260 
Fishery Bulletin 113(3) 
Figure 2 
Mean densities of age-0 southern Tanner crabs ( Chion - 
oecetes bairdi), plotted against depth, from scrape tows 
conducted during the months of May-August in 2010 
around Kodiak Island, Alaska, at (A) Holiday Beach 
and (B) Pillar Creek Cove and (C) from beam trawl 
hauls conducted at Pillar Creek Cove in July and Au- 
gust 2010. Error bars indicate standard error of the 
mean. Note that the depth range for beam trawl hauls 
on the x-axis in panel C is different from the range in 
the panels that show depths of scrape tows. 
Depth (m) 
Worm index (0-4) 
Figure 3 
(A) Mean worm index scores (averaged over months), 
which provide a measure of relative abundance of 
worm tubes on a 5-point scale (0-4), plotted against 
depth for the sites at Holiday Beach and Pillar Creek 
Cove, Kodiak Island, Alaska, for surveys conducted in 
2010. Error bars indicate standard error of the mean. 
(B) Residuals from generalized linear model analysis 
of crab density (without the worm index included as a 
covariate), plotted against worm index scores and indi- 
cating a positive relationship between crab and worm 
tube abundance after the confounding depth effect had 
been removed. 
July, then decreased from July to August (Fig. 2, A and 
B). The density of age-0 Tanner crabs was strongly in- 
fluenced by depth; crabs were generally absent from 
depths <8 m. At depths greater than 8 m, crab density 
rose with increasing depth. There was a significant in- 
teractive effect of depth and site; crabs were distrib- 
uted more deeply at Pillar Creek Cove than at Holiday 
Beach (GLM depthxsite: Wald % 2 =8.01, df=3, P=0.045). 
The depth effect also varied among months, with a more 
uniform distribution across depths in August as overall 
density decreased (depthxmonth: Wald % 2 =55.1, df=12, 
P<0.001). Crab density was generally higher at Pillar 
than at Holiday, although the magnitude of this dif- 
