Laman et al.: Correlating environmental and biogenic factors with abundance and distribution of Sebastes alutus in Alaska 
277 
Table 2 
Prevalence of biogenic structures, measured as percent frequency of occur- 
rence in trawl tows per annum, in catches of bottom trawl surveys conducted 
in the Aleutian Islands between 1997 and 2010. 
Substratum 
Survey year 
1997 
2000 
2002 
2004 
2006 
2010 
Porifera unidentified 
63 
78 
74 
78 
78 
77 
Corals 
48 
51 
52 
49 
50 
55 
Globular-papillate 
43 
52 
47 
46 
43 
55 
Globular 
43 
39 
44 
39 
37 
36 
Flabellate 
35 
37 
39 
34 
34 
47 
Arborescent 
27 
41 
34 
25 
32 
35 
Massive 
29 
30 
25 
28 
33 
38 
Papillate 
21 
30 
24 
24 
37 
28 
Bryozoans 
23 
32 
14 
15 
13 
24 
Vase 
12 
14 
16 
18 
14 
20 
Ovate 
10 
4 
10 
13 
6 
14 
Repent 
0 
0 
0 
4 
7 
10 
Tubular 
0 
0 
2 
4 
3 
4 
Encrusting 
0 
2 
2 
4 
0 
2 
Clavate 
1 
2 
1 
0 
2 
0 
Flagelliform 
0 
0 
1 
1 
1 
0 
Stipitate 
0 
0 
0 
0 
2 
1 
Pedunculate 
<1 
0 
<1 
0 
0 
0 
from P<0.05. The best-fitting GAMs 
among the 4 starting models that 
predicted juvenile or adult presence 
or abundance were selected on the 
basis of the lowest Akaike’s infor- 
mation criterion scores. 
Predicted presence of juvenile 
and adult Pacific ocean perch, based 
on the GAMs formulated with data 
from the period 1997-2010, was 
validated by comparison with ob- 
served presence of this species from 
the 2012 Aleutian Islands bottom 
trawl survey. Comparisons were 
made with Cohen’s kappa coeffi- 
cient ( k ; Manel et al., 2001), which 
measures the proportion of correct- 
ly predicted cases of presence or 
absence after accounting for chance 
effects. This coefficient ranges from 
0 to 1, corresponding with poor to 
near perfect agreement (Landis and 
Koch, 1977). 
For conditional CPUE models, 
we used a form of jack-knifing, it- 
eratively leaving out a single year 
of data, to internally validate the 
GAM. Residual deviance from each 
iteration of this cross-validation 
was used to compute a pseudocoefficient of determi- 
nation (r 2 ; O’Brien and Rago, 1996) that was used to 
measure model fit. The residual deviance (RD) is the 
deviance remaining in the data that is unexplained by 
the full model (i.e. , RD=l-r 2 ). By iteratively dropping 
terms retained in the best-fitting GAM and recalculat- 
ing r 2 , we were able to assess the relative contribution 
of that term to the deviance explained by the model, 
thereby estimating its leverage in the GAM. 
Results 
Summary of biological collections 
Between 1997 and 2010, 6 summer bottom trawl sur- 
veys were conducted in the Aleutian Islands by AFSC, 
resulting in a total of 2364 stations included in this 
study. The Pacific ocean perch was consistently ranked 
amongst the most abundant fish species collected on 
these surveys and commonly occurred in the survey 
catches; they were caught in 58% of the standard sur- 
vey tows. Of the trawl catches, 23% contained juvenile 
Pacific ocean perch and 53% contained adults. Juve- 
niles co-occurred with adults in about one-third of the 
trawl catches (34%), and the majority of trawl catches 
with juveniles also contained adults (79%). Sponges 
and corals were common in hauls of the Aleutian Is- 
lands bottom trawl surveys as well; they occurred in 
about 87% of the trawl catches. 
Sponges were the most commonly occurring struc- 
ture-forming invertebrate collected in our surveys be- 
tween 1997 and 2010, followed by corals and bryozoans. 
The majority of the trawl tows selected for analyses 
contained sponges and corals, which were also present 
in most of the catches containing Pacific ocean perch 
(about 91%). The composite corals group occurred in 
just over half (51%) of the tows in all survey years. 
Of catches containing juvenile Pacific ocean perch, 96% 
also collected sponges or corals, but the rate of co-oc- 
currence for adults was lower (90%). Bryozoans were 
less common in our catches than were sponges or cor- 
als, occurring, on average, in about 20% of the trawl 
hauls. They were also less common in hauls that con- 
tained Pacific ocean perch juveniles or adults, occurring 
in about 20% of these hauls. 
“Porifera unidentified” was the most common sponge 
category occurring in our survey tows each year (Table 
2). Globular (G) and globular-papillate (Gp) sponges oc- 
cur at similar or slightly elevated rates compared with 
some of the larger morphotypes (e.g., arborescent [A], 
flabellate [F], and massive [M] ). Clavate (C), encrusting 
(E), and tubular (Tu) sponges were less common, and 
pedunculate (Pe) sponges were the rarest morphotype 
collected. 
Summary of environmental and physical data 
Bottom trawl hauls included in this study were con- 
ducted over a wide variety of physical and oceanograph- 
ic conditions. The deepest trawl hauls during the study 
period were conducted at 488 m and the shallowest at 
