Page 3 - Cubadda2001

F. Cubadda et al. / Chemosphere 45 (2001) 561±569
Table 1
Shell sizes, body weights and mean wet weight (WW)/dry weight (DW) ratio for the organisms analysed 1.27
Species Shell size range (cm)
M. turbinata 1.0±3.0a
M. mutabilis 1.1±2.8a
P. caerulea 1.0±3.1b
P. lusitanica 1.1±3.0b
a Height.
b Length.
zinc analysis was performed by ¯ame atomic absorption
spectrometry (FAAS) with duplicate determination of
each sample. Whenever low levels occurred, copper was
determined by GFAAS.
The detection limits, coecients of variation and the
results obtained for the BCR-certi®ed reference material
278 (mussel tissue) ± included in the analytical runs to
test the accuracy of the analytical method ± are reported
elsewhere (Campanella et al., 2001). As regards analyt-
ical accuracy, for all metals the measured values were
equal to 91±106% of the certi®ed values in the CRM
used.
2.3. Statistical analysis
According to Boyden (1974, 1977), in shell®sh the
concentration of a given metal is related to body weight
as a power function. This relationship can be linearized
through a logarithmic transformation, which yields to
an equation of the form
log10‰MeŠ ˆ a ‡ b log10 W ;
where ‰MeŠ is the trace metal concentration, W is body
weight, a is equal to the logarithm of the multiplicative
coecient of the power function, and b (i.e., the slope of slopes was not met and the adjusted means were calcu-
the linear function obtained) coincides with the expo-
nent of the power function. In order to achieve this
linear relationship, all data (concentrations and dry
body weights) were log-transformed prior to subsequent
analysis. The single log±log plots thus obtained for each
species, metal and station were then examined for out-
lying values and the best ®t regression lines were calcu-
lated by the method of least squares.
Signi®cant di€erences between metal concentrations
in molluscs from di€erent stations were tested by AN-
COVA on transformed data with body weight log as
covariate. Normality and homoscedasticity, which are
required for the application of parametric methods, were
checked and resulted to be substantially improved by the
double log transformation. Parallelism of the regression
lines was tested throughout at a P level of 0.05. When
this condition was not met ANCOVA was not per-
formed and the metal levels in di€erent stations were
calculated from the regression line of each station at
563
WW/DW
1.27
DW range (g) DW mean (g) 1.26
1.32
0.049±1.551 0.442
0.055±1.081 0.351
0.050±0.808 0.228
0.096±1.019 0.306
standardized weights, using the overall mean value of
body weight for the relevant species. In all cases multiple
comparisons between stations were carried out by the
least signi®cant di€erence (LSD) method. The LSD test
was used also for comparisons between species.
Statistical analysis was performed using the package
STATISTICAÓ for Windows (Release 4.5, StatSoft,
1993).
3. Results
In all the examined molluscs species metal levels
turned out to be dependent on body size. The linear
correlation between log-transformed concentrations and
dry weights was highly signi®cant P < 0:001† in all lo-
cations whatever the metal considered. The test for
parallelism showed that generally the slopes of the re-
gression lines (b) were not signi®cantly di€erent from
one location to the other and thus mean values of log-
transformed concentrations at standardized weights
could be calculated by ANCOVA. These weight-ad-
justed means were then back-transformed and are re-
ported in Tables 2 and 3, together with ranges of the
original (untransformed) metal concentrations. In two
cases (Pb in M. turbinata, Cr in P. caerulea) equality of
lated as described in the previous section.
From data in Tables 2 and 3 it is evident that the
species belonging to the same genus exhibit similar ac-
cumulation patterns. In M. turbinata and M. mutabilis
metal concentrations decrease in the following order:
Zn > Cu > Cd > Cr  Pb, while in P. caerulea and
P. lusitanica the sequence is: Zn > Cd > Cu > Cr  Pb.
With regard to the latter two metals it may be observed
that in stations 1±4 chromium concentrations match
those of lead, while in station 5 lead levels are de®nitely
higher. The species of the genus Patella are stronger
accumulators of cadmium (P < 0:001), with P. caerulea
reaching signi®cantly higher levels than P. lusitanica
(P < 0:001). On the contrary, the species of the genus
Monodonta show higher levels of copper and zinc in
their tissues (P < 0:001).
The results of the regression analysis are reported in
Table 4. Only positive correlations between metal