HOPKINS and BAIRD: NET FEEDING IN MESOPELAGIC FISHES 



OROM MCT 



VELCHO SEAM 



FISH-CATCHE* SLEEVE 

 ( l.lcm STRETCHED MESHI 



OSn Dl AM ETE II-.133 mm MESH 

 rL ANKTON NET 



Figure 1.— Double-net closing Tucker trawl showing modified (fish-catcher sleeve) and conventional cod ends. 



items contained in the mouth, pharyngeal region, 

 esophagus, or intestine were not included in the 

 analysis. Student's /-test was used for most sta- 

 tistical comparisons, the distributions being first 

 tested for skewness and kurtosis (Sokal and Rohlf 

 1969). If the data to be tested were not normally 

 distributed, then a log or square root transform 

 was applied and the resulting distributions re- 

 tested for skewness and kurtosis. /^-tests were 

 used to determine homogeneity of variances; the 

 few exceptions to homogeneity will be discussed 

 under the appropriate section. Standard chi- 

 square tests were used to test the similarity of 

 prey size distributions. 



To compare prey availability in the trawl cod 

 end with fish stomach contents, plankton was 

 examined from tows 135, 137, and 141-144 (Table 

 2). Trawl mouth plankton net collections were used 

 in preference to cod end catches because of possi- 



ble losses of small plankton incurred in sorting 

 larger fish and shrimp from the latter. Both cod 

 end and trawl mouth nets for these particular tows 

 had the same mouth area (0.2 m^ and mesh size 

 (333 /xm) and were considered comparable in 

 fishing characteristics, at least for the smaller 

 plankton typically found in fish stomachs. Trawl 

 mouth plankton collections for tow 137 were not 

 available, consequently the catch from the cod end 

 net of the fish-catcher side of the trawl was sub- 

 stituted. This particular cod end collection had not 

 been "rough sorted" for larger organisms and was 

 in excellent condition. 



Zooplankton was identified, usually to genus, 

 and counted in each of two subsamples. One 

 plankton net collection (two subsamples) was 

 analyzed for tows 135, 137, and 142 while two 

 collections (four subsamples) were examined for 

 the remaining tows. The mean number of 



909 



