HOPKINS and BAIRD: NET FEEDING IN MESOPELAGIC FISHES 



Table 2.-ReIative abundance of principal (top 3) food items in fish stomachs and in plankton taken concurrently with 

 fish, with 333-jLun mesh nets mounted in the mouth of the double trawl (see Figure 1): F = "fish-catcher" set of fish; P 

 = cod end plankton net set. 



stomachs resulted in significant skewness in the 

 distributions of many untransformed data sets. In 

 three set comparisons there were significant 

 differences (F- tests, 0.05 >P>0. 025) in variance 

 {BentJwsema suborbitale, tow 137; C. ivanningi, 

 tow 137; Lampanijctus alatus, tow 167). In these 

 cases, tests comparing means of normal distribu- 

 tions when population variances are unequal were 

 applied as described by Johnson and Leone 

 (1964:226). 



There were significant {t-tests, P<0.05) 

 differences in 5 of 19 comparisons of number of 

 prey items per stomach. Lampanyctus alatus in 

 two collections contained more prey items per in- 

 dividual in fish taken from the plankton net cod 

 end (tow 141: 0.025 >P>0.01; tow 167: 0.05 >P 

 >0.025). However, for Gonostoma elongatum in two 

 sets (tows 144, 145: 0.05>P>0.025), and C. warm- 

 ingi in one set (tow 161:P<0.005), individuals from 

 the fish-catcher side averaged more prey per 

 stomach. Because of possible diurnal feeding 

 periodicity in mid-water fishes (Anderson 1967; 

 Hoiton 1969; DeWitt and Cailliet 1972; Baird et al. 

 1975), fish entering the trawl at different periods 

 in their feeding cycle may be satiated or have a 

 different predisposition to feed in varying 



degrees. The five sets of fish showing significant 

 differences in number of food items, howeVer, are 

 not conspicuously grouped in any single time 

 period (see Table 1) and no general relationship is 

 apparent in our results between time of capture 

 and relative abundance of prey in fish from either 

 side of the trawl. 



Mean Prey Size 



In 8 of 19 data sets, mean prey size was smaller 

 in cod end fish. The major size modes were coin- 

 cidental in all 19 set comparisons as judged from 

 visual inspection. A ^-test of the grand means 

 (mean of 19 individual means for each cod end 

 type), however, revealed no significant (P>0.05) 

 difference in mean size of food item for fishes in 

 either side of the trawl (variance of means 

 homogeneous). Though the sensitivity of this test 

 is weakened to some degree by comparing 

 different species of fish collected at different 

 times, a strong bias in prey size resulting from net 

 feeding is not apparent. 



Prey size distributions for 14 paired sets were 

 also compared using the contingency chi-square 

 test. Significant (P<0.05) differences were found 



911 



