284 



Fishery Bulletin 94(2), 1996 



tist counted rockfish through a starboard porthole 

 0.5 m above the bottom of the submersible. Rockfish 

 counts and behavior were audio- and video-recorded 

 for subsequent analysis. Rockfish densities were de- 

 rived from the number of fish counted and the sea- 

 floor area searched. This area was calculated from 

 the distance the submersible traveled and the esti- 

 mated lateral range of visibility from the starboard 

 porthole. The observer estimated that lateral visibil- 

 ity ranged from 3 to 7 m, depending on water clarity. 

 The ability of the observer to estimate distance was 

 tested by three methods: 1) a length of pipe marked 

 at meter intervals was laid on the seafloor and ob- 

 served on two dives; 2) a hand-held sonar gun pro- 

 vided distance readouts to rock formations located 

 at the maximum visibility range on six dives; and 3) 

 the submersible's sonar provided distance readouts 

 when the seafloor became visible during descents and 

 when the seafloor disappeared from view during as- 

 cents. Observer estimates of distance were consis- 

 tently within 1 m of actual distances. Accuracy to 

 within 1 m was possible because constant illumina- 

 tion was provided entirely by the submersible lights, 

 and water clarity was consistent at each site. The 

 pilot assisted in counting fish >4 m above the sea- 

 floor from a panoramic view 2 m above the scientist. 

 The pilot could see at least 10 m above the seafloor. 



Identification of Pacific ocean perch from the sub- 

 mersible was not always possible because they look 

 similar to three other red rockfish species: redstripe 

 rockfish, S. proriger; sharpchin rockfish, S. zacentrus; 

 and harlequin rockfish, S. variegatus. Therefore, 

 rockfish density estimates from submersible counts 

 and from trawling include these three species in ad- 

 dition to Pacific ocean perch. Most rockfish observed 

 from the submersible, however, were identified as Pa- 

 cific ocean perch by their symphyseal knob and body 

 shape. Trawl sampling confirmed their dominance: 72<7r 

 of the rockfish caught were Pacific ocean perch. 



A detection probability (DP) from the submersible 

 was estimated by measuring angles to targets with 

 a clinometer. Clinometer readings were collected from 

 a fixed height above the seafloor to red sea stars, 

 Luidiaster spp., on the seafloor; sea stars were se- 

 lected as targets because their color was similar to 

 red rockfish and they were abundant, stationary tar- 

 gets. The DP was assumed to equal 1.0 near the sub- 

 mersible and to decrease to zero at some distance 

 from the submersible. The DP was computed as the 

 ratio of the observed and expected density of targets. 

 The observed population density was computed as 



P,=N,/D,. 



where N : is the observed count in angle interval i, 



and D t is the distance along the seafloor correspond- 

 ing to angle interval i. The expected density was com- 

 puted as the average observed density within the full- 

 detection range (the range where the observed den- 

 sity did not decrease). 



Bottom trawling 



Trawl gear consisted of a 400-mesh Eastern otter 

 trawl equipped with 55-m bridles between the net 

 and 1.5 x 2.1-m doors weighing 386 kg each. The 

 average distance between the wingtips ( 15.8 m) and 

 the trawl doors (45.0 m) was determined during four 

 trawl hauls by using mensuration gear. Krieger 

 ( 1993) used a wingtip spread of 14 m, based on pre- 

 vious net measurements of the 400 Eastern trawl, to 

 estimate the preliminary catchability coefficent 

 (NMFS, 1993). A net height (foot-rope to head-rope) 

 of 1.8 m is based on previous mensuration gear mea- 

 surements of 1.4-1.8 m (NMFS, 1993). 



Trawl catches were processed for total number and 

 weight by species. Rockfish density estimates were 

 derived from the number of rockfish captured and 

 the seafloor area swept by the net. The seafloor area 

 swept was the distance trawled multiplied by the 

 horizontal spread of the net. Trawl distances (0.93- 

 1.74 km/haul) and trawl speeds (5.0-6.5 km/h) were 

 calculated from position fixes with a global position- 

 ing system (GPS). Trawl periods ranged from 10 to 

 18 min, depending on trawl speed and trawl distance 

 needed to intersect the submersible transects. At the 

 completion of the 10-18 min trawl period, vessel 

 speed was reduced while the trawl was retrieved. To 

 determine the sampling capabilities of the trawl dur- 

 ing retrieval, catch rates from six "standard" trawl 

 hauls were compared with catch rates from seven 

 "retrieval" trawl hauls. A "standard" trawl haul was 

 followed by a "retrieval" trawl haul at randomly se- 

 lected sites separated by <3 km distance by using 

 the same trawl depths, cable ratios, and vessel speeds 

 as those for trawling at the submersible sites. "Stan- 

 dard" trawl hauls consisted of the net sweeping the 

 seafloor for 10 min at 5.5-6.0 km/h and then being 

 retrieved at reduced vessel speed. "Retrieval" trawl 

 hauls consisted of the net sinking to the seafloor and 

 then being retrieved immediately at reduced vessel 

 speed. A sonar net-sounder was used to determine 

 when the net departed the seafloor. GPS fixes and 

 the rate of cable retrieval were used to determine the 

 speed of the trawl during retrieval. 



Data analysis 



Rockfish density estimates from submersible counts 

 and trawl catch rates were based on the total sea- 



