FOR ONLINE PUBLICATION ONLY
APPENDIX 1: CAMERA DERIVED SUPPORTING MATERIAL
In the summer of 2012, we deployed automatic wildlife cameras (ScoutGuard SG560K-8M
Infrared Digital Camera, ScoutGuard, Professional Trapping Supplies, Queensland, Australia) to
obtain time series of the presence of reindeer and other animals, and also to monitor changes in
snow depth on the plots. Two cameras were mounted on poles on each exclosure, one facing
inwards into the exclosure for monitoring snow depth (“snow cameras”) and one facing out over
the control plot for monitoring animal presence (“animal cameras”). Snow cameras were
programmed to take photographs every day from 06:00 to 18:00 at 3-hour intervals, whereas the
animal cameras were programmed to take photos around the clock at 30-minute intervals.
Pictures taken during daylight were standard RGB images, whereas an infrared system was used
during hours of darkness. Although these camera deployments do not cover the period of study
described in this paper, it nevertheless provides information supporting some of our assumptions
and conclusions.
Firstly, differences between plots in the timing or length of the season of snow cover
were negligible (Figure A1:1). Although a solid snow cover was established on all plots on the
same day (Oct 25th 2012), the time at which 50% of the ground monitored by the images had
become snow free varied by up to 3 weeks. However, there was no systematic and significant
difference between the Finnish and Norwegian sides of the border (Mann-Whitney Udf=12 = 52.5,
p=0.4205). Secondly, there were clear signs that reindeer were present to a much greater degree
on the Finnish side. Importantly, substantial numbers of pictures of reindeer were recorded on
the Finnish side during the spring and summer, whereas the much smaller number of images with
reindeer on the Norwegian side were generally recorded during autumn and winter (Figure 1:1).
Taken together, these two factors support our conclusions that there are no differences in weather
conditions and snow regime between the two sides of the border, and that the observed
differences in birch stand structure and albedo are most likely explained by differences in the
timing and intensity of reindeer grazing.
Figure A1:1. Histograms of the number of images in which reindeer were observed within each control plot. The
left-hand column represents plots on the Finnish (YRG) side whereas the right-hand column represents Norwegian
(WG) plots. Each bar represents one week. The vertical grey lines and horizontal arrow indicates the period of snow
cover.
Appendix 2: Additional figures and tables
Figure A2:1. Map of the study area in Polmak. The dark grey elevation contours are based on the 10x10 meter
Digital Elevation Model used to determine the elevation of study plots and satellite imagery grid cells. The light
grey grid represents the 250x250 meter raster at which NDVI data were available. Albedo data were available at a
similar 500x500 meter grid (not shown). These grids were both subsampled to 50x50m grids to better follow the
outlines of the country polygons. Albedo and NDVI data from grid cells where the center point fell within either of
the two country polygons were considered for analysis. The relative weighting scheme used to calculate weighted
summary statistics (mean and standard deviation) of albedo and NDVI within the Finnish and Norwegian polygons
respectively is shown as a color gradient within the polygons. The weighting scheme is based on 1) distance from
each grid cell to the centroids of the study plots on the Finnish and Norwegian sides (shown in black), and 2) the
absolute difference in elevation between each grid cell and a reference elevation (the average of the elevation at the
two centroids). The color scale indicates the relative weight, running from low (blue) to high (red).
Figure A2:2. Defoliation caused by the last geometrid outbreak (Jepsen and others 2009) in the two country
polygons. A) Weighted means and standard deviations of defoliation scores (% decrease in summer NDVI
compared to a reference year without defoliation, Jepsen and others 2009) by year for Finnish (YRG) and
Norwegian (WG) sides, and B) mean pixel-wise defoliation score across all years of the outbreak.
Table A2:1. Comprehensive List of All Variables Measured at Various Spatial Levels
Soil
Conductivity
Vegetation
Low forbs
Chamaepericlymenum
suecicum (Cs)
Trientalis europaea (Te)
Rubus chamaemorus (Rc)
Tall woody species
pH
Betula pubescens (Bp)
Calcium
Magnesium
Sorbus aucuparia (Sa)
Salix spp. (Salix)
Potassium
Dissolved
Phosphorous
Total Nitrogen
Alnus spp. (Alnus)
Total Carbon
C/N ratio
Bacterial OTUs
Bacterial diversity
Fungal OTUs
Fungal diversity
Juniperus communis (Junip) Avenella flexuosa (Af)
Grasses others (Grass)
Low woody species
Empetrum nigrum (En)
Cryptogams
Vaccinium myrtillus (Vm)
Tall ferns (Tferns)
Vaccinium vitis-idaea (Vv)
Small ferns (Sferns)
Vaccinium uliginosum (Vu) Equisetum spp. (Equi)
Arctous alpinus (Aa)
Lycopodium spp. (Lyco)
Tall forbs
Solidago virgaurea (Sv)
Mosses and Lichens
Chaemerion augustifolium
Mosses (Moss)
(Chae)
Tall forbs, others (Forbs)
Lichens (Lichens)
Linnea borealis (Lb)
Betula nana (Bn)
Ledum palustris (Lp)
Trees
Number of stems
Environment
NDVI
Number of basal shoots
Albedo
Number of browsed basal shoots
Saplings (<0.2)1
Elevation
Saplings (0.2-0.5)1
Saplings (0.5-1.3)1
Grasses
Number of functional individuals2
Tree distance2
If not otherwise indicated, soil variables were measured at plot level, vegetation at vegetation quadrat level, tree variables at tree level, herbivores at vegetation plot
level, and environmental variables (NDVI and albedo) at the overall study region level with a contrast between the Finnish and Norwegian sides.1Recorded along plot
diagonals; 2Recorded from Kite Aerial Photographs