Variation in Fennoscandian calciphile alpine vegetation. Are previous phytosociological classifications reproduced by numerical analyses?
Calciphile alpine vegetation has been a popular study object for more than a century, and numerous vegetation communities have previously been described and classified by phytosociological methods. Such available data give opportunity to explore three main questions in vegetation ecology: (1) Is it justifiable to separate vegetation types (clusters) in a nature that is assumed to vary along continuous gradients, and (2) if so, to what extent can separated phytosociological vegetation units be reproduced by a numerical classification, and (3) what are the main floristic gradients separating the communities? This study is based on data from 218 previously described communities from approximately 2270 relevés and more than 1180 vegetation stands from different parts of Fennoscandia. The sampling unit used here is plant communities, and an importance value (IV) was calculated for all species (taxa) based on their frequency and mean cover. DCA axis 1 is interpreted as a “biomass or competition axis separating tall herb vegetation from snow bed vegetation. DCA axis 2 is interpreted to represent a complex gradient with increasing length of snow cover and altitude separating snow beds, lee side vegetation and exposed vegetation. Twenty-two groups of communities were separated by the TWINSPAN classification program. Most of them show major overlap in the ordination space, and it gave limited support to previous phytosociological classifications. Communities in lee sides, snow beds and on exposed sites were, however, fairly well separated. Data on average altitudinal position and soil pH from some of the communities showed significant relationships with the community scores along the DCA axes. At high elevations the floristic gradients were strongly reduced; the lee side communities disappeared and the distinction between snow beds and communities on exposed sites became diffuse.