166 
from this site. The results are shown in Appen- 
dix Table 2. Carex scirpoidea from the dry 
heaths and S. rotundifolia from the snowbeds 
constituted 26.4 and 13%, respectively, of the 
sample compositions. Also noticeable were 
lichens, mainly 7Thamnolia vermicularis, which 
comprised about 5.7% of the diet. The relation 
between green and dead grass-like species was 
65% green Cyperaceae and 35% dead 
Cyperaceae. 
Site 3: July 23-27. This site was located in 
an area described as a wet meadow with a high 
water table early in the season, but drier with 
emerging green by the end of July. It was 
typically situated along brooks on sandy ground 
with a rich growth of willows and herbs. The 
dominant species was Dupontia fisheri. This 
vegetation type is situated close to the water, 
intermediate between the Eriophorum angusti- 
folium type and the Carex aquatilis types. Six 
EF samples were collected from this site. Appen- 
dix Table 3 shows the botanical composition: 
grass-like plants (mainly family Cyperaceae) con- 
stituted 42.6% of the sample, with C. aquatilis 
and &. angustifolium at 11.7 and 17.1%, 
respectively. Salicaceae constituted 24.7% of the 
samples and herbs 18.3%. A wider variety of 
plant groups was chosen from this than other 
sites, and the amount of dead material was very 
small. 
Site 4: July 28. This site was located in the 
sand dunes area near the coastline. Two EF 
samples were collected (Appendix Table 4), Sa/ix 
ovalifolia at 79% was the main plant species. 
Most scattered herbs were picked. Reindeer fed 
by moving from one plant cluster to another. A 
low content of dead material was noted in the 
samples. 
Direct comparison between compositions of 
EF egesta and available herbage was difficult 
because reindeer fed by slowly walking and 
nibbling at the ground cover in a non-random 
fashion. During a collection period, the animals 
might move over several vegetation types. Also, 
the distribution of plant species within vegeta- 
tion types was commonly clustered, and a valid 
quantitative evaluation of plant species selection 
is rather dubious. However, 33x100 cm plots 
were subjectively laid out and analyzed along 
the approximate route of movement of the 
reindeer after the termination of EF collections. 
Eriophorum-angustifolium (marsh) 
Vi. oD 
Carex aquatilis (marsh) 
os 
Dryas integrifolia (heath) 
Qh) es o_O 
Dupontia fisheri (brook meadows) 
mW 
Salix ovalifolia (sand dunes) 
(%) 
N 
oO 
(o>) 
je) 
NO 
j=) 
CHOMMEDRONSHINt smOnn 
S 
(>) 
oO 
60 
Shrubs Sedges/ Herbs Lichen Mosses Litter 
grasses 
ZZ [ea 
Vegetation plot Fistula sample 
Fig. 8. Frequency of occurrence of plant groups 
in esophageal fistula samples compared with 
vegetation grazed by reindeer. 
Fig. 8 shows the approximate preferences for 
the systematic plant groups from each study 
site. Each bar of the histogram represents the 
percentage composition of the particular plant 
groups from each sample. The horizontal cross- 
lines represent the median value. Gaare (pers. 
comm.) suggested the use of the median value as 
a better representation of the average than the 
mean when outlying observations occur, as in 
this case. 
From three study sites there was a median 
preference for the grass-like groups, respectively 
from the D. integrifolia-S. rotundifolia type, the 
E. angustifolium type, and the D. fisheri type. 
From the S. ovalifolia sand dunes there was 
preference for S. oval/fo/ia. f 
Table 10(a) shows the overall composition 
of 20 EF samples irrespective of vegetation type. 
Collectively, they constitute the main compo- 
nents of the diet in early, mid- and late July. 
