous but now decalcified. The community mainly occurs in northwestern Jylland, but similar vegetation, without Rosa pimpinellifolia, is found in the precipitation-poor region.

Cluster 22. Dactylis glomerata-Koeleria pyramidata grassland (87 samples) The vegetation is very species-rich. The soil pH is high and irradiation moderate. Within Denmark all of the significant indicator species occur almost exclusively in northern Jutland, where also the geographic centre of the cluster is. Vegetation resembling the core of the community occurs scattered in the precipitationpoor region. The substrate in northern Jutland and on the isle of Møn is Cretaceous chalk on slopes, more or less covered by moraine deposits. In the rest of the country the substrate is calcareous, and often sandy, till. Willems et al. (1981) investigated Danish vegetation similar to this community with traditional phytosociological methods, and described the association Solidagini-Helictotrichetum. Their relevés could surely be ascribed to cluster 22, but may be considered a special

Cluster 26. Plantago maritima-Solidago virgaurea grassland (15 samples) Some species seem to have specialized ecological races more or less restricted to this community, e.g. Dactylis glomerata (Böcher 1961), Solidago virgaurea, Equisetum hyemale, Anthyllis vulneraria (Lampinen 1990) and Hypochoeris maculata. The endemic Euphrasia dunensis occurs exclusively in this community. The vegetation is species-rich. It is found in northwesternmost Jylland on steep north-facing slopes of limestone adjacent to extensive dune areas. Eolian sand may overlay the limestone and the sand may contain calcite. This makes the transition to the sand grassland of the dunes gradual, and results in a very variable soil pH (6.7 ± 1.0). Typical of the cluster is also a very low irradiation. The cluster appeared somewhat misplaced by the

- Classification of dry grassland vegetation in Denmark first division, where it was separated from the calcareous grassland (it shares many species with Dactylis glomerata-Koeleria pyramidata grassland) and fused with acidic grassland, but the odd mixture of species gives it a transitional status. The cluster has obvious ecological similarity to the English chalk heath (Grubb et al. 1969) and limestone heath (Etherington 1981) although the particular species are different. Cluster 27. Carex arenaria-Corynephorus canescens grassland (48 samples) The vegetation is species-poor. It occurs on postglacial marine deposits all over the area north of the isostatic line where the sea has retreated during the last 5000 yr. The largest areas of such deposits are found in northern Jutland. The soil pH is moderately low and irradiation is moderate. The coarse-textured substrate is very susceptible to summer drought, which is likely to constrain productivity and species composition. The cluster includes a transition towards the vegetation of the ‘white’ dune – a few samples having species such as Ammophila arenaria and Lathyrus japonicus, indicating previous sand drift.

Discussion Our results indicate that the recently proposed symmetric indicator species analysis (Dufrêne & Legendre 1997) may be used in hierarchic clustering of vascular plant frequency data. Symmetric indicator species indicate the true cluster membership of a sample, and species presence is likely in all samples of that cluster, while asymmetric indicators only have the former feature. According to Dufrêne & Legendre (1997) the limit between asymmetric and symmetric indication corresponds to an indicator value of 55 (55% of perfect indication). This value is not supported in our study: Carex arenaria, Luzula campestris and Agrostis capillaris occur rather frequently and abundantly outside the cluster for which they should be symmetric indicators (App. 1). Gradient analysis (Ejrnæs & Bruun 2000) showed that the floristic variation in these dry grasslands was continuous and with an almost complete change in species composition from one axis end to the other along at least two axes. The change in species composition was judged from the length of DCA axes scaled in SD-units, assuming the average species to appear, reach its optimum and disappear within 6 SD units (Minchin 1987). It should be taken into account that the applied abundance scale favours presence over abundance, which has been reported to shorten DCA axes (Eilertsen et al. 1990). No natural clusters were obvious in a plot of

591

samples in the primary coenoplane (Fig. 2), except for a weak disjunction between Cluster 3: acidic grassland and Cluster 4: sand grassland. Despite this continuous variation, the clustering procedure yielded a number of floristically and ecologically well-defined clusters. This leads us to conclude that vegetation classification often, if not always, is continuum segmentation (van Groenewoud 1992 on TWINSPAN), but nevertheless feasible. We furthermore find reason to stress its usefulness, both to scientific research and to nature management and conservation. In order to achieve a classification which acknowledges that sites with a similar environment can have a somewhat differing species composition, some withincluster floristic heterogeneity should be allowed for. The breadth of our clusters seems to be suitable to allow for some stochastic variation in species occurrences under a given set of environmental conditions. The clusters are floristically defined, but ecologically interpretable. Vegetation succession implies events of local extinction and colonization, further raising the element of stochasticity, and reducing the predictability of species composition in terms of the local environment. This relationship suffers even further if the vegetation is under strong influence by man. Our clusters have deliberately been made sufficiently broad to include short-lived stages characterized by reversible overgrowing by coarse grasses, forbs and scattered shrubs. Comparison or assignment of Scandinavian vegetation ‘entities’ to the Central European syntaxonomic system has often proved a difficult task (e.g. Hallberg 1971; Diekmann 1997). For historical reasons South Scandinavia is only superficially treated in the syntaxonomic literature (Lawesson & Krienke 1996; Diekmann 1997). The climate and the available pool of grassland plant species are considerably different from the Central European. This makes assignment of Danish vegetation types, especially to higher-order categories from those areas, difficult. In the present case, differences in sampling approaches furthermore introduce some noise to the comparison: our data are frequencies, which have a non-linear relationship to ramet density, but hardly any relationship to cover, which forms the base of the above-mentioned approaches. The major advantage of frequency data is that they do not change much during the season, assuring reproducible sampling. A comparison of the end-clusters with phytosociological classifications from adjacent countries leaves at least two clusters without obvious equivalents. Our Carex arenaria-Festuca rubra grassland (cluster 25) has affinities to dune heathlands like the HieracioEmpetretum, but it also contains numerous and abundant calcicolous species like Geranium sanguineum.

592


Plantago maritima-Solidago virgaurea grassland (cluster 26) has some affinity to Gentianello-Pimpinelletum saxifragae described from South-Norwegian calcareous fixed dunes, due to constant species such as Festuca rubra and Pimpinella saxifraga and rare subordinate species such as Draba incana. But this association does not show the mixture of calcifuges and calcicoles characteristic of Cluster 26. We find it likely that these communities are unique to Denmark or southern Scandinavia. Our dry and/or calcareous grasslands (Cluster 1) are characterised by many (sub-)continentally distributed species. This is to some degree an expression of the fact that a dry and warm local climate produces a high base saturation of the soil. However, calcareous substrates appear to be preferred by continentally distributed taxa also in the absence of a dry local climate. We have anecdotal evidence that some species which are calcicolous in sub-oceanic Northwest Jylland tend to be less pH-limited in the dry Store Bælt region, e.g. Avenula pratensis, Filipendula vulgaris, Geranium sanguineum, Origanum vulgare and Helianthemum nummularium. This phenomenon has previously been reported from comparative studies along a latitudinal gradient in Europe (Etherington 1975: 278). Our Plantago maritima-Solidago virgaurea grassland (Cluster 26) has the highest calibrated Ellenbergvalue of continentality (adaptation to large diurnal and annual amplitudes in the local climate). This may be considered a paradox since this community is found in a comparatively oceanic climate. Sterner (1922) however stressed in his comprehensive treatment of the continental flora of Sweden, that continental species at their northwesternmost limits of distribution respond as much to soil conditions as to local climate. Gradient analysis showed the vegetation to respond to a set of ecological gradients in a complex manner that could not be predicted by the single gradients (Ejrnæs & Bruun 2000). A direct classification of samples constrained by the two most important explanatory variables, soil pH and relative irradiation, resulting in the four combinations of acidic/basic vs. moist/dry, would only poorly reproduce the classification presented here. Similarly, the present classification in combination with the gradient analysis of the same data indicates that a simple division according to soil types would also sacrifice the floristic homogeneity of clusters. In conclusion, we believe that an ‘indirect’ approach to classification is the best way to simultaneously optimize precision of the vegetation description and ecological validity.

Acknowledgements. We thank Henning Adsersen, Anders Michelsen, Jonas Lawesson and two referees for valuable comments on the manuscript.

References Böcher, T.W. 1945. Beiträge zur Pflanzengeographie und Ökologie dänischer Vegetation. II. Über die Waldsaumund Graskrautgesellschaften trockener und halbtrockener Böden der Insel Seeland mit besonderer Berücksichtigung der Strandabhänge und Strandebenen. Kgl. Dan. Vidensk. Selsk., Biol. Skr. 4 (1): 1-168. Böcher, T.W. 1946. Grassland communities of ‘Høje Møn’. Bot. Tidsskr. 48: 1-45. (In Danish.) Böcher, T.W. 1947. Festuca polesica Zapal., its chromosome number and occurrence in Denmark. Bot. Notiser 1947(4): 353-360. Böcher, T.W. 1954. Studies on the European calcareous fixed dune communities. Vegetatio 5-6: 562-570. Böcher, T.W. 1957. Soil and vegetation on slopes on the south coast of Bornholm. Natural History Society of Bornholm, Bornholm. (In Danish.) Böcher, T.W. 1961. Experimental and cytological studies on plant species. VI: Dactylis glomerata and Anthoxanthum odoratum. Bot. Tidsskr. 56: 314-335. Böcher, T.W., Christensen, T. & Christiansen, M.S. 1946. Slope and dune vegetation of north Jutland. I. Himmerland. Kgl. Dan. Vidensk. Selsk., Biol. Skr. 4(3): 1-78. Diekmann, M. 1997. The differentiation of alliances in South Sweden. Folia Geobot. Phytotax. 32: 193-205. Dufrêne, M. & Legendre, P. 1997. Species assemblages and indicator species: the need for a flexible asymmetrical approach. Ecol. Monogr. 67: 345-366. Eilertsen, O., Økland, R.H., Økland, T. & Pedersen, O. 1990. Data manipulation and gradient length estimation in DCA ordination. J. Veg. Sci. 1: 261-270. Ejrnæs, R. & Bruun, H.H. 1995. Prediction of grassland quality for environmental management. J. Environ. Manage. 43: 171-183. Ejrnæs, R. & Bruun, H.H. 2000. Gradient analysis of dry grassland vegetation in Denmark. J. Veg. Sci. 11: 573-584. Ellenberg, H, Weber, H.E., Düll R., Wirth, V., Werner, W., Paulißen, D. 1992. Zeigerwerte von Pflanzen in Mitteleuropa. Datenbank. Scr. Geobot. 18: 1-248. Etherington, J.R. 1975. Environment and plant ecology. J. Wiley, London. Etherington, J.R. 1981. Limestone heaths in south-west Britain: their soils and the maintenance of their calcicolecalcifuge mixtures. J. Ecol. 69: 277-294. Fredskild, B. 1954. Ecological and floristic studies on dry sand slopes in western Sealand. Oikos 5: 205-235. Grime, J.P., Hodgson, J. & Hunt, R. 1996. FIBS: An extended version of the electronic comparative plant ecology. NERC, University of Sheffield. Grubb, P.J., Green, H.E. & Merrifield, R.C.J. 1969. The ecology of chalk heath: its relevance to the calcicolecalcifuge and the soil acidification problems. J. Ecol. 57: 175-212. Hallberg, H.P. 1971. Vegetation auf den Schalenablagerungen in Bohuslän, Schweden. Acta Phytogeogr. Suec. 56: 1136. Hill, M.O. 1979. TWINSPAN: A FORTRAN program for arranging multivariate data in an ordered two-way table

- Classification of dry grassland vegetation in Denmark by classification of the individuals and attributes. Cornell University, Ithaca, NY. Jongman, R.H.G., ter Braak, C.J.F. & van Tongeren, O.F.R. (eds.) 1987. Data analysis in community and landscape ecology. Pudoc, Wageningen. Lampinen, R. 1990. Anthyllis vulneraria subsp. danica subsp. nova (Fabaceae) from Denmark. Ann. Bot. Fenn. 27: 217220. Lawesson, J.E. 1998. Quantitative vegetation ecology in Denmark. J. Veg. Sci. 9: 891-896. Lawesson, J.E. & Krienke, T. 1996. Phytosociology in Denmark – A review. Ann. Bot. (Roma) 54: 23-30. McCune, B. & Mefford, M.J. 1997. PC-ORD. Multivariate analysis of ecological data, Version 3.0. MjM Software Design, Gleneden Beach, OR. Minchin, P. 1987. Simulation of multidimensional community patterns: towards a comprehensive model. Vegetatio 71: 145-156. Mortensen, M.B. 1953. Grassland communities on the slopes of Hindsholm. Bot. Tidsskr. 49: 239-259. (In Danish.) Mortensen, M.B. 1955. Grassland communities on slopes in northern Samsø. Bot. Tidsskr. 52: 85-112. (In Danish.) Noy-Meir, I. & Whittaker, R.H. 1977. Continuous multivariate methods in community analysis: some problems and developments. Vegetatio 33: 79-98. Økland, R.H. 1986. Rescaling of ecological gradients. I. Calculation of ecological distance between vegetation stands by means of their floristic composition. Nord. J. Bot. 6: 651-660. Oksanen, J. & Minchin, P.R. 1997. Instability of ordination results under changes in input data order: explanations and remedies. J. Veg. Sci. 8: 447-454. Raunkiær, C. 1909. Investigations and statistics of plant for-

593

mations. Bot. Tidsskr. 30: 20-132. (In Danish.) – Chapt.VI, Investigations and statistics of plant formations, in: C. Raunkiær (1934). Raunkiær, C. 1934. The life forms of plants and statistical plant geography, pp. 201-281. Oxford University Press, Oxford. Shimwell, D.W. 1971. The description and classification of vegetation. Sidgwick & Jackson, London. Sterner, R. 1922. The continental flora of South Sweden. Geogr. Ann. 4: 221-441. Tutin, T.G. et al. 1964-1980. Flora Europaea. Cambridge University Press, Cambridge. van Groenewoud, H. 1992. The robustness of Correspondence, Detrended Correspondence, and TWINSPAN analysis. J. Veg. Sci. 3: 239-246. Warming, E. 1895. Plant communities. Outline of the ecological plant geography. P.G. Philipsens Forlag, Copenhagen. (In Danish.) Westhoff, V. & van der Maarel, E. 1978. The Braun-Blanquet approach. In: Whittaker, R.H. (ed.) Classification of plant communities, pp. 287-399. Junk, The Hague. Whittaker, R.H. (ed.) 1978. Classification of plant communities. Junk, The Hague. Willems, J.H., van Delft, J.M.E. & de Rijke, M.J. 1981. Observations on North-West European limestone grassland communities: IV. Phytosociological notes on chalk grasslands in Denmark. Folia Geobot. Phytotax. 16: 391-406. Wind, P. & Ballegaard, T. 1996. Monitoring of dry grassland vegetation 1995. NERI work report no. 21. Ministry of the Environment, Copenhagen. (In Danish.)

Received 22 December 1997; Revision received 28 December 1998; Final revision received 3 February 2000; Accepted 17 February 2000. Coordinating Editor: L. Mucina.

For App. 1, see pp. 594-596.

594


Poa compressa Festuca trachyphylla Bromus sterilis Alyssum alyssoides Melilotus alba Papaver argemone Anchusa officinalis Bromus erectus Melampyrum arvense Lithospermum arvense Artemisia campestris Bromus hordeaceus coll. Arenaria serpyllifolia Sedum acre Convolvulus arvensis Phleum phleoides Potentilla argentea coll. Trifolium campestre Myosotis ramosissima Acininos arvensis Allium vineale Veronica arvensis Medicago minima Vicia hirsuta Echium vulgare Sedum telephium coll. Seseli libanotis Valerianella locusta Trifolium striatum Aira caryophyllea Erophila verna Saxifraga granulata Myosotis stricta Petrorhagia prolifera Arabidopsis thaliana Cerastium semidecandrum Trifolium arvense Helichrysum arenarium Vicia lathyroides Armeria maritima Dianthus deltoides Erodium cicutarium Lychnis viscaria Trifolium dubium Veronica verna Pulsatilla pratensis Thalictrum minus coll. Veronica spicata Silene otites Scleranthus perennis Dactylis glomerata coll. Galium verum Avenula pratensis Centaurea jacea Medicago lupulina Centaurea scabiosa Phleum pratense ssp. bertolonii Ononis repens Daucus carota Filipendula vulgaris Anthyllis vulneraria Silene nutans Hypericum perforatum Geranium sanguineum Origanum vulgare Fragaria viridis Scabiosa columbaria

35*** 19*** 10*** 8*** 6** 5** 4* 4* 4** 3* 57*** 49*** 42*** 41*** 34*** 30*** 24*** 23*** 19*** 18*** 14*** 14*** 13*** 11*** 7** 4* 4* 4* 35*** 21*** 20*** 15** 12* 10** 5* 41*** 35*** 18*** 14*** 13*** 12*** 7** 5* 5* 5** 49*** 14*** 12** 7* 6* 68*** 51*** 49*** 48*** 47*** 33*** 33*** 31*** 23*** 23*** 18*** 16*** 15*** 13*** 10*** 9*** 8***

55/47 67/24 100/10 76/10 100/6 84/6 84/5 40/7 55/7 100/3 33/73 20/56 19/56 16/53 45/48 29/37 19/25 14/27 27/27 40/33 38/16 17/18 33/13 21/13 39/11 30/7 42/7 20/8 3/4 6/2 11/13 5/12 26/8 32/8 13/5 13/40 9/17 14/14 7/8 2/7 1/2 1/1 6/3 2/1 2/1 2/6 2/2

14/88 10/90 6/42 12/47 7/39 23/41 13/40 12/44 11/27 6/21 6/16 20/21 29/32 6/13 26/11 12/10 31/17

7/18 11/12

26/15

2/4

0/3

15/7

9/11 6/2

1/3

26. Plantago maritima -Solidago virgaurea grassland 27. Carex arenaria -Corynephorus canescens grassland

IV max

22. Dactylis glomerata -Koeleria pyramidata grassland 23. Rumex acetosa -Deschampsia flexuosa grassland 24. Rumex acetosa -Agrostis capillaris grassland 25. Carex arenaria -Festuca rubra grassland

Species

15. Artemisia campestris-Poa compressa grassland 17. Artemisia campestris-Trifolium striatum grassland 18. Artemisia campestris-Carex arenaria grassland 19. Dactylis glomerata -Lolium perenne grassland 20. Dactylis glomerata -Arrhenatherum elatius grassland 21. Dactylis glomerata -Ranunculus bulbosus grassland

App. 1. Indicator species analysis of the TWINSPAN tree. For each species two figures are shown separated by a slash: its relative abundance in each cluster (average abundance in the cluster over the average abundance in all samples) and its relative frequency (fraction of samples in the cluster where the species is present), both expressed as percentages (the value 0/0 not shown). The first two columns show the maximum indicator value over all clusters and the statistical significance of it computed by 1000 MonteCarlo permutations (*** = P ≤ 0.001, ** = 0.001 ≤ P ≤ 0.01, * = 0.01 ≤ P ≤ 0.05, n.s. = non-significant). 163 species with non-significant indicator values and low absolute frequencies have been omitted. Allegedly symmetric indicators (IVmax > 55%) are shown in bold.

0/1 0/1

24/4 16/2 16/2 14/2

21/4

25/6

1/1 45/3

25/63 39/78 36/75 28/82 27/25 30/33 46/45 35/51 26/25 22/12 41/20 36/33 67/20 38/24 41/8 11/2 29/4 19/2 77/45 76/27 57/35 32/45 66/18 66/16 69/8 46/80 39/53 29/25 35/25 15/43 24/24 64/10 23/14 31/16 83/6 8/12 8/6 19/10 2/2 8/63 12/98 20/80 7/25 10/37 2/10 16/41 13/45 4/6 10/24 12/35 14/18 9/12 10/18 24/14 9/4

30/74 11/44 10/37 31/96 12/15 26/33 8/15 20/22 1/4 12/19 5/11

14/11

1/8 13/46 6/23

0/3 11/32 15/48 4/13

4/17 0/2 11/29 4/20

3/3 18/6 10/19 6/6

8/13

29/23

1/4 1/6 1/4 0/2 2/4 3/6 6/6 0/2 1/2 2/2 11/8 1/2

12/23

0/1

6/8

10/10

4/3

4/3 19/1

11/31 2/8 10/8 35/23

24/11

2/4 3/7 15/31

2/11 5/22 8/15 3/19 10/15 6/4 11/26

2/2 1/4 3/8 4/2 2/2

7/15 2/15 2/8

4/4 1/6 2/2

3/17

0/7 2/13 2/20

3/3

0/2 10/35

3/8

0/1

4/2 6/4 8/7 23/32 4/2

1/1 6/8 1/5

2/8

3/7 0/1

22/13

2/3 3/1

2/4 4/10 4/2

1/13 12/10 2/3

2/6 6/5 8/3

16/90 5/72 8/48 20/76 2/22 23/40 3/16 9/38 7/16 17/36 1/6 13/18 28/28 20/32 45/22 12/12 12/4

17/97 12/97 20/77 11/55 24/74 6/23 25/65 20/42 4/13 25/45 15/42 10/6 4/6 12/19

12/75 8/82 23/89 22/67 24/82 33/46 10/33 12/37 29/43 18/39 15/44 24/25 13/17 11/20 29/16 2/2 27/8

20/12

8/8 23/100 7/69 2/38 25/62 21/69 12/54 28/85 20/46 45/46 8/23

5/19 0/3

4/3 5/5

4/2 8/3 10/10 7/3 16/19 18/32

3/3 10/45 17/10

17/15

1/3 4/9 0/3 1/7 2/2 1/2 2/3 5/14 2/2

26/18

16/6 5/19 1/3

6/8

41/23 16/4

1/5

6/2 28/4 54/8 5/15

17/44 33/63 49/37 33/19 18/41 31/19 34/4 29/7 3/4 15/4 62/78 48/30 62/19 96/7 44/15 1/15 14/96 9/52 0/7 3/19

1/8 0/3

2/5 1/3 1/3 1/5

2/3

2/1 8/17 2/7 4/3 7/17 13/14 1/1 20/9 10/7 2/3 2/2

3/2 22/3

2/3

3/2

0/7 2/7

1/7 11/17

19/41 4/7

1/7

2/7 15/33 5/7

10/7 18/23 6/8 4/4 4/2 24/4

12/10 2/3

22/3

8/13 4/3

1/35 0/8

1/11

1/5 1/3

2/22 7/73 7/40 1/7 0/2 1/4 2/15 7/24 0/2 8/23 1/7 7/7 11/11 1/4 1/1 0/1

1/4 5/17 7/13 4/8 7/17 2/2

3/24 11/90 1/3 0/3 0/3

4/27 5/73 2/7

8/69 2/13

8/33 0/7 0/2 1/7 0/7

6/10

27/45

39/47 0/7

2/6 0/2 1/8

Allium oleraceum Prunus spinosa Geranium molle Helianthemum nummularium coll. Silene vulgaris ssp. vulgaris Phleum arenarium Bellis perennis Lolium perenne Elytrigia repens Potentilla reptans Lathyrus pratensis Carex hirta Ranunculus repens Agrimonia eupatoria Geranium dissectum Vicia sativa ssp. nigra Myosotis arvensis Equisetum telmateia Galium palustre Juncus inflexus Potentilla anserina Rubus sect. Rubus Sonchus arvensis Thlaspi caerulescens Capsella bursa-pastoris Cirsium vulgare Festuca arundinacea Rubus caesius Anthriscus sylvestris Picris hieracioides Tanacetum vulgare Tragopon pratensis coll. Arrhenatherum elatius Galium mollugo Heracleum sphondylium coll. Campanula persicifolia Hedera helix Vincetoxicum hirundinaria Brachypodium sylvaticum Clinopodium vulgare Stellaria holostea Primula veris Cirsium acaule Carex flacca Viola hirta Leucanthemum vulgare Taraxacum spec. Festuca pratensis Plantago media Senecio jacobaea Knautia arvensis Avenula pubescens Trifolium pratense Fragaria vesca Ranunculus acris Leontodon hispidus Tussilago farfara Inula salicina Ranunculus bulbosus Cynosurus cristatus Cerastium fontanum ssp. triviale Thymus pulegioides Trifolium repens Taraxacum sect. Erythrosperma Astragulus danicus Potentilla cinerea Briza media Carex caryophyllea Arabis hirsuta Succisa pratensis Euphrasia stricta Geum rivale Koeleria pyramidata Campanula glomerata Sanguisorba minor ssp. minor Polygala amarella Crepis praemorsa Hieracium lactucella Polygonatum multiflorum Brachypodium pinnatum

7*** 6** 4** 4* 3* 2* 40*** 32*** 28*** 27*** 24*** 23*** 23** 18*** 14** 14*** 9** 8* 8* 8* 8* 8* 8* 8* 6* 6* 14*** 14*** 8** 5** 5* 4* 20*** 17*** 17** 15** 11** 10** 6* 6* 6* 38*** 36*** 35*** 24*** 23*** 20*** 19*** 19*** 19*** 18** 16*** 15*** 12*** 12** 11*** 5** 2* 27*** 23*** 21*** 18*** 14** 11** 6* 6* 44*** 21*** 14*** 10** 5* 4* 35*** 20*** 12** 11** 9** 7* 6* 5*

21/15 45/12 31/6 17/4 32/7 28/4

13/8 10/4 30/8 4/2 5/2 25/4

2/3 8/23 8/11 0/1 2/3 2/2 6/10

5/10 16/33 2/6

18/19 35/2

20/22

6/1 6/4 4/3 6/2

9/7 21/8

1/4 4/7

7/11

1/2 2/2 0/2 5/2 2/8 1/2

4/4 6/2 15/7 19/3

4/4 1/2

3/6 5/18 8/18

2/4

9/26

2/11 5/11

2/1

3/1 4/28 0/1 1/5 15/33

13/23 15/31 13/31 17/15 23/8 19/77 31/15 9/8 28/23 12/31 17/23 20/31 24/8 15/23 16/8

12/55 2/6 4/16 13/24 3/6 16/16 16/4 11/6 3/12 12/27 7/8

5/26

5/4

18/4

2/2

7/8

12/7

2/1

3/2

1/2 4/3 13/1 1/4 3/12 14/9

64/62 51/62 45/62 49/54 52/46 60/38 74/31 40/46 93/15 36/38 61/15 100/8 100/8 100/8 100/8 100/8 100/8 100/8 83/8 78/8 36/23 18/8 9/8 21/8 56/8 3/8

1/4 1/2 0/2 5/8 4/22 2/2

14/8 29/12 2/2 17/8 38/4

8/3 37/10 16/3

12/4

3/4

7/2 2/1 21/17 8/6

0/1 0/1 0/1 3/6 3/3 2/7 2/4 2/4 13/19 4/14 3/7

49/15

4/7 4/4 7/11 1/4

2/7 3/11 2/4

15/54 42/54 19/38 25/38 24/31 28/8 5/31 1/8

16/24 12/26 15/18 7/8 14/12 16/22 5/8

33/29 37/48 5/10 14/26 7/6

1/1 4/2

2/2 3/5

29/8

9/4 4/2

36/1 0/2 1/2 5/11 9/7 3/2

10/8

2/5 4/9 5/9 1/5 2/2 15/13

28/29 7/3 2/3

7/10

2/6

1/2

10/11 4/1

7/6 13/3

5/2 10/3

6/3 3/6

12/3 8/14 21/3 10/5 22/13

4/10 5/3 1/3 1/3 10/3 1/3


595


15. Artemisia campestris-Poa compressa grassland 17. Artemisia campestris-Trifolium striatum grassland 18. Artemisia campestris-Carex arenaria grassland 19. Dactylis glomerata -Lolium perenne grassland 20. Dactylis glomerata -Arrhenatherum elatius grassland 21. Dactylis glomerata -Ranunculus bulbosus grassland


14/7

1/3

1/2

17/1 4/2 34/20 74/18 76/10 79/4 66/8 15/6 49/42 67/26 86/20 48/32 94/12 71/14 80/8 100/6 100/6 17/64 4/20 8/30 31/26 11/24 5/28 17/16 3/10 7/16 12/34 25/28 0/2 28/18 13/24 15/14 26/4 57/2 2/22 1/4 2/16 5/12 1/4 2/2

5/26 6/24 5/6 4/2

1/2 11/8 16/6

31/2

14/6

0/1 4/2 6/2

8/5 2/3

9/2 2/2 6/6

17/7 2/7

4/2 7/3

12/3 1/1

29/58 30/48 23/45 21/32 8/19 15/52 20/23 27/39 6/19 11/35 22/45 29/45 4/3 10/16 42/19 34/6 30/90 38/61 28/74 44/42 32/45 37/29 49/13 58/10 38/74 27/55 21/16 14/6 7/3 20/3 7/6 5/3 2/3

15/44 31/52 27/52 27/25 36/45 11/40 26/24 16/37 41/33 20/41 9/18 12/23 36/15 12/20 23/18 21/6 43/3 12/52 5/25 10/9 1/6 1/2

31/64 24/51 48/25 43/23 40/13 62/6 71/49 84/24 78/16 100/11 100/9 73/9 100/6 69/7

0/5

2/8 1/5 4/8 0/3 2/3

0/3

1/5

6/8

5/20 3/8 3/11 1/1 11/9 6/22 0/1 3/4 1/3 20/46 8/13 5/13 3/2 8/17 18/8

15/63 17/28 15/46 9/16 12/23 12/12 4/2 18/2 8/20 24/46 2/2 33/18 17/6 19/1

27/6

7/17 3/14

3/14 1/3 1/7 1/3 5/14 1/3 6/24 20/21

13/33 18/33 23/53 4/7 25/73

1/6

39/40 6/20 0/7 29/40

1/2 0/2

23/33

0/10

5/27

8/31

11/27

3/13

12/28

3/13

5/8 4/4

2/3

5/20

0/7

13/2

Festuca rubra Poa pratensis Plantago lanceolata Deschampsia flexuosa Galium saxatile Carex pilulifera Agrostis vinealis Genista anglica Holcus mollis Lathyrus montanus Anemone nemorosa Vaccinium myrtillus Equisetum pratense Hypericum pulchrum Nardus stricta Carex panicea Stellaria graminea Rumex acetosa Danthonia decumbens Veronica officinalis Potentilla erecta Trifolium medium Polygala vulgaris Equisetum arvense Hypericum maculatum Rosa pimpinellifolia Ammophilia arenaria Veronica chamaedrys Linaria vulgaris Polypodium vulgare Silene vulgaris ssp. maritima Calamagrostis epigeios Pimpinella saxifraga Lotus corniculatus Holcus lanatus Salix repens coll. Vicia cracca Hippophae rhamnoides Potentilla tabernaemontani Equisetum hiemale Cochlearia officinalis Plantago maritima Solidago virga-aurea Antennaria dioica Agrostis stolonifera Linum catharticum Senecio integrifolius Empetrum nigrum Prunella vulgaris Carlina vulgaris Galium sterneri Gentianella amarella Leontodon autumnalis Erigeron acre Draba incana Hypochoeris maculata Papaver dubium Botrychium lunaria Centaurium erythraea Omalotheca sylvatica Luzula campestris Agrostis capillaris Anthoxanthum odoratum Festuca ovina Hieracium pilosella Achillea millefolium Campanula rotundifolia Viola canina Hypochoeris radicata Calluna vulgaris Rumex acetosella Aira praecox Teesdalia nudicaulis Corynephorus canescens Carex arenaria Jasione montana Hieracium umbellatum Thymus serpyllum Number of samples

37n.s. 34n.s. 30n.s. 66*** 44*** 30*** 16*** 7* 19*** 17*** 15*** 8** 4* 4** 4* 3* 20*** 42*** 34*** 30*** 29*** 23*** 18*** 11** 6** 42*** 36*** 22*** 19*** 16*** 10*** 9** 36*** 23*** 17*** 16*** 16*** 12*** 8*** 5** 4* 56*** 50*** 37*** 36*** 35*** 34*** 29*** 29*** 27*** 27*** 26*** 22*** 21*** 13*** 13*** 12** 10** 6* 6* 59*** 58*** 47*** 42*** 42** 39** 39*** 34*** 30*** 28*** 24*** 18*** 4* 71*** 64*** 49*** 48*** 39***

3/41 7/61 4/42 0/1

6/65 8/73 10/75 0/2

5/59 9/59 5/52

13/92 8/92 13/77

8/76 10/66 4/48

13/97 12/77 18/77 1/3

9/12

3/2

0/1

10/85 4/39 7/63 1/3 1/1 1/1

1/2 28/10 1/2

2/17 0/1 0/1 1/2 0/1 2/4 1/1

5/6 5/22 3/12 0/2

0/4

1/10

14/8 9/42 2/8 0/2

28/31

13/20

29/23

5/8 14/4

3/4 0/2 2/2

1/1 3/4

1/4 1/2

4/2 3/19 1/7 0/1

3/2 4/33 4/20 1/4

0/2 3/2 1/1

2/31

19/12

2/4 13/19 2/7

12/4 2/30 7/33

14/31

18/24 1/4

5/3

5/10 14/55 13/32 5/13 1/3 13/19 25/23 13/16

6/19

2/31 3/31 8/15

22/12 7/46 2/14 9/10

6/52 7/45 10/13

1/8

9/22

6/13

1/2

19/6

4/4 1/4 47/4 1/4

2/11

0/2

9/16 2/4 1/10

2/6 1/3 13/6 24/52

0/2 3/6 1/6

26/45 12/19 2/19

1/4

9/13

4/7

5/2

2/3

10/4

9/2

0/2

1/8

0/2 0/1

9/15 6/15

3/6

13/5

2/4 4/11

2/2 2/2 9/8

1/1

1/1 5/18 5/13 3/6 2/3 4/11 8/10 0/1 6/2 1/1 2/8 1/1

2/1 12/75 16/60 4/1 18/30 4/1 7/7

5/10 8/23 1/3 9/7 26/57 25/26 3/2 13/33 23/38 32/61 19/11 4/5

3/59 8/68 0/3 72/92 85/51 80/38 55/30 49/14 69/27 11/11 51/19 20/5 46/8 10/3 43/3 59/5 6/3 18/59 13/32 7/14 29/30 5/11 1/3 9/8

7/77 10/75 16/87 20/42 13/11 20/13 22/23 36/5 30/18 61/23 43/16 80/9 54/3 90/5 57/5 41/2 64/31 28/70 33/51 26/40 43/36 32/51 44/35 16/20 34/10 2/2

4/8

11/32

32/8

8/2

3/32 0/3 0/3 4/3 3/8

3/1 7/50 8/37 16/23 4/2 8/20 3/2

2/5

29/1 0/1 2/9 7/5 0/1 3/9

7/3 8/18 3/5 10/17

9/2 7/3 14/1

12/97 16/100 10/59 3/7

7/55 11/65 8/31 12/67 11/86 9/76 3/22 0/2 3/10 0/2 19/41 31/31 34/10

6/48 5/52 7/22 9/52 8/81 3/26 3/26 1/7 1/7

1/6 4/6 1/3 1/6

2/18 15/20 0/2 7/24

9/26 1/4 3/4 14/19 26/74 22/30 6/22 26/74

103

51

27

2/38 5/31 2/15 1/8 7/54 11/77 2/23

1/6

1/3

1/8

1/10 0/2

3/13 1/13

4/16 13/45

13/84 17/81 7/38 14/73 6/57 4/32 8/62 8/24 7/14 56/49 29/49 16/14 31/5 2/3 9/32 1/3 12/32 0/3

13

50

31

87

37

4/8

2/22 4/28 5/32 2/22 1/16 7/64 8/60 6/10 0/2

17/84 11/77 17/61 17/81 13/90 12/90 12/84 13/26 5/6 1/3 3/6 7/10

3/28 1/11 1/8 12/68 9/75 8/62 16/84 7/10 2/1 2/3 0/1

2/3 11/45 4/28 35/52 20/14 4/7 4/7 23/17 28/7 77/55 54/66 40/55 60/31 57/28 100/10 54/17 23/79 17/59 20/34 47/17 38/38 47/21 33/17 53/10 24/7 16/45 16/38 7/7 7/7

29/14 0/3 5/10

28/21 1/1

3/10 13/10

3/2

18/7

11/1 23/89 28/94 33/89 21/88 14/89 15/92 14/79 26/47 37/48 24/36 23/33 37/36 23/7

15/100 7/40 12/53

4/63 2/15 0/4 3/6

9/4 15/2 1/2

7/2

1/10 1/14 2/10 1/15 8/72 6/62 1/14 0/1 0/2 1/2 1/5 1/2



Bruun, H.H. & Ejrnæs, R. 15. Artemisia campestris-Poa compressa grassland 17. Artemisia campestris-Trifolium striatum grassland 18. Artemisia campestris-Carex arenaria grassland 19. Dactylis glomerata -Lolium perenne grassland 20. Dactylis glomerata -Arrhenatherum elatius grassland 21. Dactylis glomerata -Ranunculus bulbosus grassland

596

11/59 11/62 6/45

3/4 7/20 26/33 21/20 5/13 15/20 2/7 17/7 14/13 3/13 1/7 24/7

28/87 28/67 35/33 42/20 13/27 35/7 16/13 47/7 65/87 58/87 80/47 67/53 44/80 72/47 54/53 44/67 51/53 41/67 77/33 47/47 64/33 64/20 47/27 86/13 74/13 91/7 89/7 10/67 5/33 9/33 5/33 12/93 9/67 20/87 10/20 14/13 4/7

0/2 2/6

1/6 28/38 1/2 10/4 3/2

3/17 6/21

3/4 7/4 1/2

13/27 2/4 3/2 2/4 1/2 3/2 7/10

2/6 4/2 7/8 6/6 27/2 4/8 6/2

7/20 1/4 5/20 3/11

2/7 17/79 8/14 7/31 8/34

29/73 18/60

4/42 2/23 3/17 6/44 8/65 1/13 4/25 18/44 10/19 8/19 14/29 4/8 8/4 82/73 32/92 49/60 30/69 22/69

123

29

15

48

5/41 16/97 9/62 12/38 16/24 4/7 2/7 2/3

1/7

4/27