284 
Fishery Bulletin 1 14(3) 
East 
< 0.11 
More 
saline 
1) Longitude: 100°W 
Competitor; ZcomO 
2)SSS: 32.86 
West 
> 0.12 
m 
Ostracod — Cypridina 
Less americana 
saline 
m El 
Copepod— Euphausiid 
other 
B 
Figure 6 
(A) The classification tree (at a 1 SE resolution) used to predict myctophid (Myctophum niti- 
dulum, Symbolophorus reversus, Gonichthys tenuiculus) diet composition from geographic, 
oceanographic, and biologic predictor variables (see Materials and methods section for further 
explanation). The prey group identified at each terminal node is that with the highest propor- 
tional numeric composition among all the prey in the myctophid samples that were mapped to 
each terminal node. (B) A variable importance plot showing rankings of each nonzero predic- 
tor variable used in the classification tree model. Lon^longitude; ZcomO=ostracod composition 
in the zooplankton net samples; ZcomC=copepod composition in the zooplankton net samples; 
mld=mixed layer depth; SSS=sea surface salinity; Spp=myctophid species 
made up more than 80% of the community in the in- 
termediate region, perhaps reflecting a competitive 
advantage that various copepods have in moderately 
oceanographic conditions (McGowan and Walker, 1985; 
Turner, 2004). In contrast, ostracods were limited in 
their range to oligotrophic conditions, and euphausiids 
were more dominant in productive nearshore environ- 
ments (Brinton, 1979). 
Previous research has indicated that M. nitidulum 
selects amphipods and ostracods and that S. reversus 
prey on euphausiids and amphipods (Van Noord et al. 
2013b), and in fact dietary resource partitioning among 
myctophids has commonly been reported (e.g., Hopkins 
and Gartner 1992: Hopkins and Sutton 1998: Cherel 
et al., 2010), but the influence of oceanography on diet 
is less often considered. The selective feeding behavior 
observed by Van Noord et al, (2013b) is indicative of 
resource partitioning, but the current study expands 
on these initial findings and presents a more complete 
ecosystem-based analysis by including spatial, biologi- 
cal, and oceanographic variables in addition to dietary 
information. The current study indicates a very low 
level of resource partitioning among these species, as 
evidenced by the low importance of myctophid species 
in the ranking of variables. Indeed, when considering 
a fuller compliment of oceanographic, spatial, and prey 
composition data, we found that resource partition- 
ing between species is not the most important aspect 
controlling diet. Therefore, dietary resource partition- 
ing and competition among these species played minor 
roles in regulating feeding behavior, and spatial and 
oceanographic predictor variables outweighed resource 
partitioning. The importance of considering spatial, 
biological, and oceanographic variables when evaluat- 
ing feeding behavior is clear, and the findings obtained 
from these variables have implications for interpreting 
previous results. 
A diverse fish community structured through di- 
etary resource partitioning can be affected by distur- 
bance events and bottom-up forcing. For example, flying 
fish in the ETP consume many of the same prey that 
are consumed by myctophids, which could introduce 
a level of food competition (Van Noord et al., 2013a). 
During the course of our investigations, (August-No- 
vember 2007), a tropical storm bisected the sampling 
area (15-17 October), resulting in enhanced upwelling, 
productivity, and zooplankton biomass in the wake of 
the storm. The flying fish community reflected these 
changes. Feeding success increased and diet composi- 
tion changed in accordance with storm-induced chang- 
