A NEW BIOGEOGRAPHICAL CLASSIFICATION OF THE SCOTTISH UPLANDS .2. VEGETATION ENVIRONMENT RELATIONSHIPS

1. This paper describes vegetation-environment relationships in the mountains, moorland and peatlands of upland Scotland. The analysis involves 48 environmental variables, covering location. topography, climate, geology, geomorphology and land-use, and 65 plant communities mapped for 88 sample sites. These environmental variables account for much of the variation in assemblages of plant communities within nine major biogeographical groups of sites. 2. Correlations between environmental variables were investigated independently of the vegetation using principal components analysis (PCA). The first axis defines a gradient of increasing oceanicity from east to west (21.3% variance) and the second axis contrasts high sites forming extensive montane summits with somewhat warmer lower-altitude sites with little or no montane ground (17.5% variance). Precipitation has strong correlations with both axes. Grouse-moor management and small-patch burning are strongly correlated with the least oceanic areas. 3. Site-community data were ordinated without any environmental variables using detrended correspondence analysis (DCA), and then with the environmental variables included using canonical correspondence analysis (CCA). Direct gradient ordination of the vegetation data by CCA shows that these environmental variables are sufficient to account for nearly all of the community variation that can be represented on the first three DCA axes. Species-environment correlations are between 0.91 and 0.93, with a close similarity between the CCA axes and those of the unconstrained DCA. 4. The first 2 CCA axes explain 18% of the total variance in the vegetation with a clear pattern of geographical separation of sites separated on the axes. Axis 1 is strongly correlated with oceanicity (r = 0.68) and the number of wet days (r = 0.80). Axis 2 is strongly correlated with maximum altitude (r = -0.75). Axis 3 is correlated with latitude (r = 0.78) and summer temperature (r = -0.56), and gives a north-south division based on the extent of anthropogenic vegetation, showing that secondary changes have occurred regionally as well as locally. 5. Geology appears to have little influence on the broad-scale pattern of vegetation with the exception of the geologically well-defined Breadalbane hills. 6. The biogeographical analysis described here can be used to evaluate and then select sites of nature conservation interest. The CCA ordination-space can be used to detect sampling bias in groups of sites in relation to environmental variables and to show gaps in their geographical coverage.