Van Noord et al.: Oceanographic influences on the diet of myctophids in the eastern Pacific Ocean 
277 
140°W lao^w 120°W 110“W 100°W 90°W 80°W 
Species distribution of the 580 myctophid individuals of 3 myctophid species collected 
from the eastern tropical Pacific Ocean by dip net during 2006. 
fish, weighed, fixed in 3.7% formalin, and stored in 70% 
ethyl alcohol. The amount of time specimens were left 
either unfrozen or unpreserved was minimized to avoid 
degradation of stomach contents. 
Displacement volumes of small zooplankton from 
oblique net tows were measured in the laboratory. This 
measurement excluded all fishes, large cephalopods, 
pelagic crabs, and large plankters (>5 mm), including 
Thaliacea and medusae (Ohman and Smith, 1995). Net 
samples were split to a one-eighth volume by using a 
Folsom plankton splitter, and individuals were identi- 
fied under a dissecting microscope to the taxonomic 
level of order. Identifications were made to order level 
for comparison with the taxonomic resolution of most 
gut-content identifications. This taxonomic resolution 
restricted our diet analysis because we could not ex- 
clude the possibility that resource partitioning occurs 
at a lower taxonomic level. Zooplankton densities were 
standardized as numbers of individuals per m^ of wa- 
ter filtered at each station by using methods of Smith 
and Richardson (1977) and were converted to numeric 
percentages for comparison with gut contents. 
We calculated mean percentage by number (MW,) by 
using the following equation: 
N:: 
N- 
xlOO, 
( 1 ) 
where Wy = the count of prey type i in fish j; 
Q = the number of prey types in the stomach of 
fish j; and 
P = the number of fish with food in their stom- 
achs in any particular sampling stratum 
(Chipps and Garvey, 2007). 
We calculated mean percentage by weight {MWi) simi- 
larly, substituting prey weights (W) for counts (W). Per- 
cent occurrence (O,) was calculated as the number of 
fish containing a specific prey item i, divided by the 
total number of fish sampled, including those mycto- 
phids with empty stomachs, and multiplied by 100. We 
focused our analysis on the numeric predation data for 
comparison with the numeric zooplankton prey data. 
Ciassification tree analysis 
Diet composition 
We identified stomach contents to the lowest possible 
taxon and enumerated and weighed contents by taxo- 
nomic group. Pieces of plastic found in stomachs were 
not included in the analysis of the natural diet. Stom- 
achs void of all material, including unidentifiable sub- 
stance, were classified as empty. 
We applied Classification and Regression Tree (CART) 
analysis to the myctophid diet data, using the modi- 
fied approach of Kuhnert et al. (2012 [see also Olson 
et al., 2014]). CART is a nonparametric modeling ap- 
proach described by Breiman et al. (1984). Diet data 
are partitioned by forming successive splits on predic- 
tor variables in order to minimize an error criterion, 
in this case the Gini index, which represents a mea- 
