Search

This paper presents the results of a study of surface pollen deposition on 26 glacier forelands in the Jostedalsbreen–Jotunheimen region of southern Norway. Numerical techniques, including two-way indicator species analysis (TWINSPAN), detrended correspondence analysis (DCA) and canonical correspondence analysis (CCA), are used to identify distinct plant assemblages, to examine the pollen–vegetation relationship and to distinguish broad trends in the data. The source of pollen is fundamental; the majority of samples, especially those from sparsely vegetated sites, are dominated by arboreal pollen (up to 90% of the total land pollen (TLP) sum), most with a long-distance source. However, indicator taxa, notably Salix and Empetrum, although present at low frequencies, can hold the key to the true nature of the local vegetation. Indicator taxa produce strong correlations between their presence in the vegetation and representation in the pollen spectra, at times essential to distinguish plant communities. Multivariate analysis of the 197 surface pollen samples and vegetation data indicates the broad division of the sub-alpine and alpine vegetation into four major groups: pioneer communities, snowbed communities, heath communities and woodland. The primary DCA ordination axes show significant correlations with altitude and terrain age (e.g. correlation r = 0.36 between altitude and non-arboreal pollen (NAP) Axis-1). The most readily interpretable results are produced by CCA simultaneous ordination of vegetation data and NAP data. The potential for improving the interpretation of Holocene vegetation is assessed. [ABSTRACT FROM AUTHOR]