Table 2. — Occurrence of bogus prey items in the stomachs of midwaterfishes. Sizes are standard lengths in miUimeters. 



Table 3. — Haul by haul comparisons of postcapture ingestion by Stenohrachius leucopsarus. 

 Measurements are standard lengths in millimeters. 



most abundant fishes in the collection, myctophids 

 Stenohrachius leucopsarus and Tnphoturus 

 mexicanus, showed a large difference in the per- 

 centage of individuals which had ingested the 

 bogus prey items (23.2% vs. 1.5% respectively). In 

 8 of the 10 species that showed net feeding, 

 larger-than-average individuals were more likely 

 to contain artificial prey than smaller ones. 



A haul by haul comparison of data on S. leucop- 

 sarus (Table 3) shows a greater degree of net feed- 

 ing in daytime hauls than at night; although the 

 average size of specimens in nighttime hauls is 

 smaller (/-test,P<0.001). If we consider only those 

 specimens which were equal to or larger than the 

 smallest individual found with bogus prey in its 

 stomach (51 mm), the trend for greater daytime 

 net feeding still, holds. All other species also 

 showed a higher incidence of daytime net feeding 

 (Table 2). 



Discussion 



The ingestion of bogus prey items by midwater 

 fishes is direct evidence that the contamination of 

 stomach contents can and does occur in the cod 

 ends of midwater trawls. The degree to which it 

 occurs, and thus the seriousness of the bias im- 



714 



parted to dietary studies, is apparently variable. 

 Within a collection of midwater fishes, our data 

 and that of Hopkins and Baird (1975, 1977) indi- 

 cate that overall the bias may be low. However, 

 the data from the present study showed that im- 

 portant levels of contamination can occur within 

 some species. Hopkins and Baird (1975) based 

 their low estimates of net feeding on intraspecific 

 comparisons of their paired net data. The same 

 data, when e.xamined interspecifically, reveals 

 that in 14 of 19 comparisons (700 fish from 11 

 species), fishes prevented from reaching the cod 

 end had a lower average number of prey items in 

 their stomachs than fishes which had entered the 

 cod end. The probability of finding no difference in 

 the number of prey items between these samples 

 (i.e., no net feeding) is <10% (Wilcoxon matched 

 pair signed rank test); thus their data indicating 

 net feeding is significant to at least the 90% level 

 of confidence. 



Several factors may be responsible for the ob- 

 served variations in degree of net feeding. The 

 condition and viability of captured fishes is cer- 

 tainly a key factor; hardy species such as S. 

 leucopsarus and Symbolophorus califormensis 

 commonly survive capture and arrive at the sur- 

 face alive and active. The survival of other fishes 



