280 
Fishery Bulletin 114(3) 
Table 2 
Percentages of the prey composition for 3 myctophids collected at 32 stations in the eastern tropical Pacific 
Ocean. Samples of zooplankton prey were taken with an oblique haul of a bongo net. Offshore, intermedi- 
ate, and nearshore regions were identified by a classification tree analysis (Fig. 7). Standard deviations 
are shown in parentheses. 
Latitude 
Longitude 
Amphipods 
(%) 
Copepods 
(%) 
Euphausiids Euphausiids 
(%) (%) 
Other 
prey 
(%) 
Nonprey 
items 
(%) 
6.30°N 
140.72°W 
0.55 
72.9 
2.37 
12.1 
3.29 
8.78 
0.23°N 
119.92°W 
1.52 
76.6 
3.07 
3.09 
2.79 
13.0 
11.03°N 
119.28°W 
1.43 
82.8 
2.92 
1.80 
4.06 
7.00 
5.02°N 
113.58°W 
1.62 
61.0 
2.58 
22.3 
2.38 
10.1 
8.25°N 
113.17°W 
1.38 
45.0 
5.98 
33.6 
2.66 
11.3 
3.37°N 
110.85°W 
2.17 
67.5 
2.35 
9.74 
4.00 
14.3 
6.07°N 
110.70°W 
2.05 
45.2 
4.61 
37.0 
5.17 
6.04 
1.62°S 
110.65°W 
1.07 
69.8 
4.94 
5.16 
1.38 
17.6 
5.65°N 
108.00°W 
1.45 
61.1 
5.54 
19.7 
6.29 
5.84 
1.90°S 
107.58°W 
1.73 
76.7 
4.36 
2.78 
0.36 
14.1 
6.48°N 
104.38°W 
1.02 
49.3 
3.12 
32.9 
2.00 
11.6 
6.33'’N 
101.73°W 
1.15 
70.5 
3.89 
15.4 
1.78 
7.30 
7.25''N 
101.37°W 
1.35 
65.0 
5.13 
17.6 
2.19 
8.68 
2.78'’S 
96.37°W 
1.24 
78.4 
3.87 
3.11 
1.67 
11.7 
7.07'’N 
95.32°W 
0.82 
82.5 
2.35 
1.53 
3.55 
9.20 
2.68°N 
93.65°W 
0.82 
81.8 
2.50 
5.18 
4.36 
5.29 
10.8°S 
93.62°W 
0.76 
61.7 
20.7 
1.48 
1.58 
13.8 
9.02°N 
93.05°W 
2.56 
82.1 
3.63 
2.00 
7.57 
2.14 
6.15°N 
92.20°W 
1.91 
79.9 
4.37 
3.21 
3.01 
7.64 
8.37°N 
91.68°W 
1.76 
79.8 
4.01 
1.66 
2.60 
10.2 
6.23°N 
90.90°W 
2.15 
76.2 
3.07 
3.73 
2.86 
12.0 
5.30°N 
89.20°W 
1.57 
75.8 
4.20 
3.06 
1.14 
14.3 
7.35°N 
86.93°W 
1.15 
72.7 
5.51 
5.49 
1.09 
14.1 
1.73°S 
85.45°W 
1.72 
80.7 
3.29 
1.21 
4.98 
8.04 
7.47°N 
85.00°W 
1.15 
81.2 
3.74 
1.21 
1.77 
10.9 
4.77°N 
84.15°W 
1.22 
77.4 
4.41 
3.54 
0.83 
12.6 
1.17°S 
82.45°W 
0.48 
87.3 
1.78 
1.45 
3.22 
5.73 
3.72°N 
81.82°W 
1.08 
77.0 
3.73 
4.58 
1.72 
11.9 
4.57°N 
81.52°W 
1.07 
76.7 
4.07 
2.76 
5.76 
9.64 
4.73°N 
80.88°W 
1.34 
69.6 
5.29 
6.40 
2.60 
14.8 
6.83°N 
80.83°W 
1.23 
75.5 
4.21 
5.23 
1.14 
12.7 
5.18°N 
80.43°W 
0.61 
70.5 
5.44 
6.62 
3.83 
12.9 
Region 
Offshore 
1.40 
63.9 
3.90 
17.5 
3.20 
10.1 
Intermediate 
1.50 
80.5 
3.30 
2.60 
3.50 
8.60 
Nearshore 
1.11 
75.1 
4.55 
4.48 
2.34 
12.5 
Mean 
1.35 
72.2 
4.41 
8.65 
2.93 
10.5 
(±0.49) 
(±10.7) 
(±3.16) 
(±10.1) 
(±1.69) 
(±3.46) 
Classification tree analysis 
The classification tree analysis produced a tree with 
2 splits and 3 terminal nodes (Fig. 6A) and yielded 
a cross-validated error rate of 0.73 (standard er- 
ror [SE]=0.04, coefficient of multiple determination 
(i?2]=~27%). The rankings of variable importance (Fig. 
6B) indicated that longitude was the most important 
variable (i.e., rank=1.00) for predicting the diet com- 
position of these myctophids. The ostracod numeric 
composition of the zooplankton (rank=0.74), the cope- 
pod composition of the zooplankton (rank=0.61), MLD 
(rank=0.61), and SSS (rank=0.60) were the next most 
important. Myctophid species (rank=0.08) was a less 
important predictor variable in the classification tree 
given our collection of surface-migrating fishes and at 
the taxonomic level possible in this study. Latitude, 
myctophid length, SST, SCHL, zooplankton volume, 
and euphausiid composition in the zooplankton net 
samples yielded an importance rank of zero. 
The initial split in the tree provided the greatest 
reduction in deviance over the entire data set and par- 
