Phytosociology of snowbed and exposed ridge vegetation of Svalbard

Arctic vegetation still lacks a uniform system to classify its plant communities, the current arrangement of the phytosociological data being difficult to interpret. However, using modern methods, it has proved possible to organize the current data to create a suitable system based on numerical algorithms (detrended correspondence analysis and cluster analysis) and traditional methods for 135 phytosociological relevés from snowbed and exposed ridge habitats of Svalbard; 125 relevés were selected from twentieth and early twenty-first century publications, and ten relevés were collected by our group in 2012 from the Kaffiøyra Plain (Oscar II Land, NW Spitsbergen). Our analysis has shown there to be four associations and one community. Two associations and one community are distinguished within the Luzulion arcticae Gjærev. 1950: (1) the Pedicularietum hirsutae ass. nov., (2) the Deschampsietum alpinae (Nordh. 1943) Węgrzyn and Wietrzyk 2015 stat. nov., and (3) the Minuartia biflora community. Two syntaxa were assigned to the Luzulion arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015: (4) the Anthelietum juratzkanae Kobayashi ex Węgrzyn and Wietrzyk 2015 with a confirmed status and (5) the Gymnomitrietum coralloidis Hadač ex Węgrzyn and Wietrzyk 2015 stat. nov. was described by means of a new approach.


Introduction
Since the beginning of exploration of northern polar regions, researchers have paid special attention to the Arctic vegetation as being one of the most sensitive components of the polar terrestrial ecosystems. In the recent decades, rapid response of Arctic vegetation to the ongoing climate change has been observed (Stow et al. 2004;Jónsdóttir 2005;Węgrzyn et al. 2011;Olech et al. 2011).
Observing these changes appears to be easy due to the simplicity of the species composition of plant communities, but despite the relatively small number of species forming the tundra, Arctic vegetation still lacks a complete, uniform, and transparent system that will classify all the communities in terms of the classical plant phytosociology (Braun-Blanquet 1964;Walker et al. 1994;Nilsen and Thannheiser 2013).
Considerable attention to phytosociological research has been paid regarding the Svalbard flora in comparison with other elements of the Arctic vegetation (Elvebakk and Prestrud 1996;Rønning 1996;Walker et al. 2005). Although botanical research of Svalbard has been undertaken since the eighteenth century, it has solely consisted of describing new species and documenting their location (Nilsen and Thannheiser 2013). The first description of the Svalbard plant habitats was by Elton and Summerhayes (1928), and the oldest phytosociological study of the Svalbard tundra was conducted by Hadač (1946). After 1950, the vegetation of the archipelago was mainly examined by Rønning (1965), Eurola (1968), Philippi (1973). In subsequent years, phytosociological studies were continued by Brattbakk (1980), Hartmann (1980), Dubiel and Olech (1990), Möller and Thannheiser (1997), Nilsen et al. (1999), and Möller (2000). The characteristics of the plant community habitats in Svalbard were studied by Elvebakk (1994Elvebakk ( , 2005, and in a broader sense, Walker et al. (1994) assigned the circumpolar vegetation communities to specific habitat types; however, several phytosociological problems emerged in these two mentioned papers, one of which is the assignment of Luzulion arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 to various habitats. Elvebakk (1994) assigned it to snowbed and exposed ridge and Walker et al. (1994) to polar desert and scree vegetation. From the standpoint of environmental factors, scree vegetation can be considered as exposed ridge, but snowbed is a completely different type of habitat (Elvebakk 1994). The current arrangement of phytosociological data available in the literature is perplexing, and researchers have ran into numerous problems when trying to employ it (Nilsen and Thannheiser 2013). Such problems result from (1) the usage of different phytosociological systems, (2) the usage of different size of area of the relevés, (3) the questioned representativeness of the results of phytosociological research for the whole area of Svalbard, (4) the omission of cryptogamic species (mainly lichens), and (5) not taking into account the cryptogamic species as character species of syntaxa, although they are the main element of tundra vegetation. Researchers who have recently attempted to order the phytosociological system of the Spitsbergen communities still show the large gaps existing in Arctic vegetation data (Nilsen and Thannheiser 2013).
Our studies relating to these problems are based on a detailed analysis of our own data and those found in twentieth and early twenty-first century literature by organizing and standardizing information on syntaxa assigned to the snowbed and the exposed ridge. In 2012, a phytosociological study was conducted in the Kaffiøyra Plain (NW Spitsbergen) which is 15 km long and varies in width from 1.5 to 4 km. It is restricted on the eastern side by a mountain range massive in the central part of Oscar II Land and on the western side by Forlandsundet. In the south, its natural boundary is the Dahl glacier, and in the north, it is surrounded by Hornbaekbukta Bay, Aavatsmark glacier, and Dahlbrebukta Bay with the Farmsundet Passage. Seven glaciers come down to the plain from the mountain region. Kaffiøyra Plain is homogenous in terms of vegetation, but erosion processes caused by glacier rivers over the entire terrace area are highly visible (Sobota and Lankauf 2010).
The results initiated a detailed phytosociological analysis of alliances and associations related to the snowbed and exposed ridge. These types of habitats were described by Elvebakk (1994). Ordering the data and creating the hierarchical classification of tundra communities associated with snowbed and exposed ridge were undertaken in accordance with the Zurich-Montpellier School of Phytosociology (Braun-Blanquet 1964) and supported by numerical methods (Sokal and Michener 1958;Hill and Gauch 1980).

Materials and methods
Our studies, based on 10 of our own relevés and 125 relevés from the literature (Table 1), were analysed by phytosociological and numerical methods.
In July 2012, ten phytosociological relevés (each of 1 m 2 area) were taken in the Kaffiøyra plain. The location of each relevé was determined in accordance with the cardinal principle of selecting a uniform and homogenous plant community surface (Braun-Blanquet 1964). The list of species of vascular plants, bryophytes, lichens, and cyanobacteria was obtained for each relevé. Additionally, the degree of cover-abundance and the sociability for each species were defined (Braun-Blanquet 1964). Taxa which were difficult to identify and fragments of biological soil crusts were collected for further determination in the laboratory. All material has been deposited in the Herbarium of the Institute of Botany of the Jagiellonian University in Kraków (KRA).
The 125 relevés selected from publications referred to plant communities that Elvebakk (1994) assigned to two alliances: the Luzulion arcticae Gjaerev. 1950 and the L arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 ( Table 1). The analysed dataset also contained relevés of other plant communities that have a high degree of similarity in relation to those previously mentioned (Table 1). Selection was necessary because many citations, despite describing new syntaxa, did not contain the required phytosociological data (Elvebakk 1994;Weber et al. 2000); these refer to four of the communities listed in Table 1: the Salix polaris-Cetraria delisei com. Hadač 1989, the Saxifraga oppositifolia-Cetraria delisei com. Elvebakk 1985, the Cetraria delisei com. Mattick 1949 andthe ''rødsildre lavmark'' com. Brattbakk 1979. A second problem encountered during the selection process resulted from omitting or grouping together epilithic and soil lichens by some authors for two communities: the Saxifraga oppositifolia com. Möller 2000 and the Cetrarietum delisei Möller 2000. Furthermore, only relevés having an area of 1-10 m 2 were employed for the analysis. Therefore, the relevés of the Cetraria delisei com. Brossard et al. 1984, the Cetraria delisei-Saxifraga oppositifolia com. Nimis 1985, and Saxifraga oppositifolia com. Dubiel and Olech 1990 with a plot area of only 0.01 m 2 were omitted, as were the relevés of ''community C'' com. Barkman 1987, one relevé of Luzulo confusae-Salicetum polaris Hadač 1989, and two relevés of Saxifraga nivalis com. Dubiel and Olech 1990 since their areas were too large (12-100 m 2 ). The localities of selected and analysed relevés are marked on the map of Svalbard (Fig. 1).
The species of vascular plants, bryophytes, and lichens, whose presence was highly questionable, were deleted from the dataset. Cyanobacteria species were not included in the analysis because the majority of other authors also did not take them into account.
For all the alliances and associations separated in the data analyses, a synthetic table containing the constancy and fidelity for each species was made using two programs: TURBOVEG 2000 (Hennekens 2012) and JUICE (Tichý and Holt 2006). Based on these results, the character and companion species were defined (Braun-Blanquet 1964). Nomenclature of syntaxa followed the ''International Code of Phytosociological Nomenclature'' (Weber et al. 2000).
The last stage of the analysis was to prepare the phytosociological table describing the vegetation occurring in the Kaffiøyra Plain. This was made on the basis of ten relevés using TURBOVEG 2000 (Hennekens 2012) and JUICE (Tichý and Holt 2006) software.
After grouping the syntaxa with numerical methods (Figs. 2, 3), they were analysed with a phytosociological method. Table 2 classifies all obtained syntaxa based on the degree of constancy and fidelity of species of vascular plants, bryophytes, and lichens. Relevés belonging to the L. arcticae Gjaerev. 1950 were grouped into two associations and one community. The Minuartia biflora community and the Pedicularietum hirsutae ass. nov. are described for the first time (Tables 2 and 3 Two plant communities were classified into the M. biflora community: the ''community of the Kaffiøyra Plain'' and the Saxifraga oppositifolia-Cetraria delisei ''moserikt'' com. Hermansen 1979. The similarity between these relevés resulted from the participation of such species as: Lecidea ramulosa Th. Fr., Juncus biglumis L., M. biflora (L.) Schinz & Thell., Dicranoweisia crispula (Hedw.) Milde, Saxifraga hieracifolia Waldst. & Kit. ex Willd. and Draba micropetala Hook.; the last four species were considered as differential species for the community (Table 2; Fig. 3.).

Discussion
Previous phytosociological studies were based on traditional, but diverse methods of plant community classification (Du Rietz 1930;Hadač 1946;Braun-Blanquet 1964); with such literature data, it is extremely difficult to create a uniform and transparent system in which described syntaxa will be hierarchically structured, unique, and representative for the entire Svalbard area. In the present study, historical data were subjected to modern numerical analyses, which confirmed several twentieth century syntaxa descriptions and aided the organization of the existing problematic phytosociological system.
The analyses confirmed the existence of two important distinct groups considered to be equal to alliances: the L. arcticae Gjaerev. 1950 and the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 (Fig. 2), the character species of which are presented in Table 2  For the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 these are Dicranum elongatum Schleich. ex Schwägr., Stereocaulon alpinum Laurer, Luzula confusa Lindeb., Aulacomnium turgidum (Wahlenb.) Schwägr., and Psoroma hypnorum (Vahl) Gray. Gjaerevoll (1950) considered the Luzulion arciticae Gjaerev. 1950 to be the equivalent of the Luzulion nivalis Nordh. 1936. In the hierarchical classification of the plant communities of Spitsbergen proposed by Nilsen and Thannheiser (2013), the Luzulion nivalis Nordh. 1936 had a rank of suballiance (and belong to the alliances of the Saxifrago-Ranunculion nivalis Nordh. 1943 em. Dierßen 1984). However, the current work shows that this syntaxon should be raised again to the rank of alliance, due to its nature and distinctiveness of its species composition in comparison with the Saxifrago-Ranunculion nivalis Nordh. 1943 em. Dierßen 1984. It is worth noticing that the authorship of the description of this syntaxon is ambiguous: the publication of Nordhagen (1936) did not contain a description of the Luzulion nivalis Nordh. 1936; furthermore, the name of this alliance did not appear in this publication. However, (Nordhagen 1936) provides a description of an alpine association: the Ranunculo-Oxyrion Nordh. 1936, which probably corresponds to the arctic syntaxon called ''Luzulion nivalis'', as mentioned by Hadač (1989). The correct description of the alliance with the phytosociological data was available in Gjaerevoll (1950) as L. arcticae. Luzula nivalis (Laest.) Spreng. is a synonym of Luzula arctica and the current name of the species (The Plant List 2013). However, we propose using the widely published name L. arcticae Gjaerev. 1950in order to prevent confusion (Weber et al. 2000. Regarding the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015, it was proposed as provisional name by Elvebakk (1985), although it had the rank of alliance and was equal to the following: the Cerastio-Saxifragion cernuae Hartm. 1980, the Saxifrago stellaris-Oxyrion digynae Gjaerev. 1950 all. prov., andthe Ranunculo-Oxyrion Nordh. 1936 p.p. in the hierarchical classification of the Spitsbergen plant communities (Nilsen and Thannheiser 2013). However, Nilsen and Thannheiser (2013) suggest the necessity of further research. Our studies showed the need for separation of the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 from other listed syntaxa. The first syntaxon, the Cerastio-Saxifragion cernuae Hartm. 1980, differed from L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 not only in its character species, but also in a type of occupied habitat. The alliance described by Hartmann (1980) was appropriate only for ornithocoprophilous habitats located in near to bird cliffs, while the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 occurred in exposed ridge habitats. The species composition of the Saxifrago stellaris-Oxyrion digynae Gjaerev. 1950 all. prov. strongly distinguished this syntaxon from the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015. Moreover, various character species occurred in both alliances, such as Oxyria digyna and Saxifraga stellaris. The third syntaxon, the Ranunculo-Oxyrion Nordh. 1936, was similar to the L. arcticae Gjaerev. 1950 in terms of its species composition, but not to the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015.
The habitat analysis is also very important for ordering the alliances into the appropriate class. Our studies suggest that the L. arcticae Gjaerev. 1950 and the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 belong to two different classes. The L. arcticae Gjaerev. 1950 is assigned to the Salicetea herbaceae Br. -Bl. et al. 1947, which represents snowbed vegetation, as proposed by Elvebakk (1994 and Walker et al. (1994), and the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 was classified in the Thlaspietea rotundifolii Br. -Bl. et al. 1947, which includes plant communities of scree vegetation as indicated by Walker et al. (1994) and Nilsen and Thannheiser (2013).
Our syntaxonomic ordering is hierarchic and uses modern statistical analysis. In the existing phytosociology system (Elvebakk 1994) that characterizes the plant communities of snowbed and exposed ridge, there is a problem of synonymous associations and communities describing the same vegetation. Our classification is different from the one proposed by Elvebakk (1994). Plant communities assigned by Elvebakk (1994) to the L. arcticae Gjaerev. 1950 are in our classification grouped into two associations and one community (Table 2; Fig. 3). Similarly, eight plant communities assigned by Elvebakk (1994) to the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 are grouped here into only two associations (Table 2; Fig. 3).
In terms of habitat, the M. biflora community and P. hirsutae ass. nov. refer to two successional types of tundra, as noted by Elvebakk (1979) and Hermansen (1979), but have never been described as associations. The M. biflora community is the earlier, initial stage, occurring on more humid, unregulated substrates, and it is dominated by bryophytes. This stage is dominant in the Kaffiøyra Plain, where the glacier rivers often erode the substrate. Despite our study, we decided not to give association rank to this community since it requires further investigation in the field. The association of P. hirsutae ass. nov. corresponds to the later successional stage with a higher proportion of lichens and vascular plants. This syntaxon occupies more stable and drier substrates; so it can be assumed that as the habitat is stabilizing and drying the M. biflora community is gradually replaced by the P. hirsutae ass. nov.
The D. alpinae (Nordh. 1943) Węgrzyn and Wietrzyk 2015 stat. nov. (Table 2) were described for the first time by Samuelsson (1916) and later revised by Nordhagen (1943). In the hierarchical classification of Spitsbergen plant communities (Nilsen and Thannheiser 2013), this syntaxon was assigned to the Cerastio-Saxifragion cernuae Hartm. 1980 (alliance equal to the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015). The results of the current analyses indicate the necessity of moving the D. alpinae (Nordh. 1943) Węgrzyn and Wietrzyk 2015 stat. nov. to the L. arcticae Gjaerev. 1950, due to the presence of character species of this alliance in the species composition of the association (Table 2). Moreover, this syntaxon occurs on oligotrophic substrates, in this case acidic and wet soils.
The A. juratzkanae Kobayashi ex Węgrzyn and Wietrzyk 2015 seems to be adequate to the Anthelia-Cesia-riche Luzula arcuata-Ass. Nordh. 1928. In the phytosociology classification of Nilsen and Thannheiser (2013), they refer to Nordhagen (1928), but published a different name of syntaxon: the Anthelio-Luzuletum arcuatae Nordh. 1928. However, Nordhagen (1928) did not contain information on the Anthelio-Luzuletum arcuatae, but included a description of Anthelia-Cesia-riche Luzula arcuata-Ass. Nordh. 1928. Despite the suggestions that the Anthelia-Cesia-riche Luzula arcuata-Ass. Nordh. 1928 and the Anthelio-Luzuletum arcuatae Nordh. 1928 are associations, they are insufficiently published according to the ''International code of phytosociological nomenclature'' (Weber et al. 2000). The A. juratzkanae Kobayashi 1988prov. (Kobayashi et al. 1990) was a provisional name; according to our work, this name truly characterizes the described type of vegetation. In terms of habitat, the A. juratzkanae Kobayashi ex Węgrzyn and Wietrzyk 2015 covers scree areas with initial soils and because of that we assigned this association to the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015.
The G. coralloidis Hadač ex Węgrzyn and Wietrzyk 2015 stat. nov. was not included in the classification of plant communities of Spitsbergen (Nilsen and Thannheiser 2013). Hadač (1946) assigned it to the order Salicetalia herbaceae Br.-Bl. ap. Br.-Bl. and Jenny 1926, but the current analyses show considerable similarity to the L. arcuatae Elvebakk ex Węgrzyn and Wietrzyk 2015 in terms of species composition and the presence of character species (Table 2). The substrate of this association is dry and composed of gravel and small pebbles (Dubiel and Olech 1990).
In these studies, we wished to draw attention to the possibility of using historical vegetation data to solve phytosociological problems faced by scientists who are trying to create a uniform system of plant community classification. To achieve this, modern numerical methods were used. In several cases, they confirmed the correctness of schemes proposed by cited authors, who described plant communities using traditional phytosociological methods. However, our analyses have also led to the proposal of a new hierarchical classification of snowbed and exposed ridge plant communities of Svalbard and may hopefully for the foundation for revisions of syntaxa of other habitats.
Further to the hierarchical classification of plant communities of Spitsbergen elaborated by Nilsen and Thannheiser (2013), we would like to present, on the basis of our studies, the new classification of examined syntax as follows: Habitat: Snowbed  (Hadač 1946;Hofm. 1968) Association: Oxyrio-Trisetum spicati (Hadač 1946) Hadač 1989 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creative commons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.