Biased vegetation patterns and detection of vegetation changes using phytosociological databases. A case study in the forests of the Babia Gora National Park (the West Carpathians, Poland)
Holeksa, Jan; Woźniak, Gabriela
A huge number of phytosociological relevés, which have been collected in Europe during the last century, seems to be a useful point of reference for analyses of temporal changes in vegetation. The high degree of subjectivity during data collection, however, makes difficult the comparison between data sets collected at different time by different authors. Being aware of this we attempted to reveal 30-year changes in strictly protected old-growth mountain forests. We compared two sets of relevés: a subjective phytosociological sample collected in the 60s and a systematic random one from the 90s of the XX century.Standard phytosociological data gave distinct vegetation units with transitional patches poorly represented. This was observed both in the gradient between beech and coniferous forests in the montane belt and in the altitudinal gradient including two forests belts. On the contrary, the random sample did not indicate discontinuity in any of these gradients. We suggest that in spite of better representation of rare plant communities, in the subjective sample a considerable part of vegetation variability was omitted to emphasize differences between associations.Most species positively distinguishing syntaxa were more frequent in the subjective sample because of preferential sampling in patches with a high number of such species. There were few diagnostic species with higher frequency in the random sample or with similar frequencies in both samples. Such results were interpreted as an evidence of spreading of these species. Comparisons between samples revealed the expansion of heliophilous species typical for communities of Betulo-Adenostyletea and dramatic regeneration decline of Abies alba, Acer pseudoplatanus and Picea abies.