280 
Fishery Bulletin 113(3) 
o 
0) 
3= 
0 ) 
< 
CD 
2 
1 
0 
-1 
-2 
-3 
-4 
■ jJU MUB HUB 1 1 III M— i Bill I 1 MB | 
170 E 175 E 180 175W170W165W 
Longitude 
Slope 
o 
0) 
3= 
d) 
< 
CD 
Current (cm/s) 
Figure 4 
From the best-fitting generalized additive model (GAM), predictions of presence of adult Pacific ocean 
perch ( Sebastes alutus) at trawl stations in the Aleutian Islands during 1997—2010 in relation to (A) lon- 
gitude, (B) kriged bottom slope, (C) predicted tidal current velocity, and (D) the interaction between depth 
and temperature. A change in number orientation indicates that either a maximum or a minimum was 
reached for the GAM effect. 
and adult conditional abundance GAMs (Figs. 7 and 
8). Similar D,T-dependent optima were observed for ju- 
venile and adult Pacific ocean perch abundance where 
present, and predicted conditional abundance was 
highest over depths of 200-250 m and a temperature 
range of 4.5-5.0°C. As with the predictions of increased 
probability of occurrence from the presence-absence 
GAMs, some of the highest predicted conditional abun- 
dances for either life stage are associated with major 
Aleutian passes. Higher abundances were predicted 
around Buldir Pass (176°E) and Seguam Pass (173°W), 
and some decreased abundances were predicted around 
the Islands of Four Mountains (170°W). Predicted adult 
conditional abundance increased with increasing SI, up 
to an incline of around 5°, but SI was not a predictor 
retained in the juvenile conditional abundance GAM. 
Fewer biogenic structures were retained in the best- 
fitting conditional abundance GAMs, than in the pres- 
ence-absence GAMs, but the relative contribution of 
these predictors to the deviance explained in the model 
was similar within each life stage (Table 3). The com- 
bined deviance explained by the presence or absence of 
biogenic structures remained a more important propor- 
tional contributor to the total deviance explained in the 
juvenile conditional abundance GAM than in the adult 
GAM (13.1% versus 2.1%). When sponges of the C and 
V morphogroups were present, the predicted juvenile 
conditional abundance was higher, but the presence of 
sponges of the ovate [O] morphogroup was associated 
with a decrease in the abundance of juveniles where 
present. For adults, conditional abundance increased 
in the presence of Co and decreased in the presence of 
sponges of the G morphogroup. 
Validation of conditional abundance generalized additive 
modeling By cross-validating our GAM predictions it- 
