Variability of Abies alba-dominated forests in Central Europe

Transcription

1 Cent. Eur. J. Biol. 9(5) DOI: /s y Central European Journal of Biology Variability of Abies alba-dominated forests in Central Europe Krzysztof Świerkosz 1, Kamila Reczyńska 1, Karel Boublík 2 Research Article 1 Museum of Natural History, Wrocław University, PL Wrocław, Poland 2 Faculty of Environmental Sciences, Czech University of Life Sciences Prague, CZ Prague, Czech Republic Received 12 June 2013; Accepted 27 October 2013 Abstract: Using vegetation databases from Poland, the Czech Republic, Austria, Slovakia and our own material collected during a field work, we have analysed the variability of species composition of Abies alba-dominated communities in Central Europe. Analysis wasperformedusingamodifiedtwinspanalgorithminthejuicesoftware.ecologicalanalysiswasperformedonthebasisof mean Ellenberg indicator values with a MoPeT_v1.0.r script prepared in R software. In general, the findings indicated that there are at least 8 types of Central European silver fir-dominated forests with different patterns in species composition and habitat conditions. These are the oligotrophic subcontinental silver fir forests of the Polish highlands, oligotrophic (sub)montane silver fir forest, hygrophilous silver fir forests, oligo-mesotrophic (sub)montane silver fir forests, mesotrophic (sub)montane silver fir forests, submontane and montane calcicolous silver fir forests of the Alps and the Carpathians, eutrophic silver fir-beech forests and silver fir ravine forests. The main conclusion is that the diversity of species composition of Abies alba-dominated forests in Central Europe is lower than described in current regional synthesis, which was confirmed by ecological analyses. Keywords: Coniferous forests Silver fir Species composition TWINSPAN Vegetation patterns Versita Sp. z o.o. 1. Introduction An increasing number of databases containing relevés of various types of plant communities and an easier access to their resources [1,2] provides an opportunity to create synthesis of vegetation not only within particular countries [3-5], but also within geographical regions, owing to direct comparison of available data using numerical methods. Synthesis of this type has been prepared for communities of the Molinio-Arrhenatheretea class in the southern part of Central Europe [6], communities of the Mulgedio-Aconitetea class in Central and Northern Europe [7], for weed vegetation of arable land in the Czech Republic and Slovakia as well as for Balkans [8,9], inland Bulboschoenus vegetation [10], Salix tundra thickets [11], Pinus mugo-communities [12,13] and semi-dry and dry grasslands in Northern and Central Europe [14-16]. Interestingly, only a few comparative studies concerned forest communities [17-20], despite relevés of forests being the major part of most national vegetation databases [21,22]. As was rightly mentioned by Knollová et al. [4], many plant communities in Europe were distinguished in accordance with the European phytosociological tradition, based on highly detailed research, but carried out on a small area. Afterwards, syntaxonomical units created in this manner were presented as applicable to much wider areas and repeatedly formed the basis for national phytosociological classifications. Currently, having comparative material and possibilities of numerical analysis, we should avoid partial classification and rather create a pan-european syntaxonomical system including precise classification of communities on a continental scale. In the current study, we focused on Central European forest communities with a dominance of silver fir (Abies * krzysztof.swierkosz@life.pl 495

2 Variability of Abies alba-dominated forests in Central Europe alba), due to many inconsistencies resulting from the comparison of different classification systems in particular countries [23-31]. These inconsistencies have three main reasons. First, silver fir forests frequently occur within vegetation belts naturally dominated by European beech (Fagus sylvatica) or Norway spruce (Picea abies). Second, the species composition of Abies alba-dominated forests represents a transition between beech and spruce forests [26]. Third, in many areas of Central Europe it is not clear if silver fir forests represent a natural vegetation type. Some scientists present evidence that the dominance of silver fir is the result of different human activities such as forest pasture, litter raking and the preference for fir wood as a building timber [26,32,33]. For these reasons the classification of silver fir-dominated forests into higher-ranked syntaxa is questionable since they can be classified both within Vaccinio-Piceetea and Querco-Fagetea [see the differences in classification by: 24,26-28,30,31]. Some authors even suppose that secondary forests should not be described as separate communities [33]. The goals of this paper were to describe the patterns in species composition within silver fir-dominated forests occurring in Central Europe and recognize the main environmental factors affecting their vegetation variability. 2. Experimental Procedures 2.1 Relevé data The study is based on the analysis of phytosociological relevés with the occurrence of Abies alba in the tree layer, collected according to Braun-Blanquet approach [34] and originating from the vegetation databases of Austria (1521 relevés), the Czech Republic (1412 relevés), Poland (558 relevés) and Slovakia (665 relevés) [21,22,35-37]. We also used 206 relevés from Poland, which were obtained from field investigations and literature sources and stored and maintained by authors using Turboveg 2.94 software [38]. The total data set used in the preliminary analysis contained 4362 relevés and covered a major part of the Bohemian Massif, the western part of the Carpathian arc and the eastern part of the Alps. Also data from the northern border of the Abies alba distribution range in Europe, the Góry Świętokrzyskie Mts in Poland, were included. 2.2 A priori selection To obtain the final data set containing relevés of silver fir-dominated forests we applied the additional criteria in the database. Firstly, we removed all the relevés with plot size smaller than 100 m 2 and larger than 600 m 2. Using such a range of plot sizes is acceptable for heterogeneous data sets [39], and enables us to analyse more data than ranges of m 2 or m 2, which are suggested as better comparable. Thus, we selected 2408 relevés (55% from total data set). In order to restrict the analysis only to forest communities with a dominance of Abies alba, we chose those with at least 50% coverage of Abies alba (i.e. 4 and 5 in the Braun-Blanquet scale) from the set of the relevés. Finally, 718 relevés from the whole data set available have been analysed: 168 relevés from Poland, 449 relevés from the Czech Republic, 45 relevés from Austria and 56 relevés from Slovakia. The nomenclature follows Tutin et al. [40-44] for vascular plants and Ochyra et al. [45] for bryophytes. We merged some species into aggregates (marked as agg. ), due to the regionally different taxonomic approach. Most of the aggregates correspond to Chytrý and Tichý [3], however some of them were established a priori by database content (Appendix 1). The moss layer also was included in the analysis, but we excluded lichens, because they have been recorded sporadically throughout the databases. 2.3 Numerical analyses We prepared the analysis using the JUICE software [46] with a modified TWINSPAN method [47]. We transformed the cover abundance scale into a three-step ordinal scale, with pseudospecies cut levels at 0%, 5% and 25%. We merged different layers for the same species using JUICE implemented procedure. The cover of the species aggregation was calculated under the assumption that covers can overlap and that they do so independently of each other [48]. A Total Inertia method was used as a measure of cluster heterogeneity [47]. Diagnostic species for vegetation units were determined using the Φ coefficient as a measure of fidelity [20,49]. The species with threshold values of Φ coefficient >25 were considered to be diagnostic, even if their frequency was low. Fisher s exact test (P<0.05) was used to eliminate the fidelity value of species with non-significant occurrence patterns. The Φ coefficient was used for clusters of equalized size [50]. All fidelity values used in the paper were multiplied by 100. The constant species were definied as a species with an occurrence frequency of at least 50% in the described cluster. The dominant species were those having a percentage cover higher than 50% at least in one relevé of the individual vegetation unit. The final classification of the relevés was based on the TWINSPAN results (Figure 1). Preliminary results of analysis in JUICE software using 496

3 K. Świerkosz et al. Figure 1. Dendrogram of numerical classification for the 718 vegetation plots of silver fir-dominated forests analysed in Central Europe: Cluster 1 oligotrophic subcontinental silver fir forests of Polish highlands; Cluster 2 oligotrophic (sub)montane silver fir forests; Cluster 3 hygrophilous silver fir forests; Cluster 4 oligo-mesotrophic (sub)montane silver fir forests; Cluster 5 mesotrophic (sub)montane silver fir forests; Cluster 6 submontane and montane calcicolous silver fir forests of the Alps and the Carpathians; Cluster 7 eutrophic silver fir-beech forests; and Cluster 8 ravine forests dominated by silver fir. crispness of classification algorithm suggested that the optimal number of clusters was about 10. Our decision to use an 8-cluster-solution was based on subjective considerations, because our aim in this paper was to present the most ecologically and floristically consistent vegetation groups within Abies alba forests. Therefore, we decided to reduce number of clusters and obtain vegetation units that would be coherent and possible to interpret in floristic and ecological terms. Individual relevés, which significantly differed from the remaining relevés within the group, were subjectively allocated between clusters, based on their geographical location and floristic composition. We did not use any methods of geographical stratification due to the lack of geographical coordinates in relevés originating from Slovakia and Austria. In order to avoid depletion in the available information and misinterpretation of the results, we excluded random stratification methods, due to the small number of available relevés compared to the wide geographical range of the analysis. Species composition of the communities was presented in the summarised synoptic table with the modified fidelity Φ coefficient and the percentage frequency as superscript (Table 1). The diagnostic species (Φ>25) and/or species with frequency equal or higher than 80% are boldly marked in the table. Species that did not reach a fidelity of 20 and/or a frequency of 50% in at least one column were excluded. In accordance with the methodology described above, we made an additional division within each cluster to recognize their internal diversity. Distinguished units, were characterised with a short description containing differentiating species, main ecological conditions and geographical or altitudinal range. Mean Ellenberg indicator values (EIVs) were used to explain differences in species composition, as there were no direct measurements available [51]. In the first stage, the ordination analysis was performed using the Canoco 4.5/CanoDraw 4.1 software [52]. The structure of the data set was analysed with detrended correspondence analysis (DCA). The gradient length of the first DCA axis (3.681 SD units) indicated an application of unimodal ordination techniques. Therefore, the DCA was performed to obtain the sample scores used in the modified Monte Carlo permutation test. Default options and no transformation of species cover data was used in the DCA. A forward selection procedure and modified Monte Carlo permutation test with 499 unrestricted permutations were conducted to identify the statistical significance of the environmental variables and their relationship with species composition of analyzed samples [53]. The test was performed with MoPeT_v1.0.r script prepared for this purpose in R software [54] by David Zelený. Sample scores of the first four DCA axes were used as external variables. The analysis of variance (ANOVA) of mean EIVs among groups (with 499 unrestricted permutations) and multiple regression of EIVs with sample scores of the first four axes of DCA as external variables, were calculated using MoPeT_v1.0.r script and R software [53]. 497

4 Variability of Abies alba-dominated forests in Central Europe No. of cluster No. of releves Diagnostic species Trientalis europaea Pteridium aquilinum Plagiothecium cuspidatum Pinus sylvestris Quercus petraea agg Betula pubescens agg Luzula pilosa Frangula alnus Brachythecium salebrosum Calamagrostis epigejos Leucobryum glaucum Bazzania trilobata Equisetum sylvaticum Soldanella montana agg Ranunculus repens Chaerophyllum hirsutum Myosotis palustris agg Caltha palustris Lysimachia nemorum Deschampsia cespitosa Carex remota Alnus glutinosa Cardamine amara Crepis paludosa Lysimachia vulgaris Galium palustre Polytrichum commune Cirsium palustre Mnium hornum Calamagrostis villosa Anemone nemorosa Viola palustris Carex brizoides Valeriana dioica Rhytidiadelphus loreus Chrysosplenium alternifolium Table 1. Summarised synoptic table of silver fir-dominated forests in Central Europe. Species are ordered according to the diagnostic value in particular group. Φ coefficient values in the table are multiplied by 100. Their superscripts express percentage values of species frequency in particular cluster. Diagnostic species (Φ>25) and/or species with frequency equal or higher than 80% are boldly marked in the table. Species that did not reach a fidelity of 20 and/or a frequency of 50% in at least one column were excluded. 498

5 K. Świerkosz et al. No. of cluster No. of releves Rhizomnium punctatum Glyceria fluitans Sphagnum palustre Athyrium filix femina Prunella vulgaris Equisetum arvense Digitalis purpurea Ajuga reptans Luzula luzuloides Calamagrostis arundinacea Hypnum cupressiforme agg Hieracium murorum Viola riviniana agg Plagiomnium affine Moehringia trinervia Eurhynchium angustirete Rubus idaeus Chelidonium majus Rosa canina agg Valeriana tripteris Clematis alpina Salvia glutinosa Euphorbia amygdaloides Carex alba Poa stiriaca Cirsium erisithales Gentiana asclepiadea Asplenium viride Epilobium montanum Carex digitata Calamagrostis varia Lonicera xylosteum Galium sylvaticum agg Ribes alpinum Campanula trachelium Digitalis grandiflora Lonicera nigra Table 1. Summarised synoptic table of silver fir-dominated forests in Central Europe. Species are ordered according to the diagnostic continued value in particular group. Φ coefficient values in the table are multiplied by 100. Their superscripts express percentage values of species frequency in particular cluster. Diagnostic species (Φ>25) and/or species with frequency equal or higher than 80% are boldly marked in the table. Species that did not reach a fidelity of 20 and/or a frequency of 50% in at least one column were excluded. 499

6 Variability of Abies alba-dominated forests in Central Europe No. of cluster No. of releves Aruncus dioicus Asplenium trichomanes Moneses uniflora Polystichum aculeatum agg Hieracium bifidum Tanacetum clusii Fragaria vesca Mnium spinosum Ribes uva crispa Dentaria glandulosa Thalictrum aquilegiifolium Astrantia major Ribes petraeum Stachys alpina Rhytidiadelphus triquetrus Tortella tortuosa Cornus sanguinea agg Daphne mezereum Dentaria bulbifera Symphytum cordatum Carex sylvatica Ulmus glabra Acer platanoides Chaerophyllum temulum Impatiens noli tangere Taxus baccata Lamium maculatum Viola odorata Stellaria holostea Pulmonaria officinalis agg Stellaria media agg Impatiens parviflora Torilis japonica Geum urbanum Aegopodium podagraria Adoxa moschatellina Fraxinus excelsior Table 1. Summarised synoptic table of silver fir-dominated forests in Central Europe. Species are ordered according to the diagnostic continued value in particular group. Φ coefficient values in the table are multiplied by 100. Their superscripts express percentage values of species frequency in particular cluster. Diagnostic species (Φ>25) and/or species with frequency equal or higher than 80% are boldly marked in the table. Species that did not reach a fidelity of 20 and/or a frequency of 50% in at least one column were excluded. 500

7 K. Świerkosz et al. No. of cluster No. of releves Elymus caninus Carpinus betulus Hepatica nobilis Dentaria enneaphyllos Asarum europaeum agg Plagiomnium rostratum Poa nemoralis agg Sanicula europaea Ranunculus lanuginosus Diagnostic species of Cl. Vaccinio-Piceetea Picea abies Dryopteris carthusiana agg Vaccinium myrtillus Polytrichum formosum agg Pleurozium schreberi Avenella flexuosa Dicranum scoparium Carex pilulifera Vaccinium vitis idaea Sphagnum capillifolium Plagiochila asplenioides Phegopteris connectilis Sphagnum girgensohnii Diagnostic species of Cl. Querco-Fagetea Acer pseudoplatanus Senecio nemorensis agg Dryopteris filix mas agg Mycelis muralis Corylus avellana Actaea spicata Galeobdolon luteum agg Galium odoratum Geranium robertianum agg Mercurialis perennis agg Melica nutans agg Paris quadrifolia Table 1. Summarised synoptic table of silver fir-dominated forests in Central Europe. Species are ordered according to the diagnostic continued value in particular group. Φ coefficient values in the table are multiplied by 100. Their superscripts express percentage values of species frequency in particular cluster. Diagnostic species (Φ>25) and/or species with frequency equal or higher than 80% are boldly marked in the table. Species that did not reach a fidelity of 20 and/or a frequency of 50% in at least one column were excluded. 501

8 Variability of Abies alba-dominated forests in Central Europe No. of cluster No. of releves Glechoma hederacea agg Myosotis sylvatica agg Viola reichenbachiana Campanula rapunculoides Campanula persicifolia Sambucus nigra Carex pilosa Circaea lutetiana Stachys sylvatica Stellaria nemorum agg Species common in most of the Abies alba-dominated forests Abies alba Oxalis acetosella Sorbus aucuparia Fagus sylvatica Maianthemum bifolium Rubus fruticosus agg Prenanthes purpurea Atrichum undulatum Diagnostic species of fidelity > 20 Larix decidua Rumex acetosella agg Danthonia decumbens Quercus robur agg Carex pallens Pseudoscleropodium purum Circaea x intermedia Calypogeia muelleriana Juncus effusus Lysimachia nummularia Plagiomnium undulatum Carex pallescens Petasites albus Polypodium vulgare agg Campanula rotundifolia agg Senecio viscosus Solidago virgaurea agg Table 1. Summarised synoptic table of silver fir-dominated forests in Central Europe. Species are ordered according to the diagnostic continued value in particular group. Φ coefficient values in the table are multiplied by 100. Their superscripts express percentage values of species frequency in particular cluster. Diagnostic species (Φ>25) and/or species with frequency equal or higher than 80% are boldly marked in the table. Species that did not reach a fidelity of 20 and/or a frequency of 50% in at least one column were excluded. 502

9 K. Świerkosz et al. No. of cluster No. of releves Dactylis glomerata agg Isothecium alopecuroides Scrophularia nodosa Carex muricata agg Bromus benekenii Ranunculus platanifolius Phyteuma spicatum Ctenidium molluscum Cardaminopsis halleri Leontodon hispidus Fissidens dubius Laserpitium latifolium Cyclamen purpurascens Viola biflora Cardamine impatiens Rubus saxatilis Isothecium myosuroides Orthilia secunda Asplenium scolopendrium Hypericum hirsutum Table 1. Summarised synoptic table of silver fir-dominated forests in Central Europe. Species are ordered according to the diagnostic continued value in particular group. Φ coefficient values in the table are multiplied by 100. Their superscripts express percentage values of species frequency in particular cluster. Diagnostic species (Φ>25) and/or species with frequency equal or higher than 80% are boldly marked in the table. Species that did not reach a fidelity of 20 and/or a frequency of 50% in at least one column were excluded. The altitudinal differentiation among clusters was calculated using all the available data (relevés number specified in the description of each cluster) and summarized as histograms (Figures 2a-h). The analysis was performed using the PAST software [55]. 2.4 Syntaxonomical classification and nomenclature A comparison of literature sources was necessary for interpretation of distinguished vegetation units. For this purpose, we analysed synoptic tables of silver fir forests covering our study area [24,28,29,56-58], as well as list of character species and descriptions of associations from Central Europe [25,27,59,60]. 3. Results After analyses we obtained 8 clusters of different silver fir-dominated forests (Figure 1, Table 1). The species with Φ coefficient >40, as well as those with frequency in clusters higher than 80%, were boldly marked in the cluster s description. 3.1 Descriptions of the clusters Cluster 1 (Table 1, Column 1) oligotrophic subcontinental silver fir forests of Polish highlands Number of relevés: 76. Average species number in the relevé: Geographical distribution of relevés: Poland. Altitude (N=26): range m.a.s.l.; mean 417 m.a.s.l.; median 420 m.a.s.l (Figure 2a). Diagnostic species: Betula pubescens agg., Brachythecium salebrosum, Calamagrostis epigejos, Frangula alnus, Luzula pilosa, Pinus sylvestris, Plagiothecium cuspidatum, Pleurozium schreberi, Polytrichum formosum agg., Pteridium aquilinum, Quercus petraea agg., Trientalis europaea, Vaccinium myrtillus. Constant species: Abies alba, Dryopteris carthusiana agg., Frangula alnus, Luzula pilosa, Maianthemum 503

10 Variability of Abies alba-dominated forests in Central Europe Figure 2. Frequency histograms of altitudinal spectrum of units recognized within silver fir-dominated forests in Central Europe (m.a.s.l.). A: cluster 1 oligotrophic subcontinental silver fir forests of Polish highlands; B: cluster 2 oligotrophic (sub)montane silver fir forests; C: cluster 3 hygrophilous silver fir forests; D: cluster 4 oligo-mesotrophic (sub)montane silver fir forests; E: cluster 5 mesotrophic (sub)montane silver fir forests; F: cluster 6 submontane and montane calcicolous silver fir forests of the Alps and the Carpathians; G:cluster 7 eutrophic silver fir-beech forests; and H: cluster 8 ravine forests dominated by silver fir. 504

11 K. Świerkosz et al. bifolium, Oxalis acetosella, Picea abies, Pinus sylvestris, Pleurozium schreberi, Polytrichum formosum agg., Pteridium aquilinum, Quercus petraea agg., Rubus fruticosus agg., Sorbus aucuparia, Trientalis europaea, Vaccinium myrtillus. Dominant species: Abies alba, Oxalis acetosella, Picea abies, Pinus sylvestris, Pleurozium schreberi, Rubus fruticosus agg., Sphagnum girgensohnii, Vaccinium myrtillus. Three different subunits were recognized within the cluster: 1. With high frequency of Senecio nemorensis agg., Fagus sylvatica, Rubus idaeus, Corylus avellana, Sambucus racemosa, Quercus robur agg., Athyrium filix-femina, Phegopteris connectilis and Gymnocarpium dryopteris (in the Świętokrzyskie Mts and on the Silesia- Cracow Upland). 2. With high frequency of Carpinus betulus, Moehringia trinervia, Melica nutans, Mycelis muralis and Plagiothecium cuspidatum (in the Kozienicka Forest and some localities in western Poland). 3. With high frequency of Hylocomium splendens, Plagiothecium laetum, Sphagnum girgensohnii, Dicranum scoparium, Brachythecium salebrosum, Melampyrum pratense and Pleurozium schreberi (uplands of western and southern Poland). Cluster 2 (Table 1, Column 2) oligotrophic (sub) montane silver fir forests Number of relevés: 76. Average species number in the relevé: Geographical distribution of relevés: Austria and Czech Republic, a few relevés from Slovakia and Poland. Altitude (N= 76): range m.a.s.l.; mean 712 m.a.s.l.; median 685 m.a.s.l. (Figure 2b). Diagnostic species: Avenella flexuosa, Bazzania trilobata, Leucobryum glaucum, Pleurozium schreberi, Vaccinium myrtillus. Constant species: Abies alba, Avenella flexuosa, Dicranum scoparium, Dryopteris carthusiana agg., Fagus sylvatica, Oxalis acetosella, Picea abies, Pleurozium schreberi, Polytrichum formosum agg., Sorbus aucuparia, Vaccinium myrtillus. Dominant species: Abies alba, Avenella flexuosa, Calamagrostis villosa, Dicranum scoparium, Dryopteris carthusiana agg., Eurhynchium striatum, Fagus sylvatica, Hylocomium splendens, Leucobryum glaucum, Oxalis acetosella, Picea abies, Plagiomnium affine, Pleurozium schreberi, Polytrichum formosum agg., Vaccinium myrtillus, Vaccinium vitis-idaea. Two altitudinal subunits were recognizable within this cluster: 1. Community of uplands and lower mountains (mean altitude 570 m.a.s.l.) with Leucobryum glaucum, Pinus sylvestris, Quercus robur agg., Luzula luzuloides and Thuidium tamariscinum. This form is close to phytocoenoses described in cluster 1, but it occurrs only in the Czech Republic, Slovakia and Austria. 2. Mountain community (mean altitude 870 m.a.s.l.) with Galium rotundifolium and Huperzia. Cluster 3 (Table 1, Column 3) hygrophilous silver fir forests Number of relevés: 31. Average species number in the relevé: Geographical distribution of relevés: Austria, Czech Republic, Poland and a single relevé from Slovakia. Altitude (N=27): range m.a.s.l.; mean 707 m.a.s.l.; median 660 m.a.s.l. (Figure 2c). Diagnostic species: Ajuga reptans, Alnus glutinosa, Anemone nemorosa, Athyrium filix-femina, Calamagrostis villosa, Caltha palustris, Cardamine amara, Carex brizoides, Carex remota, Chaerophyllum hirsutum, Chrysosplenium alternifolium, Cirsium palustre, Crepis paludosa, Deschampsia cespitosa, Digitalis purpurea, Equisetum arvense, Equisetum sylvaticum, Galium palustre, Glyceria fluitans, Lysimachia nemorum, Lysimachia vulgaris, Maianthemum bifolium, Mnium hornum, Myosotis palustris agg., Polytrichum commune, Prunella vulgaris, Ranunculus repens, Rhizomnium punctatum, Rhytidiadelphus loreus, Soldanella montana agg., Sphagnum palustre, Vaccinium myrtillus, Valeriana dioica, Viola palustris. Constant species: Abies alba, Acer pseudoplatanus, Athyrium filix-femina, Dryopteris carthusiana agg., Equisetum sylvaticum, Maianthemum bifolium, Oxalis acetosella, Picea abies, Polytrichum formosum agg., Senecio nemorensis agg., Soldanella montana agg., Sorbus aucuparia, Vaccinium myrtillus. Dominant species: Abies alba, Alnus glutinosa, Athyrium filix-femina, Avenella flexuosa, Bazzania trilobata, Caltha palustris, Carex brizoides, Carex remota, Equisetum sylvaticum, Fagus sylvatica, Galium rotundifolium, Melica nutans agg., Mnium hornum, Oxalis acetosella, Picea abies, Polytrichum formosum agg., Rubus idaeus, Sanicula europaea, Senecio nemorensis agg., Soldanella montana agg., Sphagnum girgensohnii, Thuidium tamariscinum. No subunits were recognized. Cluster 4 (Table 1, Column 4) oligo-mesotrophic (sub) montane silver fir forests Number of relevés: 141. Average species number in the relevé: Geographical distribution of relevés: Austria, Czech Republic, Slovakia, Poland. 505

12 Variability of Abies alba-dominated forests in Central Europe Altitude (N=163): range m.a.s.l.; mean 644 m.a.s.l.; median 632 m.a.s.l. (Figure 2d). Diagnostic species: Avenella flexuosa, Calamagrostis arundinacea, Hieracium murorum, Hypnum cupressiforme agg., Luzula luzuloides. Constant species: Abies alba, Athyrium filix-femina, Avenella flexuosa, Calamagrostis arundinacea, Corylus avellana, Dicranum scoparium, Dryopteris carthusiana agg., Dryopteris filix-mas agg., Fagus sylvatica, Hieracium murorum, Hypnum cupressiforme agg., Luzula luzuloides, Mycelis muralis, Oxalis acetosella, Picea abies, Polytrichum formosum agg., Rubus idaeus, Senecio nemorensis agg., Sorbus aucuparia, Vaccinium myrtillus. Dominant species: Abies alba, Avenella flexuosa, Calamagrostis arundinacea, Fagus sylvatica, Hypnum cupressiforme agg., Oxalis acetosella, Picea abies, Plagiomnium affine, Pleurozium schreberi, Polytrichum formosum agg., Sorbus aucuparia, Thuidium tamariscinum, Vaccinium myrtillus. Two different subunits are recognizable within this cluster: 1. Mesophilous, with Mycelis muralis, Fragaria vesca, Carex digitata, Melica nutans, Galium rotundifolium, Moehringia trinervia and Viola riviniana. 2. Acidophilous, with Orthodicranum montanum, Bazzania trilobata, Prenanthes purpurea and Sphagnum girgensohnii. The floristical composition of these forms was probably determined by edaphic conditions because the forms did not show geographical and altitudinal dependence. Cluster 5 (Table 1, Column 5) mesotrophic (sub) montane silver fir forests Number of relevés: 205. Average species number in the relevé: Geographical distribution of relevés: Poland, Czech Republic; two relevés from Slovakia. Altitude (N=196): range m.a.s.l.; mean 562 m.a.s.l.; median 522 m.a.s.l. (Figure 2e). Diagnostic species: Chelidonium majus, Eurhynchium angustirete, Moehringia trinervia, Plagiomnium affine, Rosa canina agg., Rubus idaeus, Viola riviniana agg. Constant species: Abies alba, Acer pseudoplatanus, Athyrium filix-femina, Corylus avellana, Dryopteris carthusiana agg., Dryopteris filix-mas agg., Hieracium murorum, Moehringia trinervia, Mycelis muralis, Oxalis acetosella, Picea abies, Plagiomnium affine, Polytrichum formosum agg., Rubus idaeus, Senecio nemorensis agg., Sorbus aucuparia. Dominant species: Abies alba, Calamagrostis arundinacea, Carpinus betulus, Corylus avellana, Dryopteris carthusiana agg., Dryopteris filix-mas agg., Eurhynchium angustirete, Fagus sylvatica, Fragaria vesca, Hypnum cupressiforme agg., Impatiens nolitangere, Lonicera nigra, Oxalis acetosella, Picea abies, Plagiomnium affine, Sambucus racemosa, Sorbus aucuparia, Thuidium tamariscinum, Tilia cordata, Tilia platyphyllos. Three subunits were recognizable within the cluster: 1. With high frequency of Chelidonium majus, Galium sylvaticum agg., Urtica dioica, Campanula trachelium, Cardaminopsis arenosa, Hepatica nobilis, Campanula persicifolia, Campanula rapunculoides, Ribes uvacrispa and other species (hilly areas, Czech Republic). 2. With high frequency of Brachypodium sylvaticum, Ajuga reptans, Carex sylvatica, Sanicula europaea, Dentaria bulbifera and Festuca gigantea (low mountain zone, Czech Republic and Poland). 3. With higher share of acidophilous species, such as Carex pilulifera, Frangula alnus, Vaccinium myrtillus, Phegopteris connectilis and Calamagrostis villosa (mountain zone of the Czech Republic, Poland and Slovakia). Cluster 6 (Table 1, Column 6) submontane and montane calcicolous silver fir forests of the Alps and the Carpathians Number of relevés: 71. Average species number in the relevé: Geographical distribution of relevés: Austria, Slovakia, Poland. Altitude (N=67): range m.a.s.l.; mean 780 m.a.s.l.; median 720 m.a.s.l. (Figure 2f). Diagnostic species: Actaea spicata, Aruncus dioicus, Asplenium trichomanes, Asplenium viride, Astrantia major, Calamagrostis varia, Campanula trachelium, Carex alba, Carex digitata, Cirsium erisithales, Clematis alpina, Cornus sanguinea agg., Daphne mezereum, Dentaria glandulosa, Digitalis grandiflora, Epilobium montanum, Euphorbia amygdaloides, Fragaria vesca, Galeobdolon luteum agg., Galium sylvaticum agg., Gentiana asclepiadea, Hieracium bifidum, Lonicera nigra, Lonicera xylosteum, Mnium spinosum, Moneses uniflora, Poa stiriaca, Polystichum aculeatum agg., Rhytidiadelphus triquetrus, Ribes alpinum, Ribes petraeum, Ribes uva-crispa, Salvia glutinosa, Stachys alpina, Tanacetum clusii, Thalictrum aquilegiifolium, Tortella tortuosa, Valeriana tripteris. Constant species: Abies alba, Acer pseudoplatanus, Actaea spicata, Athyrium filix-femina, Carex digitata, Corylus avellana, Dryopteris carthusiana agg., Dryopteris filix-mas agg., Epilobium montanum, Fagus sylvatica, Fragaria vesca, Galeobdolon luteum agg., Geranium robertianum agg., Hieracium murorum, Lonicera xylosteum, Maianthemum bifolium, Melica nutans agg., Mycelis muralis, Oxalis acetosella, 506

13 K. Świerkosz et al. Picea abies, Prenanthes purpurea, Rubus idaeus, Salvia glutinosa, Senecio nemorensis agg., Sorbus aucuparia, Valeriana tripteris. Dominant species: Abies alba, Brachythecium velutinum, Calamagrostis arundinacea, Calamagrostis varia, Eurhynchium angustirete, Fagus sylvatica, Galium sylvaticum agg., Geranium robertianum agg., Larix decidua, Lunaria rediviva, Mercurialis perennis agg., Oxalis acetosella, Phyteuma spicatum, Picea abies, Pimpinella major agg., Plagiomnium cuspidatum, Prenanthes purpurea, Primula elatior agg., Ranunculus lanuginosus, Rhytidiadelphus triquetrus, Rubus idaeus, Rubus saxatilis, Vaccinium vitis-idaea, Valeriana tripteris. The cluster is heterogenous and the insufficient number of available relevés prevents the recognition of its component units. At least four different groups of relevés, may be recognized. 1. Phytocoenoses occurred in Austria are differentiated by share of Cyclamen purpurascens, Sorbus aria agg., Veronica urticifolia, Moehringia muscosa, Knautia maxima, Adenostyles glabra, Calamagrostis varia and many others. 2. High mountain Slovak forests are characterized by Calamagrostis arundinacea, Phyteum spicatum, Tanacetum clusii, Lilium martagon, Polygala brachyptera, Leucanthemum rotundifolium and many species of Mulgedio-Aconitetea communities. 3. High-mountain communites of the Polish side of the Carpathian Arc are differentiated by occurrence of Veronica officinalis, Campanula polymorpha agg., Luzula luzulina, Goodyera repens, Viola biflora, Moneses uniflora, Orthilia secunda, Pyrola minor and Coeloglossum viride. 4. Low-mountain zone communities of Poland and Slovakia with occurrence of Gleochoma hederacea agg., Ribes uva-crispa, Cardamine impatiens, Lonicera xylosteum, Acer platanoides and Asplenium scolopendrium. Cluster 7 (Table 1, Column 7) eutrophic silver fir-beech forests Number of relevés: 110. Average species number in the relevé: Geographical distribution of relevés: Austria, Czech Republic, Slovakia and Poland. Altitude (N=108): range m.a.s.l.; mean 648 m.a.s.l; median 625 m.a.s.l. (Figure 2g). Diagnostic species: Actaea spicata, Carex sylvatica, Dentaria bulbifera, Galeobdolon luteum agg., Galium odoratum, Symphytum cordatum. Constant species: Abies alba, Acer pseudoplatanus, Athyrium filix-femina, Dryopteris carthusiana agg., Dryopteris filix-mas agg., Fagus sylvatica, Galeobdolon luteum agg., Galium odoratum, Impatiens noli-tangere, Mercurialis perennis agg., Mycelis muralis, Oxalis acetosella, Picea abies, Rubus fruticosus agg., Rubus idaeus, Senecio nemorensis agg., Sorbus aucuparia. Dominant species: Abies alba, Acer pseudoplatanus, Athyrium filix-femina, Dentaria glandulosa, Dryopteris carthusiana agg., Dryopteris filix-mas agg., Fagus sylvatica, Festuca altissima, Galeobdolon luteum agg., Galium odoratum, Impatiens noli-tangere, Lonicera nigra, Lunaria rediviva, Mercurialis perennis agg., Oxalis acetosella, Picea abies, Plagiomnium affine, Plagiomnium undulatum, Rubus fruticosus agg., Rubus idaeus, Sambucus racemosa, Senecio nemorensis agg., Sorbus aucuparia, Urtica dioica. Three subunits of these forests were recognizable: 1. With higher occurrence of Taxus baccata, Salvia glutinosa, Dentaria glandulosa and Symphytum cordatum (Poland). 2. With higher occurrence of Dentaria enneaphyllos, Euphorbia amygdaloides, Petasites albus, Pulmonaria officinalis agg. and Polygonatum verticillatum (Slovakia). 3. With higher occurrence of Moehringia trinervia, Melica nutans agg., Festuca altissima and Poa nemoralis agg. (Austria and the Czech Republic). Cluster 8 (Table 1, Column 8) ravine forests dominated by silver fir Number of relevés: 8. Average species number in the relevé: Altitude (N=8): range m.a.s.l.; mean 410 m.a.s.l; median 330 m.a.s.l. (Figure 2h). Geographical distribution of relevés: Czech Republic. Diagnostic species: Acer platanoides, Acer pseudoplatanus, Adoxa moschatellina, Aegopodium podagraria, Asarum europaeum agg., Carpinus betulus, Chaerophyllum temulum, Dentaria enneaphyllos, Elymus caninus, Fraxinus excelsior, Galium odoratum, Geranium robertianum agg., Geum urbanum, Hepatica nobilis, Impatiens nolitangere, Impatiens parviflora, Lamium maculatum, Plagiomnium rostratum, Poa nemoralis agg., Pulmonaria officinalis agg., Ranunculus lanuginosus, Sanicula europaea, Stellaria holostea, Stellaria media agg., Taxus baccata, Torilis japonica, Ulmus glabra, Viola odorata. Constant species: Abies alba, Acer platanoides, Acer pseudoplatanus, Asarum europaeum agg., Carpinus betulus, Dryopteris filix-mas agg., Fraxinus excelsior, Galium odoratum, Geranium robertianum agg., Impatiens noli-tangere, Impatiens parviflora, Mycelis muralis, Oxalis acetosella, Pulmonaria officinalis agg., Senecio nemorensis agg., Ulmus glabra. 507

14 Variability of Abies alba-dominated forests in Central Europe Dominant species: Abies alba, Acer pseudoplatanus, Adoxa moschatellina, Carpinus betulus, Cerasus avium, Corylus avellana, Dicranum scoparium, Fraxinus excelsior, Galium odoratum, Geranium robertianum agg., Impatiens parviflora, Lamium maculatum, Mercurialis perennis agg., Oxalis acetosella, Polytrichum formosum agg., Rubus fruticosus agg., Sambucus nigra, Sorbus aucuparia, Stellaria holostea, Taxus baccata, Ulmus glabra. The cluster was homogenous. 3.2 Main environmental gradients Differences in species composition of particular vegetation types enabled general identification of the main environmental factors causing this diversity. A correlation of mean EIVs with the distribution of studied samples along the axes of DCA analysis revealed that four factors were responsible for the arrangement of studied samples along the first and the second DCA axis. These are: continentality, soil reaction, nutrients and temperature (Figures 3A and B). The distribution of samples along the third axis of DCA was correlated with light, moisture and continentality (Figure 3C). The fourth axis did not show any statistical significant correlation with environmental factors expressed by EIVs. Multiple regression results confirmed that all analyzed EIVs exerted a statistically significant influence on variability of species composition within silver firdominated forests (Table 2). Analysis of variance (ANOVA) performed within the final data set (718 relevés) using a modified permutation test revealed that none of the EIVs was statistically significant (P.modif in Table 3 and Figure 4). On the contrary, standard ANOVA confirmed statistical significance of differences in mean values of all analyzed EIVs among clusters (P.perm in Table 3 and Figure 4). 4. Discussion 4.1 Diversity of Abies alba-dominated forests in Central Europe A review of the literature conducted separately for each of the studied countries of Central Europe always produced a comprehensive overview of fir forests communities, which seemed to be consistent and unambiguous. Silver fir forests in the area of investigation were recently commented on by Moravec et al., Exner, Matuszkiewicz and Jarolímek et al. [27-30] who presented a clear interpretation of particular syntaxonomical units distinguished during field work based on a specific set of environmental factors and diagnostic species. However, a major problem arises from any attempt to compare existing classification systems and associations distinguished within them. It turns out that individual syntaxonomical units known from Poland, the Czech Republic, Austria and Slovakia almost do not overlap with each other, and syntaxonomical position of silver fir is different in these countries. Based on presented analyses we can distinguish in Central Europe at least eight types of forest communities with domination of silver fir, which differ among each other in species composition and ecological aspects. Cluster 1 contained relevés only from Poland. Not only floristic composition but also geographical (Polish uplands and the Świętokrzyskie Mts) and altitudinal range (from 180 up to 550 m.a.s.l.) were in accordance with literature sources referring to association Abietetum albae [29,56,61]. Species diagnostic for this cluster (Pinus sylvestris, Vaccinium myrtilus, Luzula pilosa, Trientalis europaea and Calamagrostis epigeios) were continental in character, which corresponds well to occurrence of this type of phytocoenoses on the north-eastern border of the silver fir natural range. Cluster 2 embraced relevés from highlands and lower montane zones across Central Europe (average altitude 712 m.a.s.l.). A large share of oligotrophic bryophytes (Bazzania trilobata, Leucobryum glaucum, Dicranum polysetum, Pleurozium schreberi, Polytrichum formosum agg.) and a constant and high share of Avenella flexuosa and Vaccinium myrtillus were distinctive features of this cluster. According to the results of the EIVs analysis, this cluster contains relevés with a high frequency of oligotrophic species and a high mean value of continentality, close to vegetation plots of Abietetum albae. That enables us to put forward a hypothesis about the subcontinental character of the phytocoenoses including in cluster 2 [see also 26]. Despite a few relevés from Poland, caused by insufficient participation of Abies alba in documented phytocoenoses, the occurrence of this community in some ranges of the Sudeten Mountains [62] and in the Carpathians [63] is likely. Species composition and habitat conditions suggest that the discussed cluster contains relevés that in most of the studied countries are treated as Vaccinio myrtilli-abietetum (syn. Vaccinio vitis-idaeae-abietetum). In Poland, the most similar phytocoenoses are included in the wide-treated Abieti- Piceetum montanum association, which comprises all spruce-fir forests in the lower mountain zone, regardless of their trophic differences [29]. In the Alps these forests were included in Bazzanio-Piceetum (subunit 1) and Calamagrostio villosae-piceetum (subunit 2) by Exner [28]. 508

15 K. Świerkosz et al. Figure 3. Results of correlation of Ellenberg indicator values (EIVs) with sample scores of the Detrended Correspondence Analysis (DCA) axis. P. modif. was calculated using modified permutation test of significance for analysis of mean EIVs. P. perm coefficient of statistical significance in Monte Carlo permutation test. rho Spearman s rank correlation coefficient which is a nonparametric measure of statistical dependence between two variables. A: correlation of EIVs with sample scores of the 1st DCA axis; B: correlation of EIVs with sample scores of the 2nd DCA axis; and C: correlation of EIVs with sample scores of the 3rd DCA axis. mean EIV AX1 AX2 AX3 AX4 R 2 R 2. adj P. modif Light Moisture Temperature Continentality Soil Reaction Nutrients Table 2. Results of multiple regression calculated with sample scores of the first four axes of Detrended Correspondence analysis (DCA) as an external variables for silver fir-dominated forests in Central Europe 509

16 Variability of Abies alba-dominated forests in Central Europe Cluster 3 contained relevés from the Czech Republic, Poland and Austria, which are of the most hygrophilous character. According to literature sources, they are usually treated as hygrophilous variants of various communities of fir forests [26,57] or as separate associations. However, due to the close resemblance among relevés originating from different areas and a rich set of diagnostic species, the latter approach was more correct. Similar phytocoenose has been reported as Equiseto sylvatici-abietetum or Carici brizoidis- Abietetum from Austria [25,28], Slovakia [30] and Croatia [64] so far. In the Czech Republic, Husová [65] and Moravec et al. [27] described such a community as Carici remotae-abietetum. Also the species composition of some forest patches from the Tatra foothills, which were included in the community Abies alba-oxalis acetosella [29], resembles this cluster. It means that the occurrence of these kinds of forests seems to be well documented in Poland and in the Czech Republic as well. This type of phytocoenoses are meso- and significantly hygrophilous in character and occur mostly in the lower montane zone. Cluster 4 contained relevés from the entire studied part of Europe. They are of (sub)mountain character and occur predominantly in highlands and lower montane zones (average 644 m.a.s.l.). Diagnostic species are difficult to identify in this cluster. Only species that are medioeuropean in character (Calamagrostis arundinacea, Luzula luzuloides and Hieracium murorum) show higher participation and abundance. Mean EIV Df F P. perm P. modif Light Temperature Continentality Moisture Soil Reaction Nutrients Table 3. Summary of the ANOVA results for mean Ellenberg indicator values (EIVs) within silver fir-dominated forests in Central Europe. Figure 4. Summary box-and-whisker plots of mean Ellenberg indicator values (EIVs) for clusters recognized within silver fir-dominated forests in Central Europe produced by one-way ANOVA. P. modif. was calculated using modified permutation test of significance for analysis of mean EIVs. P. perm coefficient of statistical significance in Monte Carlo permutation test. F test statistic. 510