202 
Fishery Bulletin 117(3) 
those from all size categories from all other 
regions (Fig. 2A). This nonmetric MDS plot also 
shows the trend of the differences in parasite 
communities among size categories of Pacific sar¬ 
dine within each region, except for those from 
Central and Southern California. 
An additional nonmetric MDS ordination 
includes the parasite communities of the 3 north¬ 
ern anchovy collections with those of Pacific sar¬ 
dine (Fig. 2B). Figure 2B shows the high similarity 
of the parasite community of northern anchovy 
collected off Grays Harbor to those of Pacific sar¬ 
dine collected off British Columbia. The parasite 
community of northern anchovy collected off Wil- 
lapa Bay was most similar to parasite communi¬ 
ties of Pacific sardine 200-209 and 210-219 mm 
SL from Washington and Oregon and of Pacific 
sardine 210-219 mm SL from Northern and Cen¬ 
tral California. The sample of northern anchovy 
collected off Port Hueneme was most similar 
to Pacific sardine 200-209 mm SL collected off 
Southern California. The overlays on this non¬ 
metric MDS plot show clustering of all parasite 
communities south of British Columbia at 60% 
similarity and clustering in 4 groups of parasite 
communities at 70% similarity. The 4 groups 
are composed of communities recovered from 1) 
Pacific sardine from British Columbia and north¬ 
ern anchovy from off Grays Harbor, 2) northern 
anchovy from Southern California and Pacific 
sardine 200-209 mm SL from Southern Califor¬ 
nia, 3) Pacific sardine <200 mm SL from all Cal¬ 
ifornia regions and Pacific sardine 200-209 mm 
SL from Central California, and 4) all Pacific 
sardine collected from Washington and Oregon, 
Pacific sardine 200-209 mm SL from Northern 
California, all Pacific sardine 210-219 mm SL 
and >220 mm SL from California, and northern 
anchovy from Willapa Bay. 
A 2-way PERMANOVA on Pacific sardine 
parasite communities showed that both region 
(pseudo-F=4.75) and fish size category (pseu- 
do-F=5.12) were significant factors (P<0.01) in 
the main effects model. There was not a signif¬ 
icant interaction between region and fish size 
(pseudo-F=0.97, P=0.50). A PERMANOVA for 
pairwise comparison of parasite communities 
showed that there was no significant difference 
between the parasite communities of Pacific sar¬ 
dine <200 mm SL and 200-209 mm SL (pseudo 
£=1.43, P=0.11) (data not shown). Thus, these 
size categories were combined into one category 
(<210 mm SL) in further analyses. In contrast, 
parasite communities from the 2 larger size cat¬ 
egories were significantly different from those of 
the smaller Pacific sardine and from each other 
(all comparisons P<0.05) (data not shown). 
Pairwise regional comparisons with PER¬ 
MANOVA between the parasite communities 
A British Columbia 
V WA-OR 
■ NorCA 
♦ CenCA 
o SoCA 
+ Anchovy-Wilapa Bay 
x Anchovy-Grays Harbor 
* Anchovy-SoCA 
Figure 2 
Nonmetric multidimensional scaling (MDS) plots of distance from 
centroids calculated from square-root transformed data of parasite 
abundance from (A) Pacific sardine (Sardinops sagax) sampled in the 
California Current System from 2005 through 2008, averaged by size 
category and region of collection (lines connect size categories within 
a region), and (B) Pacific sardine caught in 2005-2008 and northern 
anchovy (Engraulis rnordax ) caught off Washington in 2007 (2 sites) 
and Southern California in 2008 (1 site), averaged by collection loca¬ 
tion. Results of a hierarchical agglomerative clustering with similarity 
values of 60% and 70% are overlaid onto the plot. Symbol shapes rep¬ 
resent regions where Pacific sardine were sampled. Sample collection 
occurred in 5 regions: Vancouver Island, British Columbia, Canada, 
Washington and Oregon (WA-OR); Northern California (NorCA), Cen¬ 
tral California (CenCA), and Southern California (SoCA). The labels 
<200, 200, 210, and 220 indicate the size category: <200, 200-209, 
210-219, and >220 mm in standard length. The 2D stress value is a 
measure of the representation of interrelationships in a 2-dimensional 
ordination space, with stress <0.1 corresponding to a good interpreta¬ 
tion and stress <0.2 providing a useful 2-dimensional picture. 
