Begg et al.: Stock discrimination of Scomberomorus queenslandicus and S. munroi 
661 
Table 3 
Jack-knifed cross-validation classification matrix of the 
frequency of assigned cases in each area (and age) used to 
differentiate school mackerel samples. 
Classification of individual 
school mackerel by area 
Area 
Correct 
(%) 
Bowen 
(2) 
Rock 
(2) 
Rock 
(1) 
Moreton 
(2) 
Bowen (2) 
82 
18 
3 
0 
1 
Rock (2) 
50 
2 
9 
1 
6 
Rock ( 1 ) 
64 
0 
3 
16 
6 
Moreton (2) 
47 
1 
4 
5 
9 
Total 
62 
21 
19 
22 
22 
ment concentrations among spotted mackerel of the 
same age among all areas, whereas most elements 
varied significantly between samples of different ages 
(Fig. 6). The samples of 3-year-old fish from Innisfail 
and Hervey Bay had concentrations of P that were 
significantly lower than those in otoliths from Bowen 
fish. Samples from Innisfail also had concentrations 
of K that were significantly lower than those from 
Bowen fish (HSD, P<0.05). No other significant dif- 
ference in elemental concentrations were detected 
between spotted mackerel of the same age. In con- 
trast, significant differences were found for Ba, Mg, 
Na, P, S, and Sr between 1- and 3-year-old fish ( HSD, 
P<0.05), independent of the area from where the 
samples were taken, once again emphasizing the 
strong effect of fish age in determining the results of 
elemental analyses of otolith composition. Scatter- 
plots of the elements showing significant differences, 
with P as the most significant and common variable, 
also showed differences between fish of different ages 
(Fig. 7). In contrast, samples of fish of the same age 
showed no evidence of differences in otolith compo- 
sition among areas. 
Principal component analysis also showed a strong 
separation among spotted mackerel of different ages, 
but no separation of groups by area of collection 
(Fig. 8). One-year-old and 3-year-old samples were 
separated on the first principal component that ex- 
plained almost 50% of the total variation in the data, 
largely on differences in concentration of P, Sr, Ba, 
Na, and S. The second principal component explained 
another 18% of the total variation on the basis of 
differences in K concentration among the samples. 
AN OVA showed significant differences among the PC 
I scores (F=318.32, df=4, 140, P<0.0001), further con- 
firming pooling of spatial samples of the same age. 
Bowen, Innisfail, and Hervey Bay samples of 3-year- 
old fish were all significantly different from Moreton 
Bay and Hervey Bay 1-year-old fish, whereas no dif- 
ferences were found among samples of the same age 
(HSD, P<0.05). 
Spotted mackerel samples were best discriminated 
into two main groups on the basis of statistically sig- 
nificant differences in elemental composition of the 
otoliths: 1) 1-year-old fish; and 2) 3-year-old fish 
(Wilks’s lambda 0.045) (Fig. 9). ANOVA on the CV I 
scores detected significant differences between the 
different aged samples, providing further support for 
these patterns (F=472.09, df=4, 140, P<0.0001). The 
samples of 3-year-old fish from Bowen, Innisfail, and 
Hervey Bay all showed significant differences from 
the 1 -year-old fish from Moreton Bay and Hervey Bay, 
whereas no differences were found among samples 
of the same age (HSD, P<0.05). 
Phosphorus, Sr, Na, Ba, and Mg were the main 
elements that caused the grouping patterns in the 
first discriminant function that accounted for 97% 
of the total variation (Table 4). One-year-old spotted 
mackerel tended to have higher Mg and Na concen- 
trations and lower Ba, P, and Sr concentrations than 
3-year-old fish (Fig. 6). The separation patterns ob- 
served in the individual discriminations resulted in 
only 57% of the spotted mackerel samples being cor- 
rectly classified into their respective groups owing 
to the strong grouping effect of the samples into their 
specific age classes, rather than by area (Table 5). 
Pooling of the results for fish of the same age resulted 
in a classification success of 100% for each age class, 
independent of the area from where the samples were 
collected. 
TafoSe 4 
Discrimination between samples of spotted mackerel 
determined by the pooled within-groups correlations of 
length-corrected elemental concentrations with the sig- 
nificant (P<0.05) canonical variates I, II, and III, and 
the cumulative proportion of the variance accounted for 
by each function, for the seven significant elements 
(Wilks’ s lambda=0.045). 
Canonical variate 
Element 
I 
II 
III 
Ba 
-0.31 
-0.45 
0.06 
K 
-0.06 
0.27 
0.64 
Mg 
0.26 
0.40 
0.23 
Na (length corrected) 
0.33 
-0.30 
0.26 
P 
-0.35 
0.39 
0.50 
S 
-0.19 
-0.38 
0.30 
Sr (length corrected) 
-0.34 
0.13 
-0.34 
Cumulative proportion 
0.97 
0.99 
1.00 
