FISHERY BULLETIN: VOL. 81. NO. 3 



TOTAL INDIVIDUALS 



+ 

 x 



o 



< 



q 

 > 



Q 



o 



QC 



LU 

 GQ 



< 



LU 



LU 



O 



5.0-1 

 4.0- 

 3.0 

 2.04 



^ 1.0H 



4.0H 

 3.0 



2.0 



1.0 



■W- 



Depth 12 - 16m 



Depth 18 - 27m 



-■k 



+- + 



/H, 



DAY 



MONTHS (1979 - 1981) 

 — NIGHT o» 



FIGURE 4.— Monthly variation in the total number of individual fishes captured during the day and night in each of three depth blocks over 

 the study period. Each value represents the grand mean of log 10 (X + 1) transformed catches from day and night net-hauls. Vertical bars 

 depict ±2 standard errors of the respective grand mean. 



(Table 2). The probability for a date effect on E. mor- 

 dax was marginal at P = 0.09. 

 Diel, depth, and date effects were variously impor- 

 tant for four other common taxa or categories (Table 

 3). Interactions between main effects could not be 

 evaluated for these species because we were forced 

 to use Wilson's ANOVA without replication in order 

 to minimize the number of zeros in the catch data. 

 Peprilus simillimus was more abundant at night (diel 

 effect) and on certain cruises during the study period 

 (date effect). Catches of G. lineatus were higher at 

 night and at the shallow depth. Scomber japonicus was 

 more abundant offshore and at certain times of year. 



Species Associations 



Quantitative clustering of species by sample abun- 

 dance among depth blocks and diel periods pro- 

 duced four distinct groups within 21 species (Fig. 5). 

 The largest dichotomy (~1607r distance) occurred 

 between members of Groups I-II and Groups III-IV. 

 Separation within both Groups I-II and Groups III-IV 

 occurred at ~65% distance. 



In Group I E. mordax was most dissimilar, fusing 

 with other group members at ~55% distance. 

 Seriphus politus and G lineatus formed a very close 

 association, being linked at ~12% distance. Peprilus 



576 



