Gillanders: Trace metals in four structures of fish and their use for estimates of stock structure 
413 
999 permutations were performed for each test and the 
pi-obability calculated as the number of values equal to, or 
larger than, the observed value of r divided by the total 
number of permutations. 
If significant relationships were found with Mantel’s 
test, the BIOENV procedure was then used to determine 
which elements were likely to be important in describing 
the correlation between the distance matrices (Clarke and 
Ainsworth, 1993; Clarke and Warwick, 1994). BIOENV 
calculates rank correlations between a dissimilarity ma- 
trix derived from the otolith microchemistry data and ma- 
trices derived from various subsets of the elements for ei- 
ther the scale, spine, or eye lens data matrices, thereby 
defining suites of elements that “best explain” the otolith 
microchemistry data (Clarke and Ainsworth, 1993). The 
harmonic or weighted Spearman correlation p w was used. 
Results 
Otoliths, scales, spines, and eye lenses of Pcinna microl- 
epis differed in composition, both in terms of the actual 
elements present and the concentration of individual ele- 
ments. Otoliths, scales, and spines were dominated by 
calcium, whereas no calcium was detected in eye lenses. 
The microelement Sr was present in otoliths, scales, and 
spines at concentrations of 100’s to 1000’s pg/g of struc- 
ture, whereas it occurred at low concentrations in eye 
lenses. Other elements (e.g. Mn, Ba, Pb) were typically 
found in low-to-trace levels in all structures. Mercury was 
found in detectable concentrations only in eye lenses. 
Univariate analyses 
Differences in the concentration of manganese among 
structures varied over several orders of magnitude (range: 
0.05 for eye lenses to 100 pg Mn/g structure for spines). 
Concentrations of Mn in otoliths showed significant corre- 
lations with concentrations of Mn in both eye lenses and 
scales of juvenile fish, but only with scales from adult fish 
(Fig. 2, Table 2). For juveniles, fish from one site may 
have influenced the correlation analyses; therefore fish 
from this site were removed and the correlations recalcu- 
lated. This adjustment resulted in a significant correla- 
tion between Mn in otoliths and Mn in scales, as found 
previously. The relation between Mn in otoliths and Mn 
in spines was also significant when fish from this one site 
were removed (r=0.302, P<0.05). However, there was no 
longer a significant correlation for eye lenses: fish from 
one site may therefore have influenced the correlation 
result for this structure. 
Concentrations of Sr in otoliths showed significant cor- 
relations with concentrations of Sr in all the other struc- 
tures for juvenile fish; however the relation was only posi- 
tive for otoliths and spines (Fig. 3, Table 2). For scales and 
spines, fish from one site did not unduly influence correla- 
tions; when these fish were removed from analyses, cor- 
relations were still significant. However, as found for Mn, 
the correlation between Sr in otoliths and Sr in eye lenses 
was no longer significant when fish from this site were re- 
Table 2 
Correlations between concentration of elements in otoliths 
and concentration of elements in eye lenses, scales, and 
spines of juvenile and adult fish collected from ten loca- 
tions along the coast of New South Wales. Shown are 
Pearson’s correlation coefficients. Significance levels are 
indicated by * (P<0.05) and ** (P<0.01), df=48. No correc- 
tions were made for experiment-wise error rate; therefore 
1 in 20 tests would be expected to be significant by chance 
alone. 
Otolith 
Mn 
Sr 
Ba 
Pb 
Juveniles 
Eye lenses 
0.3410* 
-0.3787** 
0.1260 
Scales 
0.4018** 
-0.6237** 
0.8810** 
Spines 
0.2466 
0.6586** 
0.9001** 
Adults 
Eye lenses 
0.0429 
0.2353 
0.0239 
-0.2564 
Scales 
0.4717** 
-0.0051 
0.8839** 
-0.1487 
Spines 
0.2696 
0.8446** 
0.9121** 
-0.0732 
moved. The only significant relationship for Sr in adult 
fish was found between otoliths and spines (Fig. 3, Table 
2 ). 
Significant positive relationships were found between 
the amount of Ba in otoliths and the amount of Ba in 
scales and spines of both juvenile and adult fish (Fig. 4, 
Table 2). Lead was usually at or below detection limits of 
the instrument for juvenile fish and therefore correlations 
were not made. Correlations between the amount of Pb in 
otoliths of adult fish and the other structures showed no 
significant relationship (Fig. 5, Table 2). 
Multivariate analyses 
Mantel’s test detected a significant relationship between 
the Euclidean distances among replicates based on the 
otolith data and the distances based on data from eye 
lenses, scales, or spines for juvenile fish (Table 3). For 
adult fish, however, a significant relation was detected 
between the otolith data and the scale and spine data 
(Table 3); thus there may be either changes in assimilation 
of elements with age or resorption and remineralization 
may occur in some structures, for example in eye lenses. 
The BIOENV analyses showed that the highest rank 
correlation was found between the otolith data and the 
scale data for juvenile fish and involved Sr and Ba (Table 
3). For adult fish the highest correlation was between the 
otolith data and the spine data and involved three ele- 
ments (Mn, Sr, and Ba; Table 3). 
Comparisons of juvenile and adult fish for each struc- 
ture showed no significant relationships (Table 3). The 
Mantel test, however, was marginally nonsignificant for 
otoliths (Mantel r statistic 0.2000, P=0.060). 
