322 



Fishery Bulletin 97(2), 1999 



Figure 2 



Area covered by the B-series transects and location of the shelf edge ( 200 ml and 

 the cliffs in relation to the study area. 



range for yellowtail rockfish (Nagtegaal, 1983). We 

 also had to exclude the third of the four nocturnal 

 replicates of transect B4. The vessel had to slow and 

 change course to avoid traffic, leading to an unus- 

 able acoustic estimate. We used an average from the 

 other three nocturnal estimates in our principal 

 analyses to maintain a balanced ANOVA design. We 

 also substituted a range of values in place of the miss- 

 ing value to examine the stability of the results. 



Constant transects allowed us to compare behav- 

 ior over the same location, but in doing so we sacri- 

 ficed the improved statistical power we would have 

 achieved with a stratified random design (Jolly and 

 Hampton, 1991). 



Time of the transect was classified according to 

 the moment the vessel passed over the cliffs, the ob- 

 served center of fish abundance. These times were 

 then classified as diurnal or nocturnal according to 

 the time of sunrise and sunset for Tofino, approxi- 

 mately 100 km southeast of the study area (Atmo- 

 spheric Environment Service, Environment Canada). 



Transect length 



We derived two density estimates, based on a short 

 and a long section from each completed transect, to 

 examine the impacts of the choice of endpoints. For 

 both sections and for each transect, we chose a stan- 

 dard beginning and end in relation to the cliffs (Fig. 

 3). The ends for the short sections of the transects 

 were defined as the shoreward and seaward limits 



of the nocturnal dispersion of the cliff aggregations, 

 judged visually from the echograms for all replicates. 

 An individual transect required about one hour to 

 complete. For these sections, the seaward end was 

 inshore of the shelf break. The long sections used 

 the same shoreward or shallow end, whereas the 

 seaward or deep end was extended to a 180-m bot- 

 tom depth. This depth incorporated the apparent 

 deep-water limit of the shelf break aggregations 

 (Fig. 3). The echograms for replicate transects (Fig. 3) 

 differ in length because of differences in vessel speed 

 over the ground. We calculated an average surface 

 density (g/m'^) based on the constant length for each 

 replicate of the short and long sections from all 

 transects. 



Hydroacoustic equipment and echo processing 



Echo integration was used to estimate fish abun- 

 dance in the survey area (Forbes and Nakken, 1972; 

 Clay and Medwin, 1977; Foote, 1987; MacLennan and 

 Simmonds, 1991). The calibrated echo integration 

 system was operated from the Canadian Coast Guard 

 Ship, W. E. Richer. The acoustic processing equip- 

 ment consisted of a BioSonics 38 kHz Model 101 echo 

 sounder, BioSonics Model 111 chart recorder, 

 BioSonics Model 121 digital echo integrator, PCM/ 

 VCR tape recording system, and auxiliary equip- 

 ment. The transducer components included a towed 

 body with a Simrad ceramic transducer and armored 

 tow cable. 



