Harding et at.: Regional and seasonal patterns of epipelagic fish assemblages from the central California Current 
267 
pairwise comparisons. In MDS plots of community 
structure, stations in summer cruises separated fairly 
well by region (Fig. 3A), with NC stations forming a 
central cloud and GF stations scattered more widely 
around the perimeter and mostly to one side of the plot. 
Within the GF region there was further recognizable 
spatial structure: stations 23, 24, and 25 are the three 
easternmost nearshore stations, closest to the mouth 
of the San Francisco Bay in real space. Stations FS 
(Fanny Shoal) and NFI (North Farallon Island), the 
two westernmost GF stations, were dissimilar in the 
summer ordination. Stress was moderate at 0.13, indi- 
cating a fairly reliable plot. Samples from fall cruises 
formed a similar regional pattern, with NC stations 
falling in a (mostly) separate central cloud surrounded 
by more widely scattered and dissimilar GF stations 
(Fig. 3B). Further spatial structure within the GF was 
again apparent; the eastern nearshore trio of stations 
(23, 24, and 25) were placed together and opposite most 
of the remaining GF stations. The two westernmost GF 
stations (FS and NFI) were structurally similar to each 
other in fall. Stress was moderate at 0.19. 
Seasonal community pattern Ordinations of seasonal 
community patterns also supported their correspond- 
ing statistical tests. Along the NC there was clear 
separation of summer and fall samples (Fig. 3C), 
consistent with the significant seasonal test for this 
region, whereas in the GF there was broad overlap of 
summer and fall samples (Fig. 3D) consistent with 
the nonsignificant test for seasonal differences in 
this area. Summer samples in both regions were more 
widely spread across the plots, and hence more vari- 
able, than fall samples, and stress was moderate for 
both plots. 
Representative species The SIMPER routine identi- 
fied a small subset of species most responsible for the 
observed differences between the NC and GF communi- 
ties, and between summer and fall communities along 
the NC (Table 3). Over 80% of the total dissimilarity 
between communities was attributed to about 12 species, 
and six of these were high ranking contributors to all 
three of the paired SIMPER comparisons between sig- 
nificantly different communities. Variation in the rela- 
tive abundance and frequency of occurrence of these six 
species, in particular, allowed the multivariate tests and 
ordinations to discriminate among groups. For example, 
the three common clupeiform species (Pacific herring, 
Pacific sardine, and northern anchovy) were much more 
abundant in the GF than along the NC, regardless of 
season. Among the other principal species, jacksmelt 
and juvenile Chinook salmon were more abundant in 
the NC-summer community, whereas medusafish were 
more abundant in the GF-summer community. In the 
fall, juvenile Chinook salmon were still more abundant 
in the NC community, and jacksmelt and medusafish 
were more abundant in the GF. 
Significant seasonal differences were only observed 
along the NC. Here, the NC-summer community was 
A 
2D Stress: 0.14 
04 
V 
°°© 
05 
A 
04 
o 
05 04 
AA 
02 
01 
‘01 
A 
00 
A 
A 03 
A A 01 
•° 3 A ™ 
A NC summer A NC fall 
Q GF summer ^ GF fall 
02 
A 
CM 
O 
CL 
B 
-4 -2 0 2 4 6 
PCI (57% variance) 
Figure 2 
(A) Nonmetric multidimensional scaling (MDS) plot 
of aggregated trawl samples averaged and coded by 
region, season, and year. Data represent square-root 
transformed fish densities; resemblance was based on 
ranked Bray-Curtis similarity. (B) Principal components 
analysis (PCA) plot of seven aggregated environmen- 
tal variables: chlorophyll-a concentration, water den- 
sity, water depth, distance offshore, photosynthetically 
available radiation, salinity, and water temperature: 
log(CHL), DEN, log(DEP), sqrt(DIS), sqrt(PAR), SAL, 
and TMP, respectively, measured immediately before 
or after trawling, and averaged and coded by region, 
season, and year. Individual variable transformations 
were applied to improve normality; resemblance was 
based on Euclidian distance. 
dominated by jacksmelt, Pacific herring, and juvenile 
Chinook salmon, and the NC-fall community was 
dominated by northern anchovy, Pacific sardine, and 
medusafish. Other important contributors to regional 
dissimilarity were adult Chinook salmon, jack mack- 
erel, and Pacific saury ( Cololabis saira) (all more 
strongly associated with the NC), and Pacific tomcod 
(Microgadus proximus), surf smelt, and Pacific but- 
terfish (all more strongly associated with the GF). 
Along the NC, adult Chinook salmon and jack mack- 
erel were more abundant in summer than in fall, 
whereas Pacific tomcod were more abundant in fall 
than in summer. 
