Fróðskaparrit - 31.12.2000, Side 118

122
DOMINANT SPECIES ABUNDANCE RELATED TO ENVIRONMENTAL
FACTORS ON ROCKY SHORES IN THE FAROEISLANDS
Comparison with and interpretation ofthe
biological exposure scale methodology
The species abundance curves of the first
and third group resembled those developed
using the biological exposure scale
methodology (Expon) by Bruntse et al.
(1999b). The exact shape of the curves dif-
fered, but they generally depicted the same
trends. For the second group, the curves did
not resemble each other, again demonstrat-
ing that the curves for these species must be
interpreted with considerable caution. In
Fig. 6, the site scores on the first DCA axis
and on the biological exposure scale devel-
oped by Expon are plotted against each oth-
er. The high correlation coefficient (0.93)
demonstrates that the two techniques
arranged the sites quite similarly. The rela-
tionship does not appear to be first-order
linear, however, which may partly explain
differences in curve shapes between the
methods. For instance, DCA separated sites
with low exposure more than Expon. Most
of the differences in the curves between the
methods, however, probably depended on
the curve fitting procedures. In Expon, the
polynomial order was chosen subjectively,
and lower order polynomials were pre-
ferred since they tend to stabilise the itera-
tions. In the curve fitting to the DCA plot,
the default option of Canodraw 3.1 (in: ter
Braak and Smilauer, 1997-1999) was cho-
sen, in which the polynomial order was
chosen automatically based on a signifi-
cance test.
The high correlation between site scores
on the biological exposure scale and on the
first DCA axis implied that the DCA could
be used to interpret the biological exposure
scale for the present data. This would not
necessarily be the case for other data sets.
Expon can model bimodal responses, as
long as they can be approximated by sec-
ond- or higher-order polynomial functions,
while DCA basically relies on an unimodal
response model (e.g. ter Braak, 1995).
The results suggested that the species re-
sponse curves developed for dominant
species on hard substrates in the Faroe Is-
lands by Bruntse et al. (1999b) did reflect
the species responses to wave exposure,
which appeared to be the single most im-
portant factor structuring the species com-
position. The first DCA axis, and, thus,
probably also the biological exposure scale,
was also influenced by substrate and fjord
index. The effects of these variables on
wave exposure could at least partly explain
this influence. Further. the high amount of
species variation accounted for by the first
DCA axis (45%) suggested that the biolog-
ical exposure scale reflected a large propor-
tion of the species variation.
Ejfects of environmental factors at sites
with tidal amplitude 0.4 m or less
The nine sites with tidal amplitude 0.4 m or
less were all from the fjord area to the north
of Tórshavn. The area was atypical in other
respects than tidal range. Four of the sites
were cliffs with slope more than 60°, and
two of the sites were stony beaches. Only
three sites had the “typical” bedrock sub-
strate with slope less than 60°. The se-
quence of species centroids along the first
DCA axis for these sites resembled that for
the other sites, but the axis was mainly cor-
related with the fjord index in contrast to