Vegetation data collection
We sampled the vascular plant vegetation from all the permanent plots in mid-late July (when most plants were flowering) for each of the years from 2004-2007. All plots were sampled every year, except for one plot in 2006 and three plots in 2007 due to inadvertent human disturbance, for a total of 140 plot-level samples. Sampling was done at the sub-plot level: the nine sub-plots of each permanent plot were fully sampled for species presence/absence data. Species were identified based onDen nya nordiska floran (Stenberg and Mossberg 2003); if an individual plant could not be identified based on what remained it was not recorded; some plants could only be identified to the genus level (e.g. Taraxacum sp.) and were recorded as such. In the centre sub-plot of each permanent plot we sampled relative species abundance based on plant cover. We used a 50cm x 50cm transparent plexi-glass table with 100 randomly distributed 4mm-diameter holes: a pin of the same diameter was lowered through each hole and the number of contacts the pin had with each species was recorded (see also Olofsson et al. 2001).