Data collection was aimed at the acquisition of a complete inventory of all plant communities present in the research area, including the very rare ones, to create a reference data set for large scale vegetation maps. In other conditions, this requirement would have necessitated a large amount of random sampling points, many of which would have been inaccessible due to the rugged mountainous terrain. Thus, no special sampling design was applied, and sampling points were chosen arbitrarily. Requirements to be met were accessibility, homogeneity of species composition as well as of habitat quality, an even distribution of sampling plots over the whole research area and a coarse quantitative representativeness of relevé-counts and the area covered by each community. Relevé plots were 4x4 m in size, except for krummholz and forests (20x20 m). Data were collected according to Braun-Blanquet (1964) with seven cover-abundance values (r, +, 1, 2, 3, 4, 5). Moreover total vegetation cover, slope, aspect, altitude, soil type and-as far as possible-soil depth and soil horizon formation were recorded for each site.

The survey of Austrian plant communities (Mucina et al. 1993a, b, Grabherr and Mucina 1993) served as the basis for defining the units of the vegetation maps. Relevés were assigned to this classification as far as possible. Dominance of species, presence of floristic indicators and habitat-quality were used as criteria for identification of syntaxa. We used TWINSPAN (Hill 1979), a divisive hierarchical classification technique, for structuring the dataset.

Endemics and subendemics to be considered were selected with respect to Adler et al. (1994). From the whole set of 19 selected species, 14 are decisively restricted to the North-eastern Limestone Alps. Two of them may also be found on calcareous stands in the easternmost siliceous Alps (eastern parts of Niedere Tauern Mts.: Draba stellata, Pedicularis portenschlagii) and another two in parts of the Bavarian Alps (Berchtesgadener Alpen Mts.: Draba sauteri, Primula clusiana). Viola alpina is not an endemic or subendemic species in the strict sense but is disjunctly distributed in the Alps and the Carpathian mountains. It was taken into account because within the Alps it is restricted to the same regions as the real endemic species.

Generally, only communities represented by at least five records, and only species recorded in at least ten relevés, were included into statistical calculations.

To achieve constancy-tables, we calculated the percentage of relevés of a community, within which any one species was recorded. These constancies were summarized in five classes according to Braun-Blanquet (1964).

Regarding the syntaxonomic distribution of endemic species, we calculated their different performance in different communities by weighting their summarized constancies by their respective mean cover-abundance values giving r and + a weight of 1, 1 and 2 a weight of 2 and 3, 4 and 5 a weight of 3. The total range of values characterising the frequency of a species in a community therefore lies between 0 and 15.

To analyse the distribution of species along the altitudinal gradient we first assigned all relevés to distinct altitudinal belts of 50 m each beginning with 1400-1450 m up to 2150 to 2200 m. The overall abundance of each species in each altitudinal belt was then calculated. Overall abundance represents the summed abundance of a species in all relevés after transformation of the Braun-Blanquet-values into numerical ones (r = 1, + = 2, 1= 3 etc.). To avoid bias of data due to different relevé-counts, overall abundance of each species was divided by the sum of relevés per altitudinal belt.

Aib* = Sairb/Srb*100

with Aib* = relative overall abundance of species i in altitudinal belt b and air abundance of species i in relevé r.

Regression analysis was applied to detect associations between these relative overall abundances and altitude. We compared linear and second order polynomial regression to assess if a species performs best within or without the altitudinal range considered in this study. Thus we tried to circumvent the problem that the study area extends neither into the montane nor the upper alpine or nival belt.

To examine how altitude effects the contribution of endemic species to the overall species pool, the count of species and the count of endemics recorded in all relevés were calculated for the same altitudinal belts of 50 m.

In a similar way, the total set of relevés was subdivided with respect to total vegetation cover per plot. Five classes were arbitrarily selected (0-20%, 20-40%, 40-60%, 50-80%, 80-100% total vegetation cover) and each relevé was assigned to the appropriate class. The count of species and the count of endemics recorded in all relevés of the respective classes were calculated.

Regression analysis was performed with SPSS 7.5.

Nomenclature follows Adler et al. (1994) for flowering plants, Frahm and Frey (1992) for bryophytes, Wirth (1980) for lichens, Mucina et al. (1993 a, b) and Grabherr and Mucina (1993) for syntaxa.

Syntaxonomic analysis of the complete data set (964 relevés) revealed 73 plant communities. Details on their flora, ecology and distribution are given elsewhere (Greimler and Dirnböck 1996, Dirnböck and Greimler 1997, Dirnböck et al. 1998, 1999).

This total of 73 associations contains nine endemic plant communities. Their contributions to the overall vegetation cover differ considerably. Whereas the Drabo stellatae-Potentilletum clusianae or the Campanulo pullae-Achilleetum clusianae cover large areas, the Alchemilla anisiaca-community and the Oreochloa disticha-grasslands are restricted to small parts of the investigation area with a highly scattered distribution of these communities even within these regions (see Fig. 2).

To characterise these nine plant communities in some detail, Table 1 provides a survey of their species composition and Table 2 summarises some important ecological features.

The Drabo stellatae-Potentilletum clusianae is the predominating vegetation type of alpine rock-faces. Like any other type of alpine rock vegetation it is characterised by a very low total vegetation cover. Soils are poorly developed and cushion plants, being able to accumulate detritus and aerosols (see e.g. Körner 1999), are the dominating life-form. The community is dominated by Potentilla clusiana and Draba stellata is its character species. Unlike Draba stellata, Potentilla clusiana, though mainly distributed in the North-eastern Limestone Alps, is not an endemic or subendemic in the strict sense, because it occurs regularly in some parts of the Southeastern Calcareous Alps, too (Merxmüller 1954: Steiner Alpen Mts., Karnische Alpen Mts.). Moreover Carex firma and Festuca versicolor subsp. brachystachys, another endemic species of the region, are frequent in this community. Average plant species diversity is quite low (see Tab. 1 and 2).

The Campanulo pullae-Achilleetum clusianae association is the predominant type of calcareous snowbed-vegetation. The canopy covers usually less than 50% (see Tab. 2). The community grows on debris, typical of snowbed-associations of calcareous mountain ranges. Soils are poorly developed. Perennial herbs are the most important, among them Achillea clusiana and Campanula pulla, both endemics, are dominant. Some other typical plants of moist scree vegetation like Saxifraga stellaris, Ranunculus alpestris, Moehringia ciliata, or the endemic Soldanella austriaca, are the most frequent accompanying species (see Tab. 1).

The Campanulo pullae-Achilleetum atratae, another snowbed-community, is distinctly less frequent than the Campanulo pullae-Achilleetum clusianae. Habitat qualities of both communities are quite similar (see Tab. 2). However, the Campanulo pullae-Achilleetum atratae prefers higher altitudes, whereas the Campanulo pullae-Achilleetum clusianae is found down to the tree line. The species composition of both communities does not differ very much except for the fact that Achillea clusiana is replaced by Achillea atrata and Arabis caerulea is nearly restricted to the Campanulo pullae-Achilleetum atratae (see Tab. 1). Both associations may occasionally intermix. The syntaxonomic rank of their distinction may therefore be put up for discussion.

Initial grassland vegetation of steep and rocky slopes facing south. Total vegetation cover usually amounts to about 50% (see Tab. 2). The community takes an intermediate position between rock- and grassland vegetation. Festuca versicolor subsp. pallidula predominates, and most of the accompanying species belong to the typical flora of rock faces (Kernera saxatilis, Primula auricula, Trisetum alpestre, Carex mucronata) and debris (Athamanta cretensis and the endemic Galium meliodorum) (see Tab. 1). In fact, the community is distributed mainly below the tree line (Greimler and Mucina 1992, Grabherr et al. 1993). Hence, the sites recorded may not provide typical examples but are representing the upper fringes of the association.

Typical grassland vegetation of south facing avalanche paths and other habitats characterised by a similar combination of high radiation income and moisture. Moisture is caused mainly by specific soil properties. Soils of Helictotrichon-grasslands are often patchy, combining situations with dry and shallow soils and others with comparatively deep and loamy ones. This causes less drainage and improved water storage compared to typical rendsina soils of the related Seslerio-Caricetum sempervirentis.

The community is mainly distributed around the tree line. It rarely establishes in the alpine zone itself and may thus be a typical subalpine vegetation type. Its ecological and floristic position is exactly in between the well-known grassland communities Seslerio-Caricetum sempervirentis, preferring similarly warm but drier habitats, and Caricetum ferrrugineae, also bound to comparatively moist but usually less insolated stands. The predominance of Helictotrichon parlatorei, a large tussock graminoid, characterises not only the species composition but also the structure of the community (see Tab. 1).

A vegetation type predominated by Alchemilla anisiaca, another endemic species of the most northeastern Alps. The community is frequent in the western parts of Mount Hochschwab and was also reported from Gesäuse-Mts. (Greimler 1997), but it is missing on Mt. Rax, Mt. Schneeberg and Mt. Schneealpe. It usually covers only small areas, forming typical patchy mosaics with grasslands, especially with the Seslerio-Caricetum sempervirentis. Those vegetation patterns are controlled by microscale variation in relief and soil properties. The Alchemilla-community grows on deeper, partly loamy soils in flat depressions and shallow drains, where snow may accumulate, snow-melt is delayed, water storage is improved and a certain impact of nutrients may be expected due to microscale flow accumulation. There is some ecological resemblance to the patchy mosaics of Caricetum curvulae and Salicetum herbacea in siliceous mountains but the overall species composition of the Alchemilla anisiaca-community with Carex ferruginea, C. sempervirens, Luzula glabrata, Poa alpina, Soldanella alpina, Primula elatior, Homogyne discolor, Potentilla aurea, Campanula scheuchzeri and Helianthemum glabrum as the most important accompanying species indicates an assignment to the Caricion ferrugineae alliance. Besides the predominating Alchemilla anisiaca and a moderate constancy of Soldanella austriaca, endemic species are rare in this community (see Tab. 1).

Alchemilla anisiaca was also reported to characterise nutrient poor pastures in the region of Mt. Dachstein (Pignatti-Wikus 1959).

Grasslands predominated by the small tussocks of Agrostis alpina and Festuca pumila are frequent on flat slopes above the tree line up to 2000 m. Stands are moderately windy but the community does not grow on extremely exposed sites. Species composition and habitat requirements indicate close relationships to the Caricetum firmae. Most important for an ecological differentiation of both these communities are soil properties. Relict brown, loamy soils with a high silt fraction facilitate the establishment of Festuca-Agrostis-grassland. Hence there is a typical accumulation of moderately acidophilous species in this community (e.g. Potentilla crantzii, Carex capillaris, C. atrata, Vaccinium vitis-idaea) and the only acidophilous (sub-) endemic of the most north easterly Alps, Pedicularis portenschlagii, occurs mainly in this community. From the pool of endemics Primula clusiana and Festuca versicolor subsp. brachystachys, especially frequent in the Caricetum firmae, are also moderately constant in this community (see Tab. 1).

In alpine habitats where pronounced humus accumulation enhances soil acidification-or acidity is due to relictic loams-a specific plant community of the Caricion firmae alliance appears. This community is dominated by dwarf shrubs (Loiseleuria procumbens, Arctostaphylos alpina, Vaccinium vitis-idaea), graminoids (Agrostis rupestris, Festuca pumila, Carex capillaris) and lichens (Cetraria islandica, C. nivalis, C. cucullata). Herbs are less frequent and those that occur indicate acidic soil properties (e.g. Campanula alpina, Potentilla aurea). Typically, a mixture of acidophilous and calciphilous plant species is established, reflecting different contact to the calcareous bedrock. The area of the Homogyno discoloris-Loiseleurietum as well as of the Festuca pumila-Agrostis alpina-grasslands respectively may have been extended by former krummholz cutting. The Homogyno discoloris-Loiseleurietum is a subendemic community of the North-eastern Limestone Alps. It also occurs on isolated stands in other parts of the Northern and Southern Alps.

Grasslands usually dominated by Oreochloa disticha or, sometimes, by Agrostis rupestris or even Carex curvula are grouped into the Juncion trifidi alliance. They have a highly scattered distribution and are restricted to deep loamy soils on tertiary sediments, accumulated in flat depressions on the plateaus. The communities are exclusively developed in the alpine belt above 1800 m. Average species diversity is low in Juncion trifidi grasslands (see Tab. 2). Besides the small tussock-graminoids, which determine vegetation structure, Leontodon helveticus, Homogyne alpina, Valeriana celtica, Potentilla aurea, Geum montanum and Anthoxanthum alpinum are the most important accompanying species. The community is especially poor in endemics. Except for Festuca versicolor subsp. brachystachys, which is not an endemic species in the strict sense, none of them is present in more than 20% of the relevés (see Tab. 1).

In summary, it may be said that there is a considerable heterogeneity among the endemic plant communities of the Northeastern Alps (see Tab. 1, Tab. 2). Communities generally forming dense canopies (e.g. Homogyno discoloris-Loiseleurietum) are represented as well as particularly open vegetation types (Drabo stellatae-Potentilletum clusianae). Mean plant species diversity varies between 17 and 38 per relevé. Floristic differences are quite pronounced; Drabo stellatae-Potentilletum clusianae and Juncion trifidi-grasslands have only few species in common. There are pioneer communities growing on initial soils (e.g. Campanulo pullae-Achilleetum clusianae), on rendsina (Helictotrichon parlatorei-grasslands) and on brown earth (Juncion trifidi). Some are restricted to steep slopes (Helictotrichon parlatorei-grasslands), while others occur only on flat terrain (Juncion trifidi). Some communities are concentrated at the lower (Athamanto-Festucetum pallidulae), while others are restricted to the upper end (Juncion trifidi) of the altitudinal range considered. Common features of all communities are that they are free of woody plants and that they are part of the natural vegetation cover, i.e. none of them has developed under the influence of human land-use. There are no endemic forest nor krummholz types nor alpine pastures.

Endemic species

The brief outline of endemic plant communities demonstrates that endemic plant species are, not surprisingly, prominent and even take the part of keystone species in the bulk of endemic plant communities. However, endemic species are not at all restricted to endemic vegetation types. Table 3 summarises the weighted constancies of endemic species in the total set of plant communities present in the study area. The distribution of endemic plants confirms the trend already mentioned. Like endemic communities, endemic species occur predominantly on natural alpine stands and decrease rapidly in seminatural pastures, fens, krummholz and forests. But regarding natural alpine vegetation only, endemic species are considerably widespread. There is no species restricted to just one plant community and very few to just one habitat type. On the contrary, many of them are distributed over a wide range of plant communities and stands.

The most important habitat type for endemic species is natural alpine grassland. 10 out of 19 species are most frequent in grasslands. Among them the climax communities of the alpine belt, Seslerio-Caricetum sempervirentis and Caricetum firmae (see Grabherr 1997), are especially rich in endemics. The frequency of endemics distinctly decreases with increasing soil acidity. Whereas Festuca pumila-Agrostis alpina-grassland, representing an intermediate position on the pH-gradient, is still rich in endemics, they become rare in the Homogyno discoloris-Loiseleurietum and especially in the Juncion trifidi-grasslands. Many of the species concentrated in grassland communities are more or less frequent in rock, scree and snowbed vegetation, too. Especially, Festuca versicolor subsp. brachystachys and Alchemilla anisiaca are considerably widespread. Only a few species of natural alpine grasslands also occur with some constancy in pastures, fens, krummholz and forests. Euphorbia austriaca is the only endemic species that is as frequent in these latter habitats as in natural alpine vegetation types.

The other half of the endemic species pool (8 out of 19) focuses on so-called azonal communities, i.e. rock, scree and snowbed vegetation. Most of these species are characterised by rather specific habitat requirements (Draba stellata on rocks, Papaver alpinum subsp. alpinum on scree, Achillea clusiana, Soldanella austriaca in snowbeds) and only Campanula pulla is frequent on different stands, though it has a distinct peak in snowbed vegetation. Generally, frequency and abundance of endemics are lowest on rocks and highest in snowbeds, with scree vegetation taking an intermediate position. All of the endemic species of azonal stands also occur in grasslands, some of them infrequently (e.g. Draba stellata, Papaver alpinum subsp. alpinum) others more regularly (Campanula pulla, Soldanella austriaca). Remarkably, only snowbed species may invade alpine pastures, fens, krummholz and forests. Species of scree- and rock habitats were hardly ever found in these vegetation types.

Figure 3 shows the average total cover abundance-values of all plots the respective species were recorded on. It provides a confirmation of the dichotomy just demonstrated. There is a group of species preferring azonal habitats, characterised by the lack of a closed vegetation cover, and a second one, that performs well even in closed vegetation canopies and that is therefore able to invade climax-communities, too.

However, in general, endemic species are over-represented in open vegetation types. The contribution of endemic species to the total species pool amounts to about 3%. But there is obviously a negative, although not linear, correlation of this figure with total vegetation cover. Whereas the species pool of plots without a closed vegetation cover contains about 8-10% endemic species, the contribution of endemics drops drastically in densely vegetated areas (see Fig. 4).

The distribution of nearly all endemic species is significantly associated with altitude (see Tab. 4). Three different types of response models may be distinguished. A negative linear response to the altitudinal gradient, indicating optimal performance of the species below the lower boundaries of the investigation area, characterises a rather small group of endemics, namely Papaver alpinum subsp. alpinum, Galium meliodorum, Festuca versicolor subsp. pallidula, Euphorbia austriaca and Leucanthemum atratum (see Tab. 4 and Fig. 5). Draba sauteri on the other hand is the only species showing a pronounced linear increase of its abundance with altitude up to the summit areas (Fig. 5). However, the bulk of the endemic flora is not linearly dependent on altitude, whereas fitting a parabola to the data provides significant results. Peaks of the regression curves lie between 1750 and 1900 m (Tab. 4 and Fig. 5).

Whereas plant species diversity steadily decreases with altitude from the subalpine zone up to the summit area, the number of endemics reaches its maximum between 1600 and 1750 m (see Tab. 4). Above 1750 m the number of endemics decreases again, but this decrease is less pronounced than the decrease in the overall species diversity. Thus the percentage of endemics of the overall species pool per altitudinal belt is steadily on the increase up to 2100 m (see Fig. 6).

The formation of endemic plant communities in the north-eastern calcareous Alps can be explained first by the occurrence of endemic plant species and, secondly, through the geomorphologic peculiarities of the region. Whereas most of the endemic communities of rock, scree and snowbed habitats owe their specific character to the fact that they are predominated by species restricted to the North-eastern Limestone Alps, most of the endemic grassland-communities, being much closer to the climax vegetation of the alpine belt, lack such endemic keystone species (for the term keystone species see below and Grabherr 1989). These plant communities consist mainly of widespread taxa. This is especially true for the predominating graminoids. Festuca pumila, Agrostis alpina, Agrostis rupestris and Oreochloa disticha occur frequently in different grassland communities throughout the Alps especially in climax communities like the Caricetum curvulae, the Caricetum firmae or the Seslerio-Caricetum sempervirentis (see e.g. Grabherr and Mucina 1993, Ellenberg 1996, Ozenda 1988). They normally do not dominate these climax communities but they apparently replace the regular keystone species, i.e. Carex curvula, Carex firma, Carex sempervirens or Sesleria albicans, if habitat qualities deviate from the prevailing environmental conditions of climax stands. This replacement seems to be due to a significant loss of competitive abilities among the regular keystone species. Strong competitors with rather specific habitat requirements are replaced by species acting more like generalists with a somewhat ruderal strategy (Grime 1979, Grabherr 1989). Patches with acidic soils in the midst of a calcareous environment provide typical examples of such deviations. Consequently, these more or less isolated acidic patches are the prevailing stands of Festuca pumila-Agrostis alpina and Juncion trifidi-grasslands as well as of Homogyno discoloris-Loiseleurietum. Their occurrence is primarily due to the persistence of tertiary sediments on the vast plateau areas that are found mainly in these most north easterly parts of the calcareous Alps.

Within an extensive and geologically quite homogeneous calcareous landscape, like the North-eastern Limestone Alps, the regional species pool for acidophilous plant communities is rather small. As regional species pool and per-community-species diversity are closely correlated (Pärtel et al. 1996, Zobel 1997), it is not surprising that average plant species diversity of the Juncion trifidi-grasslands is rather low and that most of the accompanying species are not typically acidophilous, but are rather unspecific in their habitat requirements (like e.g. Poa alpina, Persicaria vivipara or Campanula scheuchzeri). Communities indicating only moderate acidification like Festuca pumila-Agrostis alpina-grasslands or Homogyno alpini-Loiseleurietum may, on the other hand, combine a considerable part of both the calcicolous and the acidophilous species pool. The result is a rather high average plant species diversity and a very characteristic intermixture of plants usually growing on different stands (e.g. Carex firma and Loisleuria procumbens, Campanula alpina and Helianthemum alpestre, Potentilla aurea and Androsace chamaejasme). In contrast, Festuca pumila and Oreochloa disticha grasslands reported from other parts of the Alps are mainly initial vegetation types of either silicious or calcareous rock habitats and therefore characterised by a rather homogenous acidophilous or calciphilous flora (see e.g. Smettan 1981, Grabherr and Mucina 1993, Pauli et al. 1999).

As the Athamanto-Festucetum pallidulae may be rather seen as part of the rock vegetation, the only endemic grassland-community growing mainly on typical calcareous soils is the Helictotrichon parlatorei-grassland. However, Helictotrichon parlatorei was reported to predominate in grassland vegetation in other parts of the Alps, too (e.g. GRABNER 1997, RÖSLER 1997). But there seems to be general agreement among the authors from outside the most North-eastern Limestone Alps, that these Helichtotrichon-grasslands are simply a variant of the Seslerio-Caricetum sempervirentis or the Caricetum ferrugineae, whereas we consider this vegetation type rather to be an association of its own.

As already mentioned, the endemic vegetation of rock, scree and snowbed habitats owes its specific character mainly to the predominance of (sub)endemic species. The remaining flora of these communities resembles the species composition of related vegetation types on similar stands outside the calcareous Alps (Grabherr and Mucina 1993, Englisch 1999). We may therefore conclude that the formation of these endemic communities is primarily due to the existence of the respective endemic plants.

Generally, endemic species, and in particular those of the Alps, are supposed to be specifically adapted to extreme environments, where stable conditions and low competition enables long lasting survival (see for example Paw#owski 1969, 1970, Ozenda 1988). Paw#owski (1970) estimated that around 40% of the total endemic flora of the Alps (including species of all altitudinal belts from the valley bottoms to the summit areas) are distinctly restricted to rock and debris habitats.

If we focus on endemic plant communities this trend is confirmed. None of them represents a climax community; they are in fact exclusively established in azonal habitats. However, with endemic species the situation differs. Endemic plants are very frequent in the climax communities of the alpine belt. This may be partially due to spatial autocorrelation, because these communities constitute the majority of the overall alpine grassland vegetation (see Figure 2). The low constancies of many species support this interpretation. Species concentrated in azonal habitat types, like Draba stellata (rocks), Papaver alpinum subsp. alpinum (scree) or Campanula pulla (snowbeds), occasionally occur in the Caricetum firmae or Seslerio-Caricetum sempervirentis, presumably due to proximity effects. But there is no general association between the frequency of endemics and the area covered by a specific plant community. Rare and patchily distributed associations like Athamanto-Festucetum pallidulae and Alchemilla anisica-ass. may be very rich in endemics, whereas endemics are, in turn, rare in Pinus mugo-krummholz, predominating in more than 25% of the total area mapped.

The frequency of endemics in climax-grasslands of the alpine belt is surely facilitated by the spatially and temporally patchy structure of these grasslands (see e.g. Pachernegg 1973), providing canopy gaps and micro terraces with habitat conditions similar to those of rock, scree or snowbed stands. But, besides endemic species that occasionally invade from azonal habitats and occupy these gaps, there are others that distinctly prefer climax grasslands or vegetation types closely resembling them like Primula clusiana, Pedicularis portenschlagii, Dianthus alpinus, Alchemilla anisiaca and Leucanthemum atratum or species with a similar frequency in both zonal and azonal habitats like Thlaspi alpestre, Doronicum calcareum, Viola alpina Festuca versicolor subsp. brachystachys or Alchemilla anisiaca (for the latter see also Pignatti-Wikus 1959). Neither spatial autocorrelation nor microhabitat diversity may therefore provide a sufficient explanation for the frequency of endemic plants in climax communities.

Competition has been demonstrated to be of major importance even in alpine grasslands (see e.g. Gigon 1971, Theodose and Bowman 1997). Such grasslands are generally structured by only a few species, mostly long-living graminoids forming dense canopies and acting as strong competitors. They were thus termed keystone species by Grabherr (1989). The bulk of the endemic plants of the North-eastern Alps is obviously able to deal with the presence of such competitors and to survive in vegetation types predominated by them. Moreover, some of them seem to act as keystone species themselves. Festuca versicolor subsp. pallidula and subsp. brachystachys, Alchemilla anisiaca, Achillea clusiana and Campanula pulla may be called keystone species in their respective communities. Indeed, all these communities are restricted to azonal stands, but the habitats they are established in are in no way unusual in mountainous terrain. On the contrary, rock and snowbed vegetation cover vast areas especially in the alpine belt of calcareous mountains, where the distinction between zonal and azonal stands becomes more and more difficult to make.

The altitudinal distribution of endemic plants does not seem to correspond with the theory of endemics being restricted to the most extreme habitats, either. The bulk of them obviously perform best in the lower alpine belt, and the summit areas are rather poor in endemics, although the mountain ranges considered hardly extend into the upper alpine zone. This concentration of endemic species in the lower alpine belt is no peculiarity of the North-eastern Limestone Alps but was also reported from other mountain ranges (see Agakhanyanz and Breckle 1995). Extreme altitudes seem to be generally dominated by widespread species (Körner 1999).

Hence it may be concluded that the concept of endemics being generally weak competitors and restricted to azonal stands does not hold true. On the other hand, the frequency of endemic plants in plots lacking a closed vegetation cover and the fact that the percentage of endemics of the overall species pool increases with altitude, apparently indicate that endemics profit from decreasing interspecific competition. But these figures are, at least in part, caused by the fact that there are two different groups of species: one that apparently avoids competition (e.g. Draba stellata, Draba sauteri, Papaver alpinum subsp. alpinum), and another one that grows well in dense grasslands. Thus, the competitive abilities of endemic plant species obviously vary considerably.

It has already been pointed out that many of the endemic plant species are not restricted to one plant community or even one habitat type. Recent studies on the distribution of relict species in the easternmost silicious Alps have provided similar results (Schneeweiss and Schönswetter 1999). This fact seems to be inconsistent with one of the standard explanations of the regional endemism in the Alps, which is that reduced genetic plasticity prevented the redispersal of endemics when the glaciers retreated (see Ehrendorfer 1965, Niklfeld 1972). Even if large parts of the North-eastern Alps were not glaciated, landscapes may be assumed to have changed significantly with climatic change during the Holocene. Nevertheless, endemic species were, at least partly, capable of reinvading most of these changing habitats. Species that had suffered a severe loss of genotypes should not be expected to do so. It is hardly plausible that species as widespread within their area of distribution as Leucanthemum atratum, Primula clusiana, Dianthus alpinus or Thlaspi alpestre, should be characterised by a particular lack of biotypes.

We therefore agree with Pils (1988, 1995), that the prevailing historical explanation of endemism in the most north-eastern calcareous Alps is in need of revision and that the importance of current ecological conditions requires reconsideration (see also Zimmermann 1972, Schneeweiss and Schönswetter 1999). The climatic peculiarities of this region (Baumgartner et al. 1983) may significantly influence plant distribution. Indeed, climatic conditions change gradually from east to west and this has often been interpreted to be inconsistent with abrupt plant distribution limits (Merxmüller 1952, Niklfeld 1967, 1972). However, this combination is a common phenomenon, for which population theory provides some possible explanations (e.g. metapopulation dynamics: Silvertown and Lovett Doust 1995, "geographical selection gradient model": Fischer 1950, Slatkin 1973, 1975, May et al. 1975, "tension zone model": Barton and Hewitt 1989). The specific climatic conditions may contribute to the competitive ability of regional endemics especially in relation to closely related vicariants growing in similar habitats outside the North-eastern Alps (e.g. Dianthus alpinus-Dianthus glacialis, Leucanthemum atratum-Leucanthemum halleri, Festuca versicolor-Festuca varia). However, it is unlikely that mechanisms of competitive exclusion should be restricted to pairs of closely related taxa. Such mechanisms normally operate within groups of species with similar habitat requirements and resource utilization (see e.g. Newman 1982, Wilson and Tilman 1991) that need not be taxonomically related.