Starr et al.: Submersible-survey and acoustic-survey estimates of fish density 



15 



Submersible belt transects followed techniques 

 used in SCUBA surveys (Brock, 1954; Brock, 1982; 

 Ebeling, 1982; Sale and Sharp, 1983; Davis and 

 Anderson, 1989) and in our previous submersible 

 surveys (Pearcy et al., 1989; Stein et al. 1992; Hixon 

 et al. 1 ). Each transect consisted of two 30-minute 

 segments, separated by a 10-minute rest period. The 

 submersible started from a preselected location and 

 traveled on a preselected course. Course headings 

 were chosen to keep transects at a uniform depth. 



During the transect, the pilot attempted to main- 

 tain a constant speed of 3 km-hr 1 and a constant 

 altitude of 2 m above the bottom. Actual speed and 

 altitude varied because of the difficulties of piloting 

 the submersible over rough terrain. However, in no 

 case did the submersible observer count fish that 

 were more than 4 m off the bottom. To plot the ac- 

 tual submersible path, the support ship RV Pirateer 

 was periodically positioned directly over the submers- 

 ible by using a Trackpoint II system, and latitude 

 and longitude were recorded with a global position- 

 ing system (GPS) receiver. 



Submersible observation techniques followed those 

 of Pearcy et al. ( 1989) and Stein et al. ( 1992) and are 

 more fully described by Hixon and Tissot. 2 Observ- 

 ers looked forward and downward through a view- 

 ing port in the bow of the DSV Delta to identify fishes. 

 In addition to identifying fish to species when pos- 

 sible, the observer counted and estimated the sizes 

 of all fish observed to the nearest decimeter. We esti- 

 mated fish density by dividing the number of fish 

 counted by the area visually surveyed. The area vi- 

 sually surveyed was calculated by multiplying the 

 length of the transect by the width of the average 

 field of view (2.3 m) along the transect. 



Acoustic surveys 



A total of 14 hours of echo integration and dual-beam 

 target strength data were collected before, during, 

 and after submersible surveys. Acoustic data col- 

 lected during submersible transects were not usable 

 because of interference from the submersible. Acous- 

 tic equipment used in this study included a 120-kHz 



1 Hixon, M. A., B. N. Tissot, and W. G. Pearcy. 1991. Fish as- 

 semblages of rocky banks of the Pacific Northwest (Heceta, 

 Coquille, and Daisy Bank]. A final report by the Department 

 of Zoology and College of Oceanography of Oregon State Uni- 

 versity for the U.S. Department of the Interior, Minerals Man- 

 agement Service Pacific OCS Office. Camarillo, CA. 



2 Hixon, M. A., and B. N. Tissot. 1992. Fish assemblages of 

 rocky banks of the Pacific Northwest [Stonewall Bank). A fi- 

 nal report supplement by the Department of Zoology of Oregon 

 State University for the U.S. Department of the Interior, Min- 

 erals Management Service Pacific OCS Office, Camarillo, CA. 



dual-beam ceramic transducer with nominal beam 

 widths of 10 and 22 degrees deployed in a towed body, 

 a BioSonics Model 101 echosounder with dual 20 log 

 R and 40 log R time-varied-gain receiver board, a 

 Sony digital tape recorder and interface, and a mi- 

 crocomputer used as a signal processor. The micro- 

 computer integrated signals in real time and stored 

 integration values in five-second intervals. Latitude 

 and longitude data obtained from the ship's GPS were 

 automatically written into echo integration files. 

 Dual-beam target strength data were taped concur- 

 rently for processing later. 



In situ dual-beam methods provide measurements 

 of target strength in the natural environment 

 (Ehrenberg and Lytle, 1977) but can be problematic 

 in surveys of schooling fishes because they require 

 resolution of individual organisms. An alternative 

 method of estimating target strength is to use the 

 mathematical relationship empirically derived by 

 Love ( 1971, 1977) to relate fish length and intensity 

 of echoes from the dorsal surface of a fish. The rela- 

 tionship, expressed in terms of acoustic frequency, is 



TS = 19.1 log (L) - 0.9 log (f) - 62.0, 



where TS = target strength in decibels (dB); L = fish 

 length (cm); andf= frequency (kHz). 



To avoid the problems associated with dual-beam 

 methods caused by schooling fishes, we chose to use 

 the mean length offish observed (Love's equation) to 

 scale echo integrator output. Although Love's equa- 

 tion provided the primary method for scaling echo 

 integrator output, we also generated target strength 

 estimates using dual-beam data. We compared tar- 

 get strength estimates obtained from Love's equa- 

 tion (Love, 1971) with those generated by dual-beam 

 acoustic methods by converting the dual-beam esti- 

 mates of target strength from logarithmic units (deci- 

 bels) to linear units, using the equation TS = 10 log 

 (<T), where TS = target strength and o- backscatter- 

 ing cross section, a linear measure of the reflective 

 nature of a target. 



Ship speed on transects ranged from 0.5 to 3 msec" 1 ; 

 thus integration sequences covered about 3-15 m of 

 linear bottom. In the vertical dimension, we summed 

 echoes in 2-m depth bins from the surface to the bot- 

 tom. Acoustic surveys provided both areal (fish-m -2 

 over water column sampled) and volumetric (fish-m -3 ) 

 estimates offish density. 



To compare the submersible and acoustic surveys, 

 we divided the acoustic data into two strata. In one 

 stratum, we summarized the echo integration data 

 collected from the surface to 4 m above the bottom. 

 These data represent fish that were above the area 

 that the submersible surveyed ("above sub" stratum ). 



