CLARKK: FKKDINC HABITS OF STOMIATOII) FISIIKS 



tain prey species much more frequently than 

 would be predicted by random nonselective feed- 

 ing. Examples include nonmyctophids, particu- 

 larly Bregmaceros spp., in the diet of Thysanactis 

 dentex; Bolinichthys longipes in Eustomias bifi- 

 lis; Howella spp. in E. enbarbatus; Diaphusspp., 

 Howella spp., and Scopelosaurus spp. in the 

 Eustomias spp. with two pectoral rays; and Noto- 

 lychnus valdiviae in Leptostomias spp. The prob- 

 ability of drawing at random, e.g., two Howella 

 spp. or two Scopelosaurus spp. out of two fish 

 from the fauna is very low. 



The relative sizes of fish prey for most species 

 of predators were 20-30% of SL. Many of the val- 

 ues over 30% were for relatively slender prey 

 such as Bregmaceros or Scopelosaurus. Only the 

 Bathophilus and Photonectes spp. and small 

 Chauliodus sloani appeared to take prey >30% 

 routinely. The values for /. fasciola were mostly 

 <20%; this species is, however, so slender that the 

 relative size in terms of head length or body 

 weight would be more like those for the other 

 species. 



If the weights of predator and prey were both 

 similarly related to the cube of the length, then 

 20-30% relative length gives a value of 0.8-2.7% of 

 body weight per item. If anything, this is prob- 

 ably an underestimate of average prey size, since 

 the predators are for the most part slenderer 

 than their prey. Also the stomiatoids seem to be 

 softer bodied than most of their prey and may 

 have a higher water content (cf. Blaxter et al. 

 1971); this would mean that relative prey size in 

 terms of dry weight is higher. Borodulina (1972) 

 gives lengths (SL ?) and wet weights of 14 fish 

 prey and stomiatoid predators. The relative 

 lengths of prey were 12-52% and relative weights 

 0.1-2.8%. The latter are probably underestimates 

 of relative weight since some losses of prey 

 weight must have occurred even in specimens 

 still intact enough to be measured. 



With the exceptions of A. indicus and T. den- 

 tex, all species with chin barbels fed exclusively 

 or nearly so on relatively large fish. The barbel is 

 rudimentary in C. sloani, which was also pisciv- 

 orous except at the smallest sizes, but C. sloani 

 has an elongated first dorsal ray with a light 

 organ on the tip. Of the other nekton-eating spe- 

 cies without barbels, fish were absent from the 

 diets of the Photostomias spp. and eaten only by 

 the largest M. niger and A. "cyaneus." The large 

 gonostomatids also lack a barbel. Fish were less 

 frequent than large crustaceans in their diets 

 and were relatively smaller than fishes eaten by 



the predators with barbels. Tactostoma macro- 

 pus, the only melanostomiatid known to eat pri- 

 marily crustaceans (Borodulina 1972), has the 

 smallest and most rudimentary barbel in the 

 family. 



Although there are no directly supportive data 

 or citations, it is undoubtedly true that in the 

 open ocean, crustaceans far outnumber fishes at 

 lengths <15-20 mm; for lengths >25-35 mm the 

 opposite is probably true. It is also probably true 

 that a fully metamorphosed fish is a faster swim- 

 mer than a similar-sized crustacean and, other 

 things being equal, more likely to evade capture 

 when attacked by a predator. Thus a predator 

 which preferred items 20-30% of its length and 

 actively searched for prey would have a diet simi- 

 lar to those of A. indicus and A. "cyaneus." The 

 small predators would encounter crustaceans 

 much more frequently and probably capture 

 those encountered more frequently than they 

 would fish, while the large predators would 

 almost be forced into piscivory due to the relative 

 rarity of appropriate-sized crustaceans. If the 

 predator preferred prey only 10% of its body 

 length, the diet would resemble those of the large 

 Gonostoma spp., where even the largest individ- 

 uals (100-200 mm) would still encounter more 

 crustaceans than fish in the appropriate size 

 range. 



Most of the fishes with barbels must either re- 

 ject crustaceans encountered or feed other than 

 by active search. A plausible and likely hypoth- 

 esis (which has been suggested by others) is that 

 they are "passive" and use the luminescent 

 bodies in the barbel to attract prey. Bertelsen 

 (1951) developed a similar hypothesis for the 

 ceratioid angler fishes. Since several of the prey 

 fishes are not known to be bioluminescent them- 

 selves, it is most probable that the barbel mimics 

 food of the prey species — most of which appear to 

 be primarily visual feeders (see above) — rather 

 than a conspecific of the prey. The large crusta- 

 ceans apparently are not similarly attracted; this 

 is not surprising in view of their very different 

 eyes and probably different diet and feeding be- 

 havior. Thus the barbel may be an adaptation for 

 attracting and perhaps aiding in capture of rela- 

 tively large fish. This mechanism could allow 

 these stomiatoids to subsist on relatively large 

 prey whose densities are quite low (on the order 

 of 1/m 2 of sea surface, see Maynard et al. 1975) 

 with less energy expenditure than would be re- 

 quired for active search and capture. Further- 

 more, assuming the findings of Pandian (1967) 



301 



