Ryer et al. : Depth distribution, habitat associations, and differential growth of Chionoecetes bairdi 
259 
ing each month, all tows at the 4 sites were conducted 
within 6 days of one another. From each scrape tow, the 
number of crabs per square meter was calculated on 
the basis of the area swept. For the purpose of analy- 
ses, scrape tows were grouped into 5-m depth bins cen- 
tered on 2.5, 7.5, 12.5, 17.5, and 22.5 m. 
During both 2010 and 2011, we also conducted a 
smaller number of tows with a 2-m plumb-staff beam 
trawl (hereafter referred to as a “trawl”) with a 3-mm- 
mesh codend. This gear allowed us to sample crabs at 
greater depths than with the scrape, which has an op- 
erational depth limit of -30 m. Tows were conducted 
at Pillar at depths of 3-80 m, parallel to shore, during 
July and August. We towed the trawl 200 m, and con- 
firmed that length with GPS readings at the beginning 
and end of each tow. In 2010, 35 tows were completed 
(July: 18, August: 17), and 46 tows were made in 2011 
(July: 23, August: 23). Crabs were sorted from the catch 
of these tows and enumerated in the laboratory. The 
number of crabs per square meter also was calculat- 
ed on the basis of the area swept, and the resulting 
density values were grouped into 10-m depth bins and 
analyzed. 
Using video from scrape tows, we characterized the 
relative abundance of worm tubes along each tow. Dur- 
ing video analysis, each tow was divided into 5-s seg- 
ments. For each segment, an observer assigned an in- 
dex score for the relative abundance of worm tubes on 
the seafloor passing between the scrape’s runners. The 
index was scored on a 5-point scale (0-4), with 0 rep- 
resenting worm absence and 4 representing a contigu- 
ous “worm turf” (Stoner et ah, 2007; Ryer et al., 2013). 
The vast majority of worm tubes were easily identifi- 
able as belonging to S. sibirica, following prior posi- 
tive identification from benthic samples taken during 
2008 (Jewett3).The tubes of S. sibirica are distinctive: 
1.00-1.25 mm in diameter and up to 12 cm long, with 
approximately 70% of the tube emergent and upright 
above the sediment surface. For each tow, the average 
worm index score was calculated. 
Crab density data followed a Poisson distribution. To 
analyze this data, we used generalized linear models 
(GLMs), incorporating a Poisson log-linear link func- 
tion, and we used the Wald chi-square statistic for tests 
of significance (Wald, 1943). Crab densities were first 
converted to an expanded integer scale through mul- 
tiplication by 10, augmentation by 1, and rounded to 
the nearest whole integer. We tested the resultant data 
for independent and interactive effects of site, month, 
and depth, with the worm index as a covariate. We 
conducted separate analyses for each year. For 2010, 
we excluded data for Womens and Kalsin from the 
analysis because of the lack of consistent temporal and 
depth coverage at these sites. For 2011, all 4 sites were 
included in analysis. Trawl data also followed a Pois- 
sion distribution and were analyzed with GLM with a 
3 Jewett, S. 2008. Personal commun. Institute of Marine 
Science, Univ. Alaska Fairbanks, Fairbanks, AK 99775-7220. 
Poisson log-linear link, which allowed us to test for the 
effects of depth on crab distribution. 
Diver estimation of crab density 
During June, July, and August of 2010, divers used 
quadrats to estimate densities of crabs and worm 
tubes. During each dive, multiple squares made of 
PVC pipe were haphazardly placed on the bottom to 
mark the 0.25-m 2 quadrats that would be surveyed. A 
second square was positioned haphazardly within each 
of the 0.25-m 2 quadrats, marking an area of 0.10 m 2 
for enumeration of worm tubes. After worm tubes were 
counted, the 0.10-m 2 square was removed. Then, using 
a metal tooth comb, a diver methodically worked over 
the sediment within the 0.25-m 2 quadrat, dislodging 
crabs and enumerating them. During the course of 10 
dives, 222 paired (crab and worm tube) quadrats were 
sampled, with 69, 74, 59, and 20 pairs from depths of 
8, 13, 17, and 23 m, respectively. Data followed a Pois- 
son distribution and were analyzed with GLM with a 
Poisson log-linear link. We tested for independent and 
interactive effects of month and depth on crab density, 
with worm density included as a covariate. 
Measurements of crab carapace width 
We measured the carapace width of crabs from scrape 
tows conducted during only July and August of 2010 
and during all 4 months in 2011. For tows from which 
more than 100 crabs were sampled, we measured a 
subsample of 50-100 crabs. Carapace width was mea- 
sured to the nearest 0.1 mm with digital calipers. On 
the basis of visual examination of a cumulative size- 
frequency distribution, break points between peaks 
(instars) were identified and used to classify individu- 
als into molt stages (C1-C5 instars). Using this clas- 
sification scheme and G-tests (Sokal and Rohlf, 1969), 
we tested for differences in the frequency distribution 
of molt stages between sites for each month. In addi- 
tion, we compared mean carapace width using ANOVA 
(Sokal and Rohlf, 1969). Where the ANOVA indicated 
a significant effect, group means were compared with 
Tukey’s honestly significant difference (HSD) test (So- 
kal and Rohlf, 1969). Where data did not meet paramet- 
ric assumptions, a Kruskal-Wallis ANOVA was used, 
and when significant effects were found, the associated 
multiple range test for differences between means was 
used (Conover, 1971). In these comparisons, we elimi- 
nated from consideration those crabs that were not, in 
our estimation, representative of the age-0 cohort (i.e., 
crabs of age 1 or older). 
Results 
Distribution of age-0 crabs 
Crab densities estimated from catches of scrape tows 
conducted during 2010 increased from May through 
