CLARKE; DIETS OF FOURTEEN SPECIES OF MESOPELAGIC FISHES 



Oncaea spp. (^0.6 mm) frequently and had the 

 lowest fractions of macrozooplankton, had much 

 finer gill rakers and somewhat larger lenses than 

 similar-sized Benthosema suborbitale, which took 

 a wider variety of sizes. Diaphus trachops, in con- 

 trast to the other three, ate mostly large prey. 

 ASV's of most of its prey were also much higher 

 than those of the other species. Its gape was the 

 largest of all species examined, consonant with 

 large prey size, but its relatively finely spaced gill 

 rakers and high raker area indicate it is equip- 

 ped to retain small prey as well. Diaphus trachops 

 was the only species caught only at 170 m where 

 zooplankton densities and particularly microzoo- 

 plankton were much lower. While the large lenses 

 of the other three species seem related to increased 

 ability to detect small prey, D. trachops' seem re- 

 lated to detection of relatively large, less dense 

 prey from greater distances. Lower light levels in 

 its depth range would also favor large lenses. 



Diaphus schmidti andD. fragilis were similar to 

 each other other and intermediate to other myc- 

 tophids in all four features. Diet of D. schmidti was 

 generally similar to that of D. perspicillatus, i.e., 

 very general, but it differed in that high ASV's 

 were not associated with many types of small 

 copepods preferred by D. perspicillatus. This is 

 consonant with D. perspicillatus' much finer gill 

 rakers and larger lenses. Although data are few, 

 the diet ofD. fragilis seems most similar to that of 

 D. schmidti. Diaphus fragilis is uncommon near 

 Hawaii but very abundant in more productive 

 waters near the Equator (Hartmann and Clarke 

 1975). It is also larger than D. schmidti. Ebeling 

 (1962) has suggested that "dwarf" species of 

 melamphaids are adapted to the less productive 

 central water masses. The above indicates that 

 similarly the larger of two otherwise similar myc- 

 tophids is less successful in the central water 

 mass. 



Bregmaceros japonicus was the most distinct 

 morphologically of all species considered. It had no 

 gill rakers and the smallest lenses and gape of all 

 species. Though it ate chaetognaths fairly fre- 

 quently, the ASV's indicated that it prefers large 

 crustaceans. Apparently the small mouth of this 

 species does not inhibit it from ingesting large 

 prey, and in spite of its small lenses, it is able to 

 detect and capture at least a fraction of the trans- 

 lucent chaetognaths encountered. 



Diet of M. danae was quite distinct from that of 

 the others. The most frequent and preferred items 

 included large and small forms and taxa other 



than crustaceans — many of which were rarely 

 eaten by other fishes. Also, certain prey such as 

 Pleuromamma and Oncaea spp., which appeared 

 in diets of almost all other fishes, were absent or 

 nearly so from that of M. danae. Not a great deal 

 can be gleaned from its morphological features; in 

 spite of its small mouth and lens, M. danae is 

 obviously capable of ingesting fairly large items 

 and detecting small or translucent prey, but there 

 is no clear indication of why certain prey types 

 were not eaten. 



Among the myctophids, differences in lens size 

 and gill raker space were most obviously and fre- 

 quently correlated with differences in diet and 

 preference. These indicate that ability to visually 

 detect and to retain prey in the mouth are impor- 

 tant factors affecting frequency and preference. 

 The general lack of correlation of dietary features 

 with differences in gill raker area indicates 

 these fish are probably not simply filtering. Mor- 

 phological relationships within the myctophids, 

 however, do not seem to extend to the sole repre- 

 sentatives of the other two families considered 

 here. Bregmaceros japonicus and M. danae appear 

 basically different; whether their morphological 

 features are in any way related to diet must await 

 data on other species of these families. 



Aside from the correlations of lens size with diet 

 and lack thereof for gill raker area, the prefer- 

 ences observed and absolute values of ASV's also 

 indicate that these fishes feed in a particulate, 

 visually oriented mode (O'Connell 1972) as op- 

 posed to filtering. That the fishes are selective 

 precludes simple filtration unless it is assumed 

 that the differences between diet and available 

 prey are due entirely to differential escape 

 capabilities of the prey, and the general absence of 

 small or translucent prey from the diets implicates 

 vision. In many cases, the ASV's, which are mini- 

 mal estimates of the volume searched, seem too 

 high to have resulted from filtering alone. Even 

 assuming that the area filtered is as large as the 

 square of the premaxillary and that the fish swam 

 at 2.5 body lengths/s (Ware 1978) for 5 h, a 50 mm 

 D. trachops, C. warmingi, D. perspicillatus, or L. 

 steinbecki would search only 0.25-0.32 m-' (de- 

 pending on premaxillary length). Yet ASV's were 

 as high as 1.0 m^ for several prey of these species. 

 To search 1.0 m^ visually would require that the 

 fish detect prey within only about 12 mm. Simi- 

 larly, a 20 mm Diogenichthys atlanticus could at 

 best filter only about 0.008 m^, while ASV's of at 

 least five times this were associated with several of 



637 



