Jacobson et al.: Use of parasites to clarify migration of Sardinops sagax 
203 
of Pacific sardine from the 5 different regions were ana¬ 
lyzed separately for the different size categories, following 
the results of the pairwise comparison on size categories 
described above. For Pacific sardine <210 mm SL (the 
combination of the 2 smallest size categories), the parasite 
communities collected off British Columbia were signifi¬ 
cantly different from those from all other sampled regions 
of the CCS (all comparisons: P<0.01) (Table 5). No other 
regional comparisons were statistically significant for this 
size category of Pacific sardine. For Pacific sardine 210- 
219 mm SL, the parasite communities of fish from British 
Columbia were significantly different from those collected 
off Washington and Oregon (pseudo-£=2.41, P<0.01) and 
off Northern California (pseudo-£=1.89, P=0.01). P-values 
for pairwise regional comparisons with PERMANOVA 
between Pacific sardine in this size category from British 
Columbia and Central and Southern California were 0.08. 
For Pacific sardine >220 mm SL, a significant difference 
between parasite communities was found between fish 
caught off British Columbia and those collected off Wash¬ 
ington and Oregon (pseudo-£=1.71, P=0.04) (Table 5). 
Because the pairwise PERMANOVA for regions showed 
no differences among our regions of California, we per¬ 
formed a CAP analysis with those regions condensed 
into one California region. The CAP analysis further sup¬ 
ported significant differences among all regions for the 
Pacific sardine <210 mm SL (tr=0. 63182, P=0.0001). The 
selected orthonormal PCO axes (m=3 axes) described 89% 
of the variation in the data cloud. Cross-validation results 
of this CAP model show the distinctiveness of the Pacific 
sardine <210 mm SL collected off British Columbia with 
correct allocation to their own group at 76.5% and of the 
Pacific sardine <210 mm SL collected off California with 
correct allocation to California at 82.2% (Table 6). Similar 
results of CAP analysis were observed with Pacific sardine 
210-219 mm SL (tr=0. 67626, P=0.0001) and >220 mm 
SL (tr= 0.50237, P=0.0001). Correct allocations to British 
Columbia were 80% and 70% for these 2 size categories of 
large Pacific sardine when the m =3 and m =4 orthonormal 
PCO axes, respectively, were used. The correct allocation 
of Pacific sardine >220 mm SL back to California was also 
high at 80%. Correct allocations to Washington and Ore¬ 
gon were fewer, with assignments also going to both Brit¬ 
ish Columbia and California, but were greater than the 
33.3% that would be expected from chance alone, except 
for fish >220 mm SL (Table 6). 
Regional differences in abundance of specific parasites 
observed among Pacific sardine played a differential role 
in defining regional parasite communities. For example, 3 
trematode species, L. gibossus, Pronoprymna petrowi, and 
Parahemiurus merus, were more abundant in samples 
from British Columbia than in those from other regions 
(see shade differences presented in Figure 3). In con¬ 
trast, another trematode, M. ecaude, had high abundance 
in Pacific sardine from California regions. Abundance of 
Anisakis spp. increased in concert with increasing size of 
Pacific sardine from all regions, but especially among large 
sardines collected off Central and Southern California. 
In all size categories of Pacific sardine caught off Brit¬ 
ish Columbia, the trematode L. gibossus provided the 
highest percent contribution to the parasite communi¬ 
ties, from 71.5% in the Pacific sardine <210 mm SL to 
Table 5 
Results of pairwise comparisons with permutational multivariate analysis of variance 
(PERMANOVA) between parasite communities of Pacific sardine (Sardinops sagax) 
caught from 2005 through 2008 in 5 regions of the California Current System and ana¬ 
lyzed separately by size category: <210 mm standard length (SL), 210-219 mm SL, and 
>220 mm SL. The regions are British Columbia (1), Washington and Oregon (2), North¬ 
ern California (3), Central California (4), and Southern California (5). The pairwise 
PERMANOVA pseudo-1 test statistic is a direct multivariate analogue to the univariate 
^-statistic. This value is provided with a permutation P-value. 
Regional 
comparisons 
<210 mm SL 
210-219 mm SL 
>220 mm SL 
Pseudo-^ 
P 
(perm) 
Pseudo-i 
P 
(perm) 
Pseudo-f 
P 
(perm) 
1,2 
2.60 
<0.01 
2.41 
<0.01 
1.71 
0.04 
1,3 
3.69 
<0.01 
1.89 
0.01 
1.50 
0.12 
1,4 
3.08 
<0.01 
1.52 
0.08 
1.36 
0.20 
1,5 
2.85 
<0.01 
1.67 
0.08 
1.40 
0.95 
2,3 
1.32 
0.17 
0.59 
0.77 
0.38 
0.95 
2,4 
1.25 
0.06 
0.75 
0.55 
0.47 
0.83 
2,5 
1.66 
0.22 
1.18 
0.24 
0.53 
0.75 
3,4 
1.36 
0.17 
0.77 
0.66 
0.60 
0.75 
3,5 
1.35 
0.17 
0.85 
0.67 
0.71 
0.50 
4,5 
0.81 
0.50 
0.57 
0.90 
0.63 
0.85 
