FISHERY BULLETIN: VOL. 71. NO. 2 



twice and others not at all during the September 

 1970 series. This also occurred to a lesser extent 

 during the December 1970 series. All night 

 tows above 200 m were made during new moon 

 for the March 1971 and June 1971 series. The 

 March 1971 series was incomplete due to a 

 storm, and the 400-700 m zone was not sampled 

 at night. Such limitations are taken into account 

 in interpreting the data. 



The CT tows were taken in conjunction with 

 the March 1971 IK series. Eight night tows 

 were taken at 25-m intervals between 25 and 

 200 m during the same three nights that IK 

 tows were taken at or near these depths. The 

 IK-CT pairs were unfortunately not all taken 

 simultaneously. All tows but one were 2 hr at 

 depth; the 25-m CT tow was only 15 min long, 

 but catches of some species were sufficient for 

 useful analyses. Two day tows at 300 and 400 m 

 were also taken with the CT during this period. 



To determine effects of moon phase on vertical 

 distribution and avoidance at night, lO-ft IK 

 tows of IV2 hr each were made at depths of 

 15, 45, 65, 80, 100, 125, 145, and 165 m on 17-19 

 September 1971 (new moon) and at 20, 50, 75, 

 100, 130, 170, and 190 m on 4-6 October 1971 

 (full moon). This series, designated September 

 1971, also provided useful data on other aspects. 



During December 1970 and June 1971, three 

 and four, respectively, short oblique tows were 

 made to depths of 250-300 m at night in order 

 to roughly estimate the catch due to ascent 

 and descent through the upper layers. The 

 cable was paid out as described above, but the 

 ship was slowed and the trawl retrieved im- 

 mediately. A longer oblique tow was made in 

 June 1971 to 330 m at night. Fifty meters of 

 cable was paid out every 15 min in an attempt 

 to sample all depths equally. 



The dusk and dawn tows taken during July 

 1970 indicated that myctophids had completed 

 their upward migration by 2000 hr and had 

 begun to descend by 0400-0500 hr. All sub- 

 sequent night tows were taken between these 

 times. Differences between tows taken in 

 sequence at 50 and 100 m through the night 

 were not markedly greater than differences 

 between tows taken at the same hour on two 

 successive nights. There was no trend among 

 the species which indicated that anything 



analogous to a "mid-night dispersal" occurred. 

 Therefore, all night tows were considered to- 

 gether regardless of what period of the night 

 they were taken. No comparable study was 

 made to check on possible changes in depth 

 distribution during the day. It was assumed 

 that no changes took place during the day 

 between 0800 and 1600 hr. 



Specimens were preserved in Formalin. 

 Standard length was measured to the nearest 

 mm for all collected by IK. For some CT collec- 

 tions of over 1,000 individuals per species, a 

 subsampl.e of several hundred was measured. 

 Wet weights of blotted specimens were deter- 

 mined to the nearest mg for small individuals 

 and to the nearest 10 mg for larger fish or 

 whole samples. 



Insufficient replicate samples were taken to 

 reliably specify the variance associated with the 

 estimate of abundance from a single tow. There 

 was considerable variability in the catches of 

 the two series of four tows each taken at 50 

 and 100 m in July 1970. (Appendix Table 1.) 

 In particular, one tow in the 100 m series 

 caught much higher numbers of several species. 

 This variability, probably related to patchiness 

 in the organisms sampled and differences in 

 avoidance due to differences in ship's speed, 

 cloud cover (light), etc., limits interpretation 

 of some of the data. Where I have drawn con- 

 clusions without statements of statistical sig- 

 nificance, I have attempted to be conservative. 

 Features not consistently evident from a series 

 of tows should be regarded as tentative. 



The significance of differences between two 

 samples in size composition was determined 

 using the Kolmogorov-Smirnov test (Tate and 

 Clelland. 1957). Two samples were considered 

 significantly different if the probability (one- 

 tailed) associated with the maximum difference 

 between the cumulative size-frequency curves 

 was less than 0.05. The overall significance of 

 the differences is somewhat altered because 

 multiple tests were often made with the same 

 data. In most cases, the trends observed were 

 obvious and consistent, and it is doubtful that 

 large errors resulted from the procedure. 



There were relatively few significant dif- 

 ferences in size composition between replicates 

 at the same depth. For 10 sets of data for eight 



404 



