270 
Fishery Bulletin 116(3-4) 
S. auriculatus 
S. emphaeus 
S. goodei 
S. rastrelliger 
S. wilsoni 
Both forms of ID 
'S. aleutianus' 
S. aurora 
S. babcocki 
S. crameri 
S. diploproa 
S. elongatus 
S. entomelas 
S. helvomaculatus 
S. jordani 
S. pinniger 
S. proriger 
S. ruberrimus 
S. alutus 
5. brevispinis 
S. ensifer 
5. fla vidus 
S. maliger 
S. melanostictus 
S. melanostomus 
S. miniatus 
S. nebuiosus 
S. paucispinis 
S. reedi 
S. rosaceus 
Molecular ID 
Morphological ID 
Figure 2 
Venn diagram of species of Sebastes identified morphologically, by 
molecular technique, or both, from samples collected during 2005- 
2008 off Oregon and Washington. 
the Sprensen distance measure, a maximum of 4 axes, 
and 200 iterations. 
We examined monthly averages of temperature, sa¬ 
linity, and large-scale environmental indices by year 
to assess how the physical environment varied dur¬ 
ing the study. To determine the degree to which these 
variables were correlated with one another, we used 
Pearson correlation analysis. We also used correlation 
analysis to assess the relationship of the environment 
with the juvenile rockfish community by examining 
relationships between NMS scores and environmental 
covariates measured at the level of an individual haul. 
Physical variables included surface (1 m) and depth 
(30 m) temperature and salinity, as well as the large- 
scale indices of upwelling, PDO and NPGO. Distance 
from shore (km) was included as a spatial variable in 
the multivariate analysis because it has been shown 
in other studies that ichthyoplankton communities can 
vary along inshore and offshore gradients (Auth et ah, 
2011). Temporal variables included month, year, and 
season (spring=May and June, summer=July-October). 
We also conducted indicator species analysis on each 
rockfish species matrix to detect within- and between- 
year variability and to identify statistically significant 
indicators for each month and year (Dufrene and Leg¬ 
endre, 1997). An indicator species analysis combines in¬ 
formation on relative abundance and frequency of occur¬ 
rence to identify indicators that are most characteristic 
of a group. Indicator species analysis was performed 
separately for sampling months and years. Levels of sig¬ 
nificance were determined by using a Bonferroni adjust¬ 
ment and based on Monte Carlo tests with 5000 permu¬ 
tations for comparison with observed indicator values. 
Results 
Rockfish identification 
Late-larval and juvenile rockfish were 
caught in 196 of the 271 tows made (72%) 
for a total of 78,407 specimens from 2005 
through 2008. Using strictly visual means 
(meristics and pigmentation), we were able 
to identify 3266 (4.1% of total) late-larval/ 
juvenile rockfish from 17 distinct taxa found 
in 93 hauls (Fig. 2). By far the most abun¬ 
dant species (81% of the identified rockfish) 
was the rosethorn rockfish (S. helvomacu¬ 
latus), of which 2633 were identified. Other 
abundant rockfish species identified visually 
were the greenstriped rockfish (S. elongatus, 
213 individuals), rougheye rockfish ( S. aleu¬ 
tianus, 72 individuals), redbanded rockfish 
(S. babcocki, 66 individuals), brown rockfish 
(S. auriculatus, 66 individuals), and dark- 
blotched rockfish ( S. crameri, 52 individu¬ 
als). Collectively, these 6 species accounted 
for 95% of the fish visually identified from 
2005 through 2008. 
Size distributions of late-larval and ju¬ 
venile rockfish measured during this study 
ranged from 6 to 69 mm SL (Suppl. Figs. 1 and 2) (on¬ 
line only). There was a tendency toward greater visual 
identification of larger individuals, especially in May, 
when all specimens >35 mm SL were identifiable but 
none below this threshold (Suppl. Fig. 1) (online only). 
Similar patterns were observed in the annual size 
ranges that could be identified by visual means (Suppl. 
Fig. 2) (online only). 
Subsampling among the visually unidentifiable 
larvae and juveniles, we successfully sequenced full- 
length (782 bp) cytochrome b fragments from a total of 
2534 juvenile rockfish collected in 144 hauls, including 
105 from the NH-100 station. We identified 20 species 
according to our high-confidence stringency criterion 
(Suppl. Fig. 3) (online only), and confidently assigned a 
large number (1489 individuals, 58.7%) of specimens 
to the multispecies WEVZ complex. Another 213 un¬ 
knowns could not be assigned at high confidence and 
were coded as unassigned. Using the low-confidence 
(>50% bootstrap) criterion, we were able to identify 
24 species (Fig. 2, Suppl. Fig 3 (online only) and assign 
many of the previous unknowns to the WEVZ group. 
The remaining 25 unknowns could not be assigned at 
low confidence and were coded as unassigned. 
Of the species that occurred in the set of low- 
confidence identifications that were not found in the 
high-confidence set, 3 were represented by a single in¬ 
dividual (aurora rockfish, S. aurora ; rosy rockfish, S. 
rosaceus) or a few individuals (darkblotched rockfish, 
3 individuals). Only the splitnose rockfish was present 
in greater numbers (43 individuals). Other differences 
between the sets of high- and low-confidence identifi¬ 
cations were proportions of individuals that were as- 
