Fróðskaparrit - 01.01.2007, Side 125

GRÓÐURIN Á FLAGTEKJUM í FØROYUM OG Á GRASLENDINUM UTTANUM
- EIN KANNING ÚR SANDOYNNI
123
field samples. This proximity reflects the
fact that there is low species diversity and
that the original turfs may have come from
the infield. This is, of course, in contra-dis-
tinction to Ostenfeld’s (1905-08) observa-
tion that the turfs are removed from out-
field sources. There is likely to have been
much variation in such practices in the past
and the present situation may not follow
the same pattern anyway.
Although some of the grass roofs are
subjected to grazing, the insignificant re-
sult obtained by the z-test suggests that this
factor is not responsible for differences in
diversity and tliat the second hypothesis is
not therefore supported. In addition, graz-
ing, slope and aspect appear to have no
correlation with either of the DCA axes in
the turf-roof ordination (Fig. 5). Slope and
aspect only become significant when the
ground, and especially the mire, vegetation
is included in the ordination data set. Slope
differentiates between the poorly drained
nrire samples to the right of Figure 6, and
grazing separates the heath-like outfield
from the infield vegetation. Despite an ap-
parent hydrological gradient on this axis,
improvement to the infield areas is respon-
sible for this distinction and explains the
position of these samples to the left of the
relatively unimproved roof turfs.
The absence of a correlation between
aspect, grazing, slope and the DCA axes,
when the roof data alone are analysed, sug-
gests that an unmeasured variable is re-
sponsible for the composition of the turf-
roof vegetation. In light of this, the refuta-
tion of hypotheses two and three, and com-
ments by previous authors, it would be rea-
sonable to assume that summer drought is
involved in the low species diversity. This
may be expected to correlate with the slope
of the roof and its aspect - in the absence of
such a relationship, it may be presumed
that the effect of a rapidly-draining sub-
stratum in an exposed situation is of
greater significance than either of these
two variables.
Soine of the variation that does exist
within the turf-roof vegetation, appears to
be attributable to the senescence of the
dominant grass species and the condition
of the roofs, but this could not be tested in
the absence of the necessary information.
Observation of abandoned, turf-roofed
buildings suggests that as they age or fall
into disrepair, the roofs lose their dense
grass cover and cryptogams, such as
lichens of the genus Peltigeria sp. or hyp-
naceous mosses such as Rhytidiadelphus
sqiiarrosus and Isothecium myosuroides,
become increasingly prevalent. In addi-
tion, shoots of the grasses Anthoxanthum
odoratum and Holcus lanatus become
more frequent amongst those of the
favoured Festuca rubra víiv. fratercula and
Poa trivialis as well as a number of herbs,
especially Rumex acetosa and Thymus
praecox. The diversity of the turfs in-
creases as a more heterogeneous structure
becomes available for colonisation by a
range of species. The ensuing abundance
of ciyptogam and herbaceous species that
are able to withstand drought conditions
lends further support to the acceptance of
hypothesis one, that the low diversity of
the roofs is a result of summer drought
conditions.