8 
Fishery Bulletin 106(1 ) 
A 
E 
Cl 
(D 
Q 
X 
o 
o 
o 
X 
□ 
X 
X 
o 
400 x Aleutian Islands-shelf 
X Central GOA-shelf 
- 350 O Aleutian Islands-break 
A Southeastern Alaska-break 
’ i,MI o Aleutian Islands-slope 
X 
■ 250 
A 
■200 A 
■ 15 * 
6 
■ 100 
• 50 
2000 1500 1000 500 0 -500 -1000 
Position (km) 
o 
Position (km) 
Figure 4 
Plots of mean weighted (by catch per unit of effort) distributions of each 
rockfish species-group (n = 17) along three environmental variables. Mean 
weighted distributions of rockfish species-groups are shown for (A) depth 
versus position, (B) temperature versus position, and (C) temperature versus 
depth. Position is the distance from Hinchinbrook Island, Alaska; positive 
values are west of this central point in the trawl surveys and negative values 
are southeastward. Dashed lines indicate break points in the mean-weighted 
environmental variables. Symbols indicate the assemblage membership (from 
Fig. 3) of each species-group as belonging to the Aleutian Islands-shelf, central 
Gulf of Alaska (GOA shelf), Aleutian Islands-break, southeastern Alaska- 
break, or Aleutian Islands-slope assemblage. 
rockfish, adult rougheye rockfish, 
shortraker rockfish, sharpchin 
rockfish, and dusky rockfish. For 
these species the co-occurrence in 
trawl survey catches had a strong 
linear relationship to the amount of 
overlap in their distributions across 
environmental variables, and over 
50% of the variance in co-occur- 
rence was explained by species 
group overlap (Fig. 5). When all 
species subgroups were considered, 
the relationship was significant, 
but the overlap index explained 
only 32% of the variance in co-oc- 
currence among pairs of species 
subgroups. This demonstrates that 
the co-occurrence of rockfish in the 
trawl survey data is positively cor- 
related to their overlap in environ- 
mental preferences. 
Discussion 
The method used in this study 
provided resolution of the species 
composition of rockfish assemblages 
across large-scale environmental 
gradients that influence fish dis- 
tribution (depth and geographical 
position). The results of these anal- 
yses were similar to findings on fish 
assemblages in other areas. Rock- 
fish (and other species) on the west 
coast of North America distribute 
themselves by depth and latitude 
into distinct assemblages (Wein- 
berg, 1994; Williams and Ralston, 
2002; Tolimieri and Levin, 2006). 
In the Gulf of Alaska, an analysis 
conducted on 72 groundfish species 
(including rockfish species) with 
data from five NMFS trawl surveys 
revealed that the major gradients 
for variation in species diversity 
were depth and alongshore distance, 
whereas temperature and temporal 
gradients had only minor effects 
on species composition (Mueter and 
Norcross, 2002). As in the current 
study, Mueter and Norcross (2002) 
and Williams and Ralston (2002) 
found a peak in groundfish species 
richness near the shelf break (200 
m), in the region of overlap between 
shallower species and those occur- 
ring at deeper depths. 
The method of defining rockfish 
assemblages described here is very 
