FISHERY BULLETIN: VOL. 76, NO. 3 



capture and were not counted, but the fish and any 

 other contents were used. Items part way down the 

 esophagus with appendages flattened against the 

 body or the body folded were assumed to have been 

 eaten before capture and were included. I consi- 

 dered such "esophagus" items unlikely to have 

 been eaten after capture because concurrent 

 analyses of diet (Clarke in prep.) on the same 

 species collected by the same net indicate that 

 there is no difference in species composition be- 

 tween such items and items clearly in the stomach 

 and partially digested. 



Stomach fullness values for a single species and 

 single time period were rarely distributed nor- 

 mally. Usually the values were skewed to the left, 

 but variably so — the mean being sometimes close 

 to the median and sometimes close to the 75th 

 percentile. Consequently, the entire set of 

 stomach fullness values for each species were 

 ranked and tested for between-period differences 

 by the Kruskal-Wallis nonparametric equivalent 

 of analysis of variance (//-test). The test is mainly 

 sensitive to differences in position (Tate and Clel- 

 land 1957), and significance implies differences 

 among the medians for the separate time periods 

 but does not single out which sets of data are 

 different. Each adjacent (in time) pair of data sets 

 was tested for differences in the median with the 

 Mann-Whitney or Rank sum test (Tate and Clel- 

 land 1957); however, because of multiple testing 

 on the same data, the significance levels from this 

 cannot be taken rigorously. 



Neither test used is sensitive to differences in 

 variability, and no separate testing was done. 

 Some idea of differences in frequency distribution 

 can be gleaned from relative position of the mean 

 and median. Other gross differences, e.g., bi- 

 modality vs. unimodality, will be pointed out in 

 the results. Likewise, I did not test for possible 

 correlations between sex or size of the fish and 

 stomach fullness. The data from each period were, 

 however, ranked and compared (by inspection) 

 with sex and rank in length; no obvious correla- 

 tions were found. 



RESULTS 



A total of 15 vertically migrating species (10 

 myctophids, 4 stomiatoids, and 1 melamphaid) 

 and 1 nonmigrating stomiatoid were investigated. 

 These included species for which 20 individuals 

 were collected at most of the nine periods sampled 

 plus a few, less frequently taken species selected to 



498 



give broader coverage with respect to systematic 

 position or vertical distil jution pattern. In addi- 

 tion to graphical presentations (cited specifically 

 below), ancillary data for all species are sum- 

 marized in Table 2. In the subsequent presenta- 

 tion, stomachs were considered "empty" if 

 stomach fullness was <0.1'7r. This included both 

 visually empty stomachs and those with only a 

 trace of digested remains in the pyloric end of the 

 stomach. Types of prey organisms, state of diges- 

 tion, and other aspects not obvious from the 

 figures or Table 2 are considered in individual 

 species accounts below. 



Comments on vertical distribution of prey items 

 are based on preliminary analyses of opening- 

 closing plankton tows taken in the study area and 

 their general agreement with data in the litera- 

 ture for the same or closely related species in other 

 central water mass localities. The plankton 

 tows — 16 taken in September 1973 and 20 in 

 November 1974 — covered the depth ranges of the 

 fishes considered both day and night. Euphausiids 

 from all samples have been counted and identified, 

 and copepods either counted (shallow night sam- 

 ples) or sufficiently examined to at least roughly 

 determine the depth ranges of the important prey 

 species. The apparent depth ranges agree gener- 

 ally with those given by Brinton (1967) and Roe 

 (1972). These two important types of prey can, 

 with a high degree of certainty, be classified as 

 shallow nonmigrators (above 200-300 m both day 

 and night), vertical migrators (above 200-300 m at 

 night and below this depth by day), and deep living 

 (below 300 m day and night). Similar statements 

 cannot be made for ostracods, the other important 

 crustacean group, nor for other taxa of zooplank- 

 ton. 



Myctophidae 



Bentbosema suhorhitale (Figure 1) 



The //-test indicated highly significant 

 (P <0.005) differences in stomach fullness over the 

 diel cycle. The data from SS and Nl were charac- 

 terized by low averages, narrow percentile limits, 

 and high proportions of empty stomachs. Sub- 

 sequently stomach fullness generally increased 

 until SR and decreased throughout the day. The 

 most frequent prey items were copepods of the 

 genera Pleuromamma, Candacia, and Paracan- 

 dacia. Euphausia spp. and occasionally small de- 

 capods contributed significantly to the weight of 



