FISHERY BULLETIN: VOL. 73, NO. 4 



in only two pairs: Lepidophane^ guentheri, tow 167 

 (P<0.001) and Valenciennellus. fripunctulafus, 

 tow 135 (0.005 >P>0.001). In the former, those in- 

 dividuals from the cod end took more prey items in 

 smaller size classes while in the latter the reverse 

 occurred. We have no simple explanation for these 

 results. It is difficult to attribute them, however, to 

 net feeding since other diet characteristics tested 

 showed no significant differences for these same 

 sets. Additionally, paired samples of the same 

 species from other collections revealed no sig- 

 nificant differences. 



Prey Diversity 



In comparison to the coarse mesh fish-catcher, 

 the unobstructed cod end net of the adjacent trawl 

 contained a much greater variety of plankton and 

 consequently a more diverse potential food source 

 for net feeding. A comparison was made of diver- 

 sity of food items in stomachs of fish from each 

 side of the trawl using 12 (of 19) species-pair 

 collections represented by sets of approximately 

 equal numbers of individuals for each cod end 

 type. Total diversity was scored for each set of 

 fishes, yielding two diversity values for each 

 species-pair collection. Diversity scores were then 

 summed to give grand means for each cod end 

 type. 



On the basis of a f-test on logjo transformed 

 data, no significant (P>0.05) difference was in- 

 dicated for the two cod end types though total 

 diversity was considerably greater in fishes from 

 the plankton net cod end in some sets (e.g., Ar- 

 gyropelecus aculeatus, tow 144; L. alatus, tow 

 141). 



Fish Scales 



Anderson (1967), in his analysis of the diet of 

 Bafh)jlagi(>i stilbiiis, frequently encountered fish 

 scales in stomachs yet no other remains of fish of 

 the size indicated by the scales. This, in addition to 

 the absence of scales in intestines and the oc- 

 currence of scales and copepods in the mouths of 

 fish, he considered as evidence of net feeding. In 

 the present study, fish in half the sets of samples (6 

 of 12) for which data are presented contained no 

 fish scales. In four of the remaining six pairs, more 

 scales were found in fish from the cod end where 

 scales would be expected to accumulate during the 

 course of a tow, but n one of the differences were 

 significant {f-test on \/A' + 0.5 transformed data; P 

 >0.05). 



The occurrence of fish scales in stomachs does 

 not necessarily stem from predation on smaller 

 fish or from eating scales abraded from fish within 

 the trawl. Fish scales appear to be common in the 

 water column and thus available as separate 

 forage items. In a series of paired 30-liter bottle 

 casts made between and 1,000 m in August 1972, 

 in the eastern Gulf of Mexico where most of the 

 fish examined were taken, scales occurred in 

 collections (60 liters/sample) from 7 of 15 depths 

 sampled at densities of 17-83 per m^. Scales ranged 

 from 0.5 to 5 mm in diameter. No fish were taken in 

 the sample bottles and the probability of con- 

 tamination from other sources appears low. 



Taxonomic Composition of 

 Stomach Contents and Plankton 



Table 2 presents the principal taxonomic com- 

 ponents of prey found in nine sets of fish from both 

 sides of the trawl. The principal diet item was the 

 same in both sets of fish in six of nine collections, 

 the same prey constituted the top three food items 

 by number in seven of nine collections and the 

 prey taxa were in the same rank order in five of 

 nine collections. The principal three prey taxa in 

 fish from either side of the trawl were within ±3% 

 of the mean value for both sides from each tow in 

 25 of 29 food item comparisons and all values were 

 within ±10% of the means. These results show 

 that the taxonomic composition of at least the 

 principal components of the diet was similar in fish 

 from both sides of the trawl for all comparisons. 



Comparison of food items in stomachs of fish 

 from the cod end, where net feeding is assumed 

 mostly likely to occur, with plankton catches 

 reveals little similarity in the top three taxonomic 

 components. In none of the nine collections was 

 the principal taxon the same in either the plankton 

 net catch or in the stomachs of fish from the cod 

 end. Of particular importance are tows 137, 141, 

 144, 152, and 161 which sampled relatively narrow 

 depth zones and consequently were potentially 

 less influenced by vertical stratification of plank- 

 ton. Also, three species of fish collected in the same 

 haul (tow 137) each contained a different principal 

 food item, none of which matched the most abun- 

 dant taxon in the cod end plankton catch. The 

 major diet components for Bciifliosema siihorbi- 

 fale, C. icarmhigi, and L. guenfheri from tow 137 

 were Oncaea, Limacina, and Pleuromamma; the 

 most abundant plankton in the cod end net were 

 ostracods (Conchoecinae). This particular haul' was 



912 



