412 
Fishery Bulletin 99(3) 
Table 1 
Summary of collections of Parma microlepis. Sample locations are shown in Figure 1. Sample sizes were n= 5 for juveniles and 
adults at all locations, except for adult fish at Henry Head where n= 7. Lengths refer to standard length in mm. 
Sample no. 
Location 
Date of collection 
Juvenile mean length 
(range) 
Adult mean length 
(range) 
1 
Norah Head 
Mar 1998 
72.0 (56-89) 
125.6 (117-133) 
2 
Terrigal 
Mar 1998 
83.0(61-95) 
122.6 (105-132) 
3 
Barrenjoey Head 
Feb 1998 
64.0 (49-89) 
118.8(106-129) 
4 
Middle Head 
Feb 1998 
88.6 (83-94) 
125.6(115-136) 
5 
Henry Head 
Jan 1998 
83.6 (71-94) 
122.6 (114-136) 
6 
Sutherland Point 
Jan 1998 
85.4(76-93) 
118.6(107-125) 
7 
Jibbon Head 
Mar 1998 
83.2 (65-95) 
126.4(114-136) 
8 
Flinders Islet 
Mar 1998 
76.0 (73-80) 
110.4(105-124) 
9 
Jervis Bay 
Apr 1998 
81.2 (75-95) 
117.8 (112-125) 
10 
Ulladulla 
Apr 1998 
80.8 (76-90) 
111.2(102-120) 
All samples and standards were prepared for ion analy- 
sis in a laminar flow cabinet. Otoliths and eye lenses, and 
scales and spines were individually weighed on an ana- 
lytical balance (to 0.0001 g) and a microbalance (to 0.001 
mg), respectively. All samples were rinsed in 1% nitric ac- 
id for 15 s prior to dissolution. Structures were digested in 
nitric acid (Aristar) either overnight (otoliths and scales) 
or for 36 h (eye lenses and spines). For eye lenses, sul- 
phuric acid (Aristar) was added following dissolution and 
each sample was heated to ~90°C for 1 hour. For all struc- 
tures varying amounts of MilliQ water were then added 
followed by additional dilution with 1% nitric acid where 
necessary. This was necessary so that concentrations of 
the sample solutions were standardized before analysis 
with ICP-MS. 
Samples were analyzed by solution-based ICP-MS (Per- 
kin Elmer SCIEX ELAN 5000). Initially, the four struc- 
tures of fish collected from one location (Henry Head, Fig. 
1) were analyzed in “totalquant II” mode {n = 7 fish). This 
procedure provided a rapid-survey-analysis to semiquanti- 
tatively determine the elements present (and their range) 
within each of the four structures. The instrument was 
calibrated at six points spanning the mass range from 9 to 
209. Accuracy and drift of the machine were determined 
from spiked samples. 
Samples were then analyzed in “quantitative analysis” 
mode after the instrument was calibrated for either exter- 
nal standardization or addition calibration, depending on 
the structure and expected concentration range of each el- 
ement. Measurement parameters for the ICP-MS were as 
detailed in Gillanders and Kingsford (1996). Seven blank 
solutions were run at the beginning of each session to de- 
termine detection limits, which were calculated from the 
concentration of analyte, yielding a signal equivalent to 
three times the standard deviation of the blank signal. 
Blank solutions were 1% HN0 3 for otoliths, scales, and 
spines, and 2% acid (1%HN0 3 and 1%H 2 S0 4 ) for eye lens- 
es. Standards were run through the machine at the begin- 
ning of each session. Spiked samples were run every ten 
samples to measure sensitivity changes to the machine 
through use and for recovery verification. 
All samples were initially analyzed by using addition 
calibration mode. The data were then reprocessed in exter- 
nal calibration mode to obtain concentrations of microele- 
ments. Each sample took around 3 to 4 min to analyze, in- 
cluding a delay before starting to read a sample (70 s) and 
the rinse ( 1% HN0 3 , 0.1% Triton X) between samples of 60 
s. Samples from each structure (otoliths, scales, spines, and 
eye lenses) and size class (juveniles and adults) were ana- 
lyzed together in blocks, but all samples within each block 
were randomized to ensure that resulting patterns did not 
reflect variation in the performance of the instrument. 
Univariate and multivariate techniques were used to 
test hypotheses concerning individual elements and multi- 
element fingerprints of Parma microlepis. Pearson’s cor- 
relations were used to determine the relation between 
otoliths and each of the other structures for individual ele- 
ments (Mn, Sr, Ba, and Pb). Separate analyses were per- 
formed for each size class and structure. 
The match between the otolith microchemistry data and 
the microchemistry data for each of the other three struc- 
tures was examined by using two types of nonparametric 
multivariate analysis. For both types of analysis, the data 
were standardized by subtracting the mean and by divid- 
ing by the standard deviation of the variable, so that each 
element would have equal weight. A matrix of Euclidean 
distances among all pairs of samples was then calculated. 
Mantel’s test (from the R package) and the BIOENV pro- 
cedure (included in the PRIMER computer program; copy- 
right M.R. Carr and K.R. Clarke, Marine Biological Labo- 
ratory, Plymouth, UK) were used to determine the degree 
of correlation between pairs of distance matrices (otoliths 
and each of the other three structures). 
Mantel’s test describes the relation between two dis- 
tance matrices with Pearson’s correlation coefficient, r 
(Mantel, 1967; Legendre and Fortin, 1989). A test of sig- 
nificance for r is obtained by random permutations of the 
replicate sample units for one of the matrices. A total of 
