Syah et al.; Predicting potential fishing zones for Cololabis saira 
335 
14rE 144°E 147“E 150°E 153°E 14rE 144°E 147°E ISO'E 153°E 14rE 144°E 147°E 150°E 153°E 
Figure 3 
Spatial distribution of fishing locations for Pacific saury (Cololabis saira) in the western 
North Pacific pooled during (A) August, (B) September, (C) October, (D) November, and (E) 
December for the period 2005—2013. 
ern Hokkaido, whereas the number of vessels off the 
Sanriku coast decreased. In October (Fig. 3C), fishing 
vessels were widely distributed in Hokkaido and San- 
riku waters. The distribution of fishing vessels moved 
slightly to the south and approached the shores of 
southeastern Hokkaido and Sanriku (38-41°N). During 
this same month, the offshore fishing locations (148°E 
and 48°N) also increased and extended northeast to 
154°E and 43°N. 
During November (Fig. 3D), fishing vessels moved 
southward. The number of fishing vessels in eastern 
Hokkaido waters decreased, but the number of fishing 
vessels around the Sanriku coast increased (38-41°N) 
and moved northeast to 155°E and 43°N. A small num- 
ber of fishing vessels also appeared off the Joban coast. 
In December (Fig. 3E), fishing vessels appeared mostly 
along the Sanriku coast and were distributed in near- 
shore waters between 38°N and 40°N; however, a small 
number of fishing vessels still remained offshore and 
along the Joban coast. 
The monthly averaged time series of environmental 
data for the period 2005-2013 are shown in Figure 4. 
Mean SST values indicated a decreasing trend of tem- 
perature on the fishing locations from August through 
December (Fig. 4A). Mean chl-a concentrations (Fig. 
4B) increased in September but declined in December. 
The mean chl-a concentration was highest in Septem- 
ber (0.93 mg/m^), when most vessels were concentrated 
off the eastern coast of Hokkaido and near the south- 
ern Kuril Islands. Mean EKE and SSHA values (Fig. 4, 
C-D) increased in trends that corresponded with the 
southward shift of fishing vessels, especially from Sep- 
tember until November. 
Model performance and potential fish habitat 
All monthly maximum entropy models significantly 
fitted better than they were fitted by chance as sup- 
ported by the modest values of the performance metric 
(AUC>0.5; Table 4). This outcome indicates the high 
predictive success of these models (Elith et al., 2006; 
Phillips et al., 2006). The relative contribution of each 
environmental variable to model prediction is shown 
in Table 5. Model results indicate that the 2 most im- 
portant factors in August and October were SST and 
EKE, and in September the most important factors 
were SST and chl-a. In November and December, the 
2 highest contributions to model gain were SST and 
SSHA. 
Figure 5 provides the model-derived preferred rang- 
es for each environmental variable. The plots in this 
figure show the performance and contribution of the 
various environmental data to model fit. High prob- 
abilities of occurrence of Pacific saury were observed 
in varied ranges for each month. In general, occurrence 
of Pacific saury had the highest probabilities in cool 
(14-16“C) waters with chl-a concentrations of 0. 5-2.0 
mg/m^. In addition, there were high probabilities of oc- 
currence of Pacific saury at low to moderate EKE and 
positive SSHA values. 
