Diversity of Mulgedio-Aconitetea communities in the Sudetes Mts. (SW Poland) in the Central European context

Aims: To describe the compositional and ecological diversity of Mulgedio-Aconitetea communities in the Sudetes Mts. and their foothills. Study area: The Sudetes Mts. (Southwestern Poland). Methods: A total of 399 vegetation relevés from own field studies and the literature were sorted into groups that match the higher syntaxa of the EuroVegChecklist and associations described in the literature. Diagnostic species of the so delimited associations were determined with the phi-coefficient of association, and maps of the associations produced. Direct ordination methods were applied to identify the main environmental gradients shaping the plant communities. Results: We distinguished nine associations, belonging to four alliances: submontane and colline communities (Petasition officinalis: Geranio phaei-Urticetum dioicae, Petasitetum hybridi, Chaerophyllo hirsuti-Petasitetum albi, Prenanthetum purpureae), upper montane nitrophilous communities (Rumicion alpini: Rumicetum alpini); subalpine communities with a dominance of graminoids and ferns (Calamagrostion villosae: Poo chaixii-Deschampsietum cespitosae, Crepido conyzifoliae-Calamagrostietum villosae, Athyrietum filicis-feminae) and subalpine tall-herb communities (Adenostylion alliariae: Cicerbitetum alpinae). Altitude, light availability, and bedrock type, which determines nutrient availability and soil reaction, played an important role in differentiating the studied communities. Conclusions: For convenience, we placed the four alliances in four separate orders as in the EuroVegChecklist. The fact that our ordination diagram separated only two main groups suggests the need of further research in this matter. Taxonomic reference: Euro+Med (2006-) for vascular plants. Syntaxonomic reference: Higher syntaxa follow Mucina et al. (2016). Abbreviations: db-RDA = distance-based redundancy analysis; EIV = Ellenberg indicator value; pANOVA = permutational analysis of variance; PCoA = principal coordinates analysis.


Introduction
Montane to subalpine tall-herb communities of Europe, Siberia and Greenland are classified in the class Mulgedio-Aconitetea Hadač et Klika in Klika et Hadač of eutrophic habitats on raw alluvia of streams in the upper colline to supramontane belts (Mucina et al. 2016). Due to the mosaic character of montane vegetation, which is often due to microhabitats created by glacial relief or intense erosion, communities of different phytosociological classes may co-occur with each other even at small spatial scales. Therefore, the classification of tall-herb communities is extremely complicated, as reflected by different systems throughout Central European countries that are usually incompatible with one another. For instance, in Poland, 15 associations and two alliances are distinguished within the class Mulgedio-Aconitetea (Matuszkiewicz 2012), in Czechia 13-16 associations and five alliances (Kočí 2001(Kočí , 2007Kočí et al. 2003), in Slovakia 32 associations within eight alliances (Kliment et al. 2004;Jarolímek et al. 2008;Šibíková et al. 2008) and in Austria (excluding the Alnion viridis Schnyder 1930) 12 associations and four alliances (Karner and Mucina 1993). An ambitious attempt to organize these jigsaw puzzles down to the level of alliances and associations across Central and Northern Europe was undertaken by Michl et al. (2010). A noticeable consequence of the study by Michl et al. (2010) was a significant reduction in the number of associations described from individual mountain ranges and their inclusion in syntaxa with broader distribution. From Western and Central Europe (including the Pyrenees, Massif Central, Alps, Central European highlands and Carpathians) only 19 associations within five alliances were distinguished. Later, Mucina et al. (2016) proposed a system for the higher syntaxa (alliances to class) for the whole of Europe.
The aims of this study are thus: i) to conduct a comprehensive analysis of diversity of tall-herb communities of the Mulgedio-Aconitetea in the Sudetes Mts., assign the terminal units to described associations and place them into the higher syntaxa of the EuroVegChecklist (https:// www.synbiosys.alterra.nl/evc/; Mucina et al. 2016); ii) to determine diagnostic species of these communities in the regional context; and iii) to identify the ecological factors influencing community species composition.

Study area
Our research was conducted on the whole area of the Sudetes, a mountain range in southwestern Poland extending over a length of 300 km and covering together with its foothills approximately 5,550 km 2 ( Figure 1). The altitude extends across 1400 m (the highest peak is Śnieżka Mt., 1602 m a.s.l.), which is sufficient for the formation of diverse vegetation belts, from the foothills to the subalpine zone, marked by typical communities. In turn, the longitudinal location of the range causes suboceanic species to occur in the western part (e.g., Meum athamanticum, Chrysosplenium oppositifolium), while in the eastern part, Carpathian and Alpine elements occur (e.g., Sesleria sadlerana subsp. tatrae, Scabiosa lucida, Cardamine trifolia). The geological structure of the Sudetes Mts. is complex. Acidic rocks, both plutonic (granites) and sedimentary (gneisses, shales, sandstones), predominate here, but there are numerous intrusions of nutrient-rich, effusive rocks (basalts, trachytes, andesites) and metamorphic rocks (greenstones and greenschists, marbles, crystal limestones and dolomites). Thus, different types of soils derive from the diverse bedrocksfrom initial soils and podzols, which are extremely poor in nutrients, through brown soils, to nutrient-rich calcareous rendzinas (Blachowski et al. 2005). The average annual temperature ranges from 8.9°C in the foothills to 0.7°C on the top of Śnieżka Mt., and the precipitation reaches from 600 to 1200 mm/year. The relief is also notably diverse, with deep gorges next to vast plateaus and cones of volcanic origin. The hydrological network is well developed, with several major rivers (Nysa Łużycka, Kwisa, Bóbr, Kaczawa, Bystrzyca and Nysa Kłodzka), all belonging to the Odra basin. The valleys of larger rivers are usually wide and flat, and the rivers in such sections are often partially regulated. The ravines of both larger rivers and streams are often marked by a natural hydrological regime. Due to the diversity of habitats, plant communities with vastly different ecological requirements can develop here in close proximity to each other. This also applies to tall-herb communities.

Field sampling and literature data
Between 1991 and 2020, we sampled 212 vegetation plots of the class in the Sudetes Mts. and their foothills (coordinates 15.32°E-17.23°E and 50.20°N-51.25°N), at elevations from 240 to 1400 m a.s.l. The plots were located in the terraces or banks of the stream valleys as well as within mires with tall-herbs. We chose stands that con-tained diagnostic species for the class Mulgedio-Aconitetea and its subordinate syntaxa as known from the literature (Suppl. material 1). In accordance with Michl et al. (2010) and Mucina et al. (2016), we excluded vegetation plots with a shrub cover of more than 30%. According to Mucina et al. (2016) and following Kliment et al. (2010), we also considered the order Petasito-Chaerophylletalia Morariu 1967 ex Kopecký 1969 as belonging to the class Mulgedio-Aconitetea. The phytosociological material was collected using the Braun-Blanquet approach (Mueller-Dombois and Ellenberg 2002). The plot size varied from 5 m 2 to 60 m 2 (mean 21.5 m 2 ). All the relevés were stored in a TUR-BOVEG database (Hennekens and Schaminée 2001).
The relevés are available via the Polish Vegetation Database (Global Index of Vegetation-Plot Databases, ID: EU-PL-001; Kącki and Śliwiński 2012). Additionally, our own original relevés are available upon request through the VESTA Database (ID: EU-PL-004 in the Global Index of Vegetation-Plot Databases).

Environmental variables
In order to identify the ecological conditions of the tallherb communities within the study area, different environmental variables were analyzed. Elevation (measured in m a.s.l. and divided by 1000 for presentation), slope, heat load and bedrock type were used as explanatory variables. The bedrock type at each site was obtained from the Detailed Geological Map of the Sudetes Mts. (Polish Geological Institute, National Research Institute, http:// sudety.pgi.gov.pl). Based on the criterion of mineral composition and major geological processes (Bolewski and Parachoniak 1988), six main categories of rocks were considered as explanatory variables: Quaternary Pleistocene deposits (postglacial sands, clays and gravels), Quaternary Holocene deposits in the stream valleys, metamorphic rocks (gneisses, schists), plutonic (granitoids), acidic sedimentary (sandstones) and calcicolous rocks (limestone mudstones and marbles). According to McCune and Keon (2002) the heat load was used instead of aspect because aspect is a "circular" variable with 0° = 360°. As there were no direct measurements of light and soil conditions, we used Ellenberg indicator values (EIVs) (Ellenberg et al. 1991), corrected by datasets of Berg et al. (2017) with reference to values of continentality. EIVs weighted by percentage species' cover were calculated for each relevé using the JUICE software (Tichý and Holt 2006).

Phytosociological analysis
Prior to the analyses, occurrences of the same species in different vertical layers were merged using the procedure implemented in JUICE, under the assumption that the overlap of layers is random (Fischer 2015). Bryophytes were excluded from the analysis as they had been recorded only in part of the plots.
We conducted an unsupervised classification with the modified TWINSPAN algorithm (Roleček et al. 2009) with total inertia measure of heterogeneity as implemented in JUICE software (Tichý 2002). Based on the expert evaluation of the intitial clusters, we merged and re-arranged some of them (Suppl. material 2). Diagnostic species of the so derived terminal clusters (associations) were then determined using the Φ coefficient as a measure of fidelity for clusters of equalized size (Chytrý et al. 2002;Tichý and Chytrý 2006;Willner et al. 2009). Species with Φ ≥ 0.25, constancy ≥ 20%, constancy ratio (Dengler et al. 2005) ≥ 1.3 and statistically significant concentration in a particular cluster, tested by the Fisher's exact test (p < 0.05), were considered diagnostic. A species was considered diagnostic for more than one cluster with Φ > 0.25 in at least two clusters and a constancy ratio of ≥ 1.3. Species with constancy ratio < 1.3 and/or Φ > 0.25 in one or more clusters were not considered diagnostic but are also presented in Table 1. Constant species were defined as species with frequency of at least 50% in a particular cluster and dominant species as those with cover > 75% (5 on a Braun-Blanquet's cover scale) at least in a single relevé in this cluster. Based on literature sources (Kočí 2007;Stachurska-Swakoń 2009;Michl et al. 2010;Kliment et al. 2010) we classified the diagnostic species as either regional character or differential species. Then, we produced distribution maps of the syntaxa using QGIS 3.16 (with support of © MapTiler © OpenStreetMap contributors). For the associations, we used names acccording to the Michl et al. (2010) and Kliment et al. (2010) without checking in detail their validity according to the ICPN (Theurillat et al. 2021). Finally, the associations were arranged into the higher syntaxa of the EuroVegChecklist (https://www.synbiosys.alterra.nl/evc/; Mucina et al. 2016) by expert judgement.

Ecological analysis
A principal coordinates analysis (PCoA) was performed in CANOCO (Ter Braak and Smilauer 2012) both to explore differentiation of recognized groups and to check the percentage of variation explained. Species percentage cover data were transformed with log (x + 1), and the distance matrix (399 × 399) was calculated using the percentage difference (SQRT Jaccard binary distance). The distribution of the sample groups was visualized with a PCoA diagram. To identify the statistical significance of correlations (using Spearman's coefficient) between the PCoA sample scores obtained from CANOCO and mean randomized EIVs for relevés, a modified permutation test with 499 unrestricted permutations was conducted. The test was performed with MoPeT_v1.2.r script (Zelený and Schaffers 2012) in the R software (R Development Core Team 2021). Permutational analysis of variance (one-way pANOVA on the mean randomized EIVs) and modified permutation test (with 499 unrestricted permutations) were also calculated using MoPeT_v1.2.r (Zelený and Schaffers 2012), to determine which EIVs differentiate the selected communities. Using pANOVA is as an alternative to other tests under non-normal conditions, because it does not operate under the assumption of normality and uses actual scores (Gleason 2013).
To identify the main ecological drivers affecting the diversity of distinct groups, a distance-based redundancy analysis (db-RDA), embedded in CANOCO 5.0 (Ter Braak and Smilauer 2012) with SQRT Jaccard binary distance, and the variation was explained (Jupke and Schäfer 2020). A standard Monte Carlo permutation test with 499 unrestricted permutations under the full model was conducted to identify the significance of the simple term and conditional effects of environmental variables (elevation, slope, heat load and the main bedrock type) on the species composition of the analyzed samples (Ter Braak and Smilauer 2012). The conditional effect expresses the variation explained by a single explanatory variable, whereas the others are used as covariables. The simple effect expresses the variation explained by the single explanatory variable without covariables.

Results
In the analyzed data we could distinguish nine groups of tall-herb communities (Table 1, Figure 2, Suppl. material 3). We interpreted the nine groups as associations described in the literature and placed them in the higher hierarchies of the EuroVegChecklist as follows:  Pawłowski et al. 1928 All. Calamagrostion villosae Pawłowski et al. 1928  The nine associations clearly differ in terms of diagnostic species (Table 1

Description of distinguished groups of relevés
In the following, among the diagnostic species, the character species (in bold) are highlighted (see Table 1). The association was found at elevations between 300 and 964 m a.s.l. (mean 532 m a.s.l.), indicating its submountain and lower-mountain character. It was recorded mainly in stream valleys over Quaternary Holocene deposits, and only exceptionally over postglacial formations. Most of the documented stands occurred in the vicinity of submontane riparian forests, but in large, open and sunny clearings. The succession of trees and shrubs in such tall-forb patches is low, which may indicate regular floods of watercourses, increasing the mortality rate of seedlings of woody forms. The association was found at elevations of 375-995 m a.s.l. (mean 695 m a.s.l.), indicating its submontane/lower-montane character. It usually accompanies nitrophilous, deciduous forest communities of the Carpino-Fagetea class -mainly alluvial forests, but also nitrophilous beech or ravine woods. This association was recorded from all types of analyzed bedrock, most often on Holocene gravels in stream valleys, but also on sedimentary and metamorphic rocks. Therefore, it does not appear to show any preferences regarding bedrock type and is one of the most widely distributed communities within the alliance. Stands over calcareous rocks are richer in species, with the inclusion of Delphinium elatum, Ranunculus platanifolius or Aconitum variegatum, and resemble the Delphinietum elatae Beger ex Sutter 1978. However, so far only one locality of that type is known in the Kleśnica Valley (Eastern Sudetes, Śnieżnik Massif, about 800 m a.s.l.). The association was found at elevations of (405-) 650-925 m a.s.l. (mean 758 m a.s.l.), therefore is montane in character. However, in contrast to the previous association, it accompanies acidophilous beech forests of the Luzulo-Fagetalia sylvaticeae Scamoni et Passarge 1959 order and spruce forest communities of natural or anthropogenic origin. Stands were mainly reported from sedimentary and metamorphic rocks, rarely from Holocene gravels or granites. The association is less frequent than the previous one and is restricted to a few mountain ranges (Śnieżnik Massif, Stołowe Mts., Orlickie Mts., Izerskie Mts. and Karkonosze Mts.).  montane and subalpine zones. In the Sudetes Mts. it was reported from the Śnieżnik Massif, Izerskie Mts., Bystrzyckie and Orlickie Mts. mainly on metamorphic rocks (schists) and, rarely, Holocene deposits.

Group 5 -Rumicetum alpini
Group 7 -Crepido conyzifoliae-Calamagrostietum villosae (Zlatník 1925   The association was found at elevations from 770 to 1380 m a.s.l. (mean 1168 m a.s.l.), in the upper montane and subalpine zones. In the Sudetes Mts. it is reported from the Karkonosze Mts. on granites, mainly on steep slopes of the postglacial circles, and from the Orlickie and Izerskie Mts. on metamorphic bedrocks.

Ecological differentiation
As shown in the PCoA ordination diagram, the nine associations fall within two clusters that were clearly separated along the first PCoA axis ( Figure 6). Associations 1-6 were in the cluster with low PCoA1 values, while assocations 7-9 with the exception of few relevés of association 8, were in the cluster with high PCoA1 values. Within the left-hand cluster (low PCoA1 values), the associations largely overlapped, particularly association 1 and 2, as well as associations 3, 4 and 6. By contrast, the three associations of the right-hand cluster were quite well separated from each other. The first and the second PCoA axes explained 8.8% and 4.2% of compositional variability of studied communities, respectively. The first PCoA axis was significantly negatively correlated with the EIVs for soil reaction (p < 0.05), nutrients (p < 0.05) and temperature (p < 0.01). The second and third PCoA axes were significantly and positively correlated with the EIVs for light (p < 0.01) and moisture (p < 0.05), respectively ( Table 2).
The pANOVA revealed that temperature played a significant role in the floristic differentiation of the nine associations (p < 0.05). In contrast, the other analyzed EIVs for nutrients, soil reaction, moisture and light were not significant (Figure 7).
The db-RDA (Figure 8) revealed that the explanatory variables used in the analysis accounted for 13.5% of the total variation in species composition (while R 2 adj .   Figure 2. Clusters 1 and 2 refer to submontane communities on nutrient-rich, Quaternary deposits; cluster 3 and 4 represent phytocoenoses in shaded sites, located within forest stands in the lower alpine zone; clusters 5 to 9 represent light-demanding and slightly acidophilic phytocoenoses developing from the subalpine to lower alpine zone. Symbols are grouped in alliances: circles:  was 10.3%). However, the contribution of the three environmental variables (elevation, slope and the main types of bedrock) to the explained variance depended on whether we consider simple term effects or conditional effects (Table 3). Most important was elevation, but slope and bedrock type were also significant.

Delimitation of the Mulgedio-Aconitetea against other classes
Separation of communities of the class Mulgedio-Aconitetea from other tall-herb communities of the classes Hitherto, in synthetic studies concerning the Mulgedio-Aconitetea class, character species were distinguished based on a priori prepared lists (Michl et al. 2010) or numerical analyses (Kliment et al. 2010). For the territory of Poland, Kącki et al. (2013) attempted to identify species diagnostic for the whole class as well as orders and alliances. The authors analyzed 127 relevés of tall-herb but also shrub communities accompanying watercourses. Regardless of differences in methodology, we used a similar set of diagnostic species when preparing Suppl. material 1. The Mulgedio-Aconitetea class is well defined by a wide range of alpine and subalpine species whose abundance increases with altitude. Therefore, at lower altitudes it may be more difficult to distinguish tall-herb phytocoenoses whose species composition refers to communities of the classes Epilobietea angustifolii or Molinio-Arrhenatheretea Tx. 1937. However, even in such localities, species recognized as diagnostic for the Mulgedio-Aconitetea class, such as Silene dioica, Stellaria nemorum, Chaerophyllum hirsutum, Anthriscus nitida or Petasites albus regularly appear (Table 1). In this respect, an important aspect of our adopted classification system is that we included in the Mulgedio-Aconitetea class the order Petasito-Chaerophylletalia, which is in line with the concept of Kliment et al. (2010) and Mucina et al. (2016), but in contrast to Michl et al. (2010). This solution is supported by the presence of species of this class in communities belonging to this order, although their proportion gradually decreases with decreasing elevation. Moreover, these are communities associated with the valleys of montane and submontane watercourses, in contrast to other tall-herb communities of the Epilobietea angustifolii or Artemisietea vulgaris classes.

Subdivision of the Mulgedio-Aconitetea into orders and alliances
Despite recognition at both regional and supra-regional scales, there is still no general agreement on the syntaxonomy of tall-herb communities of the class Mulgedio-Aconitetea. The synthesis carried out by Michl et al. (2010)    Central Europe, the Balkans and the Apennines of order with two Central European alliances: Petasition officinalis (vegetation on raw alluvia of streams in the upper colline to supramontane belts of the Carpathians and the Hercynicum) and Arunco dioici-Petasition albi (in the montane and supramontane belts of the Alps). 4. Senecioni rupestris-Rumicetalia alpini: Tall-herb anthropogenic vegetation on nutrient-rich soils in the upper montane to alpine belts with the single alliance Rumicion alpini.
Here we adopted the concept of four orders of Mucina et al. (2016), because it is the only consistent proposal for a syntaxonomic classification of all plant communities in Europe at the level of classes, orders and alliances. We acknowledge that this proposal is not based on a detailed phytosociological analysis, and may change in the future. As the syntaxonomic division of the Mulgedio-Aconitetea class presented by Mucina et al. (2016) is complex, our discussion focuses on the alliances and orders known from Central Europe (including the Hercynicum), and omits higher units typical of Southern Europe, the Balkans and the boreal-subarctic group of orders. What distinguishes our study from those of Michl et al. (2010) and Mucina et al. (2016) is the presence of two, clearly distinctive groups of tall-herb communities. The first group includes colline-montane tall-herb communities accompanying watercourses at elevations between 200 and 1000 m a.s.l. which are rich in nemoral and nitrophilic species. We associate this group with the order Petasito-Chaerophyletalia, which Michl et al. (2010) do not distinguish at all. The second group consists of alpine communities that according to Michl et al. (2010) belong to one order Calamagrostietalia villosae, and according to Mucina et al. (2016) belong to three different orders. Some deviations of the classification proposed by us from these earlier studies are due to several facts. First, as we already mentioned, the material analyzed was strongly differentiated along the elevational gradient by inclusion of submontane tall-herb communities. Second, in the case of the Rumicetum alpini and Poo chaixii-Deschampsietum cespitosae associations, we faced the limited number of relevés, which may also affect the final classification. Therefore, it is difficult to determine which of these two above-mentioned supra-regional classifications better reflects the actual diversity of tall-herb communities in the Sudetes Mts. Nevertheless, our presented findings indicate a need for future syntheses at larger spatial extent to evaluate whether the pattern observed in the Sudetes Mts. is a local phenomenon or may contribute to changes in a general syntaxonomic scheme.

Justification and circumscription of associations
The main discrepancies occur for the communities classified here in the Petasito-Chaerophylletalia order (Groups 1-4). The first problems concern the assignment of the Geranio phaei-Urticetum to the order Petasito-Chaerophylletalia. Phytocoenoses of this type were described for the first time by Hadač et al. (1969) from the Dolina Siedmich Prameňov (Belaer Tatras) at elevations of 1265-1310 m a.s.l. The association is also listed as quite common in the Austrian Alps and their foothills, especially on calcareous substratum , and reported from the Tatra Mts. in Poland, (Balcerkiewicz 1978). Due to floristic composition of this association, Michl et al. (2010), included it in the Rumicion alpini. However, their decision was based only on 6 relevés in total [2 relevés of Hadač et al. (1969) and four relevés of Kliment (1989Kliment ( , 1991], missing for example the relevés published by Świer kosz et al. (2002). The latter material documents also submontane and even colline forms of the association. The data analyzed in the present study (33 relevés) show a wide altitudinal range of this community, which is not limited to the highest mountain parts (such as the Rumicetum alpini), but develops from the foothills to the subalpine zone, on the stream terraces covered by soils enriched with nitrogen due to high anthropogenic pressure. Altitudinal variation translates into internal variation in species composition of the association (Świerkosz et al. 2002). The phytocoenoses described by Hadač et al. (1969) and Balcerkiewicz (1978) from higher elevation in the Tatra Mts. differ from those described from the Sudetes only in the presence of three species from the class Mulgedio-Aconieteta (Carduus personata, Rumex arifolius and Epilobium alpestre). The last two species occur sporadically. On the other hand, the species composition is dominated, as in the case of forms at lower elevations, by Urtica dioica, Geum urbanum, Rumex obtusifolius, Dactylis glomerata, Aegopodium podagraria, Chaerophyllum aromaticum and Ranunculus repens (see Hadač et al. 1969, pp. 216-217). Nevertheless, it appears that Geranio phaei-Urticetum is a well-differentiated floristically and coherent unit, with a large group of its own diagnostic species. At the same time, due to the large number of nitrophilous species considered to be distinctive for the Petasition officinalis, and which occur also in other associations of this alliance, it seems reasonable for us to include the group 1 in the order Petasitio-Chaerophylletalia rather than in Senecioni rupestris-Rumicetalia alpini. This is especially prudent since the latter aggregates tall-herb anthropogenic vegetation on nutrient-rich soils in the upper montane to alpine belts, with common occurrence of subalpine species (e.g., Ochlopoa supina, Peucedanum ostruthium, Rumex arifolius, Senecio nemorensis, Alchemilla glabra, Adenostyles alliaria and Athyrium distentifolium).
An additional point of debate is the placement of the Petasitetum hybridi. Traditionally, this association is not placed in the class Mulgedio-Aconitetea, but considered as a lowland community, and thus placed in tall-herb classes of the lowlands. A typical stand of the Petasition officinalis according to Kliment and Jarolímek (2002: p. 107) is represented by the relevé 1 on page 134 in Sillinger (1933). According to Michl et al. (2010), this type relevé should be assigned to the lowland tall-herb communities due to the prevalence of many diagnostic species of the former Artemisietea vulgaris and Filipendulo ulmariae-Calystegietea sepium Géhu et Géhu-Franck 1987 (e.g., Aegopodium podagraria, Anthriscus sylvestris, Filipendula ulmaria, Galium aparine, Petasites hybridus). Moreover, Michl et al. (2010) suggest classifying the Petasition officinalis within the Filipendulo ulmariae-Calystegietea sepium, largely corresponding to the order Convolvuletalia sepium Tx. ex Moor 1958 according to Mucina et al. (2016). A similar solution is applied e.g. in Czechia where the whole alliance Petasition hybridi Sillinger 1933 is included in the class Galio-Urticetea Passarge ex Kopecký 1969(Laníková et al. 2009), corresponding to the Epilobietea angustifolii in Mucina et al. (2016). In Poland, the association is included in the alliance Aegopodion podagrariae within the class Artemisietea vulgaris (Matuszkiewicz 2012), whereas Austrian and German synthesis typically place it in the order Lamio albi-Chenopodietalia boni-henrici Kopecký 1969 within the class Galio-Urticetea (Hilbig et al. 1972;Pott 1993). A contrasting approach is presented by Mucina et al. (2016) who describe the Petasition officinalis as "tall-herb vegetation on raw alluvia of streams in the upper colline to supramontane belts of the Carpathians and the Hercynicum" and include it in the order Petasito-Chaerophylletalia within the class Mulgedio-Aconitetea. This solution has also been supported in other regional classifications, especially in Slovakia Kliment et al. 2010). Kliment et al. 2010 (Table 1, col. 8) indicated both Geranium phaeum and Petasites hybridus as diagnostic species for the order Petasito-Chaerophylletalia. Similar to the studies of Jarolímek et al. (2002) and Kliment et al. (2010), in our dataset, diagnostic species for the Mulgedio-Aconitetea still appear in the Petasitetum hybridi regularly (e.g., Petasites albus, Chaerophyllum hirsutum, Primula elatior, Stellaria nemorum), or sporadically (Silene dioica, Geranium sylvaticum, Veratrum album, Aconitum variegatum, Delphinium elatum). There are clear connections among these species, through the presence of low-mountain and nitrophilous species to the communities of the classes Epilobietea angustifolii and Molinio-Arrhenatheretea, particularly to the alliance Filipendulo-Petasition Br.-Bl. ex Duvigneaud 1949. However, this reference is also found in Table 1 col, 8 in Kliment et al. (2010), which presents the similar set of species as we see in our data (e.g., Chaerophyllum aromaticum, Aegopodium podagraria, Filipendula ulmaria, Rumex obtusifolius, Lamium maculatum, Cirsium oleraceum, Galium aparine, Lysimachia nummularia, Schedonorus giganteus and others). It should also be noted that the latest study on the differentiation of the class Molinio-Arrhenathereta in Poland (Kącki et al. 2021) does not list herbaceous communities with Petasites hybridus within this class at all.
The next two associations that we distinguished (Chaerophyllo hirsuti-Petasitetum albi and Prenanthetum purpureae) overlap with the range of the broadly defined Chaerophyllo hirsuti-Cicerbitetum alpinae. However, in our opinion, differences in species composition and ecological characteristics fully justify their separation. The association Chaerophyllo hirsuti-Petasitetum albi is the central unit of the alliance, because of the large amplitude of the occupied habitat types, wide altitudinal range, and the species composition, determined by the dominance of Petasites albus, which is common in the Sudetes Mts., as well as the constant presence of species of the Carpino-Fagetea class. Since the Petasitetum albi is recognized as nomen ambiguum (Koči 2001(Koči , 2007 we propose to restore the name Chaerophyllo hirsuti-Petasitetum albi, because it almost perfectly fits the species combination (most of the species occurring in the type relevé (Sýkora and Hadač 1984) occurred also in the Table 1, col. 3) and ecological characteristics (Hadač and Soldán 1989). Despite that, some authors included this association in the mire vegetation of the class Montio-Cardaminetea Br.-Bl. et Tx. ex Klika et Hadač 1944(Hrivnák et al. 2005, which is not consistent with our knowledge about phytocoenoses discussed here. They are usually present in valleys of streams flowing through deciduous forests, hence the randomized EIVs for nutrients and soil reaction calculated for this association are clearly higher than that for the Prenanthetum purpureae. The latter is similar in terms of habitat but present in acidophilous beech forests (especially Calamagrostio villosae-Fagetum Mikyška 1972 and Calamagrostio arundinaceae-Fagetum Sýkora 1971) and in artificial spruce forests replacing them. Moreover, it is also marked by lower EIVs for temperature, which are connected to the higher elevations the association occupies in relation to the Chaerophyllo hirsuti-Petasitetum albi. Therefore, we think that the separation of these two associations is fully justified as in their original description (Sýkora and Hadač 1984). The species composition of the Prenanthetum purpureae corresponds to the original diagnosis of Bolleter (1921, p. 86, relevés 4 and 5), although several Alpine species (e.g., Aconitum lycoctonum, Ranunculus aconitifolius or Crepis pyrenaica) are absent from the Sudetes Mts.
We propose to include both these associations in the order Petasito-Chaerophylletalia and in the alliance Petasition officinalis, instead of in the Calamagrostietalia villosae and the alliance Arunco dioici-Petasition albi, as proposed by Michl et al. (2010). We decided on the first option based on the concept of Mucina et al. (2016), who consider the alliance Arunco-Petasition albi as restricted to the Alps, while the Petasition officinalis alliance includes tall-herb vegetation on raw alluvia of streams in the upper colline to supramontane belts of the Carpathians and the Hercynicum. Many differential species of this alliance occurred in both associations (Table 1), whereas species from subalpine alliances (Calamagrostion villosae, Adenostylion alliariae) were scarce.
Group 5 embraces montane, nitrophilous phytocoenoses with a dominance of Rumex alpinus and corresponds to the Rumicetum alpini, an association regularly mentioned both in regional studies (Karner and Mucina 1993;Kočí 2001Kočí , 2007Stachurska-Swakoń 2008;Matuszkiewicz 2012) and the supraregional synthesis by Michl et al. (2010). However, Stachurska-Swakoń (2009) split the Carpathian communities with Rumex alpinus between two classes. She distinguished two associations (Aconito firmi-Rumicetum alpini Unar in Unar et al. 1985 andHeracleo palmati-Rumicetum alpini Oltean et Dihoru 1986) with natural character and included them in the class Mulgedio-Aconitetea, while placing the montane synanthropic association Rumicetum alpini in the Galio-Urticetea class. The diversity of these tall-herb phytocoenoses in the Sudetes Mts. is not as high as in the Carpathians and clearly corresponds to Rumicetum alpini. However, similar to Michl et al. (2010), we treat the latter, as belonging to the alliance Rumicion alpini within the Mulgedio-Aconitetea class.
Group 6 includes relevés of mountain tall-grass communities that most often accompany local wetlands, spring zones and stream valleys in open areas. Kočí (2001Kočí ( , 2007 includes them in the Violo sudeticae-Deschampsietum cespitosae (Jeník et al. 1980) Kočí 2001, an association considered endemic to the Eastern Sudetes, which has been synonymized with the Poo chaixii-Deschampsietum cespitosae in the supra-regional study of Michl et al. (2010).
Group 7 represents alpine tall-grass stands in the Karkonosze Mts. The analyzed data was mainly obtained from the literature (Matuszkiewicz and Matuszkiewicz 1974). Fabiszewski and Wojtuń (2001) documented a substantial decrease in the mean number of species per plot from 21.8 to 13.7 in comparison to data collected in the 1970s. This decline was reflected by a complete loss of species such as Aconitum plicatum, Lactuca alpina, Geum montanum, Hieracium alpinum, Hypericum maculatum, Potentilla aurea and a significant decrease in abundance of Adenostyles alliariae, Huperzia selago, Pulsatilla alba and Ranunculus platanifolius. Similarly, Dunajski et al. (2016) reported a 28% decrease in the total number of species within 10 permanent plots. This indicates a significant impoverishment of the composition of the tall-grass phytocoenoses in the Karkonosze Mts., which are probably caused by excessive depositions of atmospheric nitrogen (1138 mg/m 2 /year), favouring the expansion of graminoids, especially Calamagrostis villosa and Avenella flexuosa (Fabiszewski and Wojtuń 2002). In the numerical analysis, the previously described association Bupleuro-Calamagrostietum arundinaceae is not distinguishable and is a part of the cluster of the Crepido-Calamagrostietum. In recent years, phytocoenoses whose species composition relates to the former association have not been examined. Therefore, both its presence in the Karkonosze Mts. and distinctness from the Crepido-Calamagrostietum require further research.
Group 9 includes relevés from the highest parts of the Karkonosze Mts. We classified the communities from this group to the Cicerbitetum alpinae (within the Adenostylion alliariae alliance), which was also reported by Michl et al. (2010) from the Sudetes Mts. The species composition of vegetation plots reported from the Karkonosze Mts. corresponds well to the original diagnosis of Bolleter (1921, Table p. 86, relevés 6-7) with occurrence of Adenostyles alliariae, Lactuca alpina, Rumex arifolius, Geranium sylvaticum, Chaerophyllum hirsutum, Dryopteris filix-mas, Viola biflora, Silene dioica and others. However, similar to the Prenanthetum purpureae, some Alpine species do not occur here. Therefore, distinguishing the separate associations such as Adenostyletum alliariae, Aconitetum firmi and Petasitetum kablikiani (compare Matuszkiewicz 2012) does not seem to be justified.
It should be noted that we had no data of the Arunco-Doronicetum austriaci from our study region (neither own records, nor literature data), which has been reported from other parts of the Sudetes Mts. (Matuszkiewicz 2012). Most of the phytocoenoses with the dominance of Aruncus sylvestris belong to the Arunco vulgaris-Lunarietum redivivae Sádlo et Petřík in , which are synanthropic in the Polish part of the Sudetes Mts., and not directly connected to watercourses. They usually occupy places under gaps in the canopy (e.g., in rich beech forests), forest clearings, or unforested steep slopes, where they form one of the successional stages (in Świerkosz and Reczyńska 2016 as comm. Polygonatum verticillatum-Ranunculus platanifolius; mentioned also in Kącki et al. 2019).

Conclusions
In this comprehensive regional typology of the Mulgedio-Aconitetea in the Polish Sudetes Mts. we distinguished nine associations with relatively clear floristic and ecological separation. For convenience, we largely adopted the higher syntaxa of the current EuroVegChecklist (Mucina et al. 2016), thus, accepting four orders with one alliance each. It should be noted that this "conservative" approach contrasts to our ordination diagram and partly also our vegetation table, according to which there are rather two main groups, namely colline-montane and alpine communities. This differs from the four groups as found in Mucina et al. (2016), and the one as in Michl et al. (2010). The first group embraces phytocoenoses accompanying watercourses of elevations between 200 and 1000 m a.s.l. and with high proportion of species such as Petasistes albus, P. hybridus, Geranium phaeum, Prenanthes purpurea, Aegopodum podagraria, Urtica dioica and, depending on local conditions, also numerous species typical of forests or meadows. The second group includes alpine communities, most often developing above the upper forest zone (1100-1470 m a.s.l., exceptionally lower), in which alpine species dominate, such as Athyrium distentifolium, Aconitum plicatum, Adenostyles alliariae, Ranunculus platanifolius, Lactuca alpina, Rumex alpinus or Senecio nemorensis. These groupings may suggest that the division of alpine communities into three independent orders as adopted by Mucina et al. (2016) will not be confirmed during the analysis of a broader phytosociological material. It will be the task for future syntheses at larger spatial extents to evaluate whether this pattern is a Sudetian idiosyncrasy, or more widespread, and should thus be reflected in a general syntaxonomic scheme.

Data availability
All analyzed relevés (including environmental variables) are available on request through the VESTA Database (Global Index of Vegetation-Plot Databases, ID: EU-PL-004) and the Polish Vegetation Database (Global Index of Vegetation-Plot Databases, ID: EU-PL-001).

Author contributions
K.Ś. and K.R planned the research, K.Ś. and K.R conducted the field sampling, K.Ś. performed the statistical analyses and led the writing, while both authors critically revised the manuscript.