68 



Fishery Bulletin 93(1), 1995 



Still and video cameras (baited and unbaited) have 

 been used to sample other marine habitats, e.g. to 

 compare data collected with other methods and to 

 reduce biases associated with remote collection tech- 

 niques (Uzmann et al., 1978; Grimes et al., 1982; 

 Matlock et al., 1991; Moffitt and Parrish, 1992), and 

 to study the natural history and estimate the abun- 

 dance indices of different species (Miller, 1975; Priede 

 et al., 1990; Auster et al., 1991; Armstrong et al., 

 1992; Michalopoulos et al., 1992). Baited video sys- 

 tems have some inherent biases (such as loss of odor 

 attractant (or bait; Isaacs and Schwartzlose, 1975), 

 and intra- or inter-specific competition for bait 

 (Armstrong et al., 1992). Towed video systems and 

 remotely operated vehicles (ROVs) have been used 

 to collect analogous data without the use of bait. A 

 towed system or ROV can sample a known area and 

 assess organismal densities directly within that area. 

 Although a towed and unbaited video system might 

 be preferable if the objective is to estimate absolute 

 fish density, we chose a stationary, baited system both 

 in order to minimize the disturbance of bottom habi- 

 tat and to provide an index of relative abundance simi- 

 lar to catch per unit of effort (CPUE) that would be 

 sufficient for temporal comparisons. 



In this paper, we assess the precision, accuracy, 

 and efficiency of a baited video camera system for 

 producing indices of fish abundance based on a vi- 

 sual record, and we compare these video data with a 

 more traditional abundance index of longline CPUE. 

 We first compared the two methods by pair-sampling 

 during an initial cruise in 1992. Preliminary esti- 

 mates of video precision were made with these data. 

 Power and sample sizes for video sampling were 

 again estimated with data obtained during 1993. 

 Effects of station and depth for the 1993 cruise were 

 also examined. We emphasize data for the target 

 species, opakapaka, but include complementary data 

 for Bleeker's balloonfish ("puffers"), Torquigener 

 florealis, because of its numerical dominance in both 

 types of samples. 



Materials and methods 



Sampling 



Simultaneous longline and video camera operations 

 were conducted for four days offshore of windward 

 Oahu embayments (Fig. 1) and three days offshore 

 of embayments at each of two other islands (Maui 

 and Kauai) during a May 1992 cruise of the NOAA 

 ship Townsend Cromwell. Additional video camera 

 operations were conducted from small craft during 

 May 1993 outside Kaneohe Bay, Oahu (Fig. 1). Sam- 



pling was conducted on flat, unconsolidated bottoms 

 (Parrish, 1989; Moffitt and Parrish 4 ) between 0800 

 and 1530 hours. Longline depths ranged from 54 m 

 to 107 m and video camera depths ranged from 52 m 

 to 87 m for the 1992 cruise, the approximate depth 

 range for juvenile opakapaka (Parrish 1989; Ellis et 

 al. 5 ). Paired video-longline samples were collected in 

 1992 over a longshore spatial scale of 10 nmi (18.5 

 km) off Oahu (Fig. 1). The longshore spatial scales 

 for Kahului Bay, Maui, and Hanalei Bay, Kauai, were 

 9.1 nmi (16.9 km) and 9.5 nmi (17.6 km), respectively. 

 From one to three video deployments were made per 

 station at shallow, mid-, and deep depths along the 

 longline set. A total of 39 longline and video stations 

 were completed (15 at Oahu, and 24 combined for 

 Maui and Kauai). A complete set of three (shallow, 

 mid-, and deep) video camera deployments were com- 

 pleted per station for 10 of the 15 stations off Oahu. 

 In 1993, video sampling was refocused on a finer 

 spatial scale (2.4 nmi or 4.4 km) off Kaneohe Bay, 

 Oahu, over known bottom topography to increase the 

 probability of sampling more homogeneous habitat. 

 Two video deployments per station were made at 

 shallow (73 m to 77 m) and deep (83 m to 85 m) posi- 

 tions in 1993. A total of 18 stations were completed 

 in 1993. 



Video camera and longline stations were parallel 

 within 50 to 75 m of each other. The spacing between 

 video-longline stations during any particular day was 

 approximately 1 km (0.5 nmi) longshore during 1992. 

 Spacing between video deployments was approxi- 

 mately 100 to 300 m. Longshore distance between 

 video stations in 1993 was about 200 m (0.1 nmi). At 

 distances of >100 m separating adjacent stations, 

 successive video deployments were likely indepen- 

 dent, because the greatest distance offish attraction 

 to the bait was only 48 to 90 m. This estimate was 

 based on average maximum bottom current speeds 

 of 0.1 to 0.2 m/s respectively (Bathen, 1978), a soak 

 time of 10 minutes, and a swimming speed for 

 opakapaka of 0.6 m/s (or approximately 3 body 

 lengths (BL) per second, where one BL=20 cm; 

 Videler, 1993). Depths of all video and longline sets 

 were determined by depth sounders aboard the re- 

 search vessels, and positions were determined by 

 GPS (Global Positioning System) or sighting com- 

 pass, as Loran-C capabilities were unavailable. 



Longlines were deployed approximately perpen- 

 dicular to depth contours. Bottom longline operations 

 used modified Kali longlines, 6 each with 30 individu- 



6 Shiota, P. M. 1987. A comparison of bottom longline and deep- 

 sea handline for sampling bottom fishes in the Hawaiian Ar- 

 chipelago. Honolulu Lab., Southwest Fish. Sci. Cent., Natl. Mar. 

 Fish. Serv., NOAA, Honolulu, HI 96822-2396. Southwest Fish. 

 Cent. Admin. Rep. H-87-5, 18 p. 



