Fróðskaparrit - 31.12.2000, Blaðsíða 108

112
DOMINANT SPECIES ABUNDANCE RELATED TO ENVIRONMENTAL
FACTORS ON ROCKY SHORES IN THE FAROEISLANDS
Fig. 2. Partial Correspondence Analysis (partial CA)
for rocky shore communities in the Faroe Islands,
showing the variation unaccounted for by the vari-
ables selected in the CCA (cf. Fig. 1), i.e. exposure,
substrate, fjord index, current and tide. The diagram
shows the species and ordination axes 1 (horizontal)
and 2 (vertical), which explain 26% ofthe total inertia
(= weighted variance) in species abundances. The
eigenvalues ofaxes I and 2 are 0.16 and 0.08, respec-
tively. Species names are given infull in Table 1.
carpus stellatus. The species at the right
were most likely found in conditions at the
opposite end of the scale, i.e. low exposure,
stony or bouldery substrate and/or within
fjords. These included Pelvetia canalicula-
ta, F. vesiculosus, Ascophyllum nodosum,
Littorina obtusata, Cladophora rupestris,
F. evanescens, F. spiralis, Verrucaria mu-
cosa, and Nucella lapillus. Species near the
centre were either favoured by intermediate
levels of these environmental factors, or
they were found in a wide range of condi-
tions. These included Semibalanus bal-
anoides, Palmaria palmata, Patella vulga-
ta, Laminaria digitata, and Mytilus edulis.
In the partial Correspondence Analysis
(partial CA), with the environmental vari-
ables entered as co-variables, the two first
axes explained 17% and 8%, respectively,
of the total species variation (eigenvalues
0.16 and 0.08). In the ordination plot (Fig.
2) Ascophyllum nodosum, Pelvetia canalic-
ulata, Littorina obtusata, Fucus vesiculo-
sus and Cladophora rupestris formed one
group with high scores on the first axis,
while the other species had scores much
closer to zero. The same group of species
was conspicuous on the CCA plot. Partly
due to this, the species sequences along the
two first axes in the partial CA plot reflect-
ed their sequences along the corresponding
axes in the CCA plot (Spearman’s Rank
Correlation Coefficients, rs = 0.67 and -
0.76, respectively, between the two first-
axes and between the two second-axes,
p<0.01 for both correlations). This is an un-
usual result. One possible explanation
might be that the same pattern that was
shown in the CCA was repeated at a differ-
ent scale, i.e. that different sequences of
sites or groups of sites yielded similar se-
quences of species centroids. Another ex-
planation might be that the environmental
variables estimated the underlying factor(s)
imperfectly, and that the partial CA reflect-
ed the unexplained part of the same pattern,
i.e. that there was one main underlying se-
quence of sites reflected in both plots. This
might also occur if an unknown factor, not
correlated to the environmental variables,