Multivariate analyses were performed to examine changes both in
the environmental conditions and in the structure of the zooplanktonic
community, combining abundance and composition. The environmental
conditions throughout the study period were described by a
principal components analysis (PCA). We created a matrix with 17 environmental
variables for each station (mean values in the upper
100 m of the water column of temperature, Chl-a concentration, dissolved
oxygen concentration, and turbidity; surface and maximum
values of salinity; maximum Brunt–Väisälä frequency value; mixed
layer depth; depth Chl-a maximum; surface values of temperature
and density; maximum values of Chl-a concentration and transmittance;
depths at which were recorded the maximum values of salinity,
turbidity and Brunt–Väisälä frequency; and density difference in the
upper 100 m). A correlation matrix was performed to identify redundancies
between variables, and the original matrix was narrowed
down to the first 9 variables listed before (Fig. 2). PCA was made on
this subset after the variables were mean centered and scaled to unit
variance. A PCA was also carried out on the taxonomic dataset to identify
the temporal variation in composition and abundance of the zooplankton
community and to determine the relative contribution of
each taxon to the changing structure observed. Before the analysis, we
applied the Hellinger transformation, which appears to be the most
appropriate one for linear ordination of community composition data
(Legendre and Gallagher, 2001).