Adams et al.: Population estimates of Pacific coast groundfishes 



447 



tained from the catch-at-age data to determine how 

 close the population is to a threshold value of over- 

 fishing (Kimura, 1989). Survey data also yield valu- 

 able information on migration routes, or biological 

 parameters such as age-at-maturity, fecundity, and 

 feeding. Biased survey estimates can result in very 

 precise estimates of population abundance which 

 are either lower or higher than the true population 

 abundance. 



For environmental surveys, the problem of preci- 

 sion is the major concern. Here, the goal is to detect 

 a trend in population size, often of all the fish in the 

 habitat, and to distinguish that trend from natural 

 variation in fish populations. Abundance estimates 

 from trawl surveys tend to have very large variances; 

 often means and variances are correlated (see Lenarz 

 and Adams, 1980). Resulting confidence intervals 

 around means range from 50 to 100% for flatfish spe- 

 cies and are greater than 100% for rockfish, Sebastes 

 spp. (Raymore and Weinberg, 1990). As a result, all 

 but the most extreme changes in population size are 

 masked by these large confidence intervals. Meth- 

 ods commonly used to deal with this variability are 

 transformations using the negative binomial (Lenarz 

 and Adams, 1980) or the Delta distribution (Pen- 

 nington, 1986). Data transformed by using these dis- 

 tributions often result in the variance being inde- 

 pendent of the mean; however, the large confidence 

 intervals and low statistical power remain. 



In this study we examined video transects con- 

 ducted from a remotely operated vehicle (ROV) as 

 an alternative method for making population estimates 

 of demersal fishes and compared these estimates to 

 those from a conventional swept-area trawl survey. 



Methods 



Study site 



The study site off central California lies along a cone- 

 shaped ridge which runs southwest from Santa Cruz, 

 California, and separates Monterey Canyon to the 

 south from Ascension Canyon to the north (Fig. 1). 

 Surface topography of the ridge is smooth and rela- 

 tively unbroken (Greene 1 ). The ridge is composed of 

 sandstone, is covered with mud of pelagic origin, and 

 is characterized by occasional areas of exposed bed- 

 rock. The stations where ROV and trawl operations 

 were conducted were located along the ridge at depths 

 of 200, 400, and 600 m (see Fig. 1). 



ROV operations 



ROV operations were conducted by using the 

 Monterey Bay Aquarium Research Institute's ROV 

 Ventana aboard the RV Point Lobos. The ROV was 

 equipped with a Sony DXC-3000 video camera with 

 a 5.5-44 mm zoom lens, illuminated by four 400-W 

 sodium scandium lights. The zoom lens was used only 

 for identification off the transect. Fiber-optic cable was 

 used for viewing and recording images. The ROV was 

 also equipped with a combined dual signal, a global 

 positioning system and sonar system, which recorded 

 the ROV position every 10 seconds. Depth, altitude off 

 bottom, and various camera settings were also recorded. 

 At least three replicate transects were made at 

 each depth. Transects at the 200-m, 400-m, and 600-m 

 depths were sampled (for dates, see Table 1). Because 

 a video transect covering the same total area as a 

 trawl was not practical, a transect length of similar 

 distance covered by a trawl was chosen (approxi- 

 mately 1.8 km or 1 nmi). Strip transects were used 

 rather than line transects because the orientation of 

 the lights produced a very sharp boundary between 

 illuminated width of the transect and the darkness 

 (Burnhametal., 1980; Butler etal., 1991). Transects 

 were made at a speed of approximately 1.8 km/hr (1 

 knot) parallel to the isobath, interrupted occasion- 

 ally by stops for fish identification or vehicle main- 

 tenance. Transects were made with a camera angle 

 of 30° off the parallel horizon to the bottom and with 

 a camera height averaging 0.7 m off the bottom. Fish 

 were identified from videotapes by two independent 

 viewers, and the response of each fish to the ROV 

 was recorded as follows: 1) strongly attracted (rap- 

 idly moving into the frame); 2) weakly attracted 



1 Greene, G. U.S. Geological Service, Pacific Marine Geology, 345 

 Middlefield Rd., Menlo Park, CA 94025. Personal commun 

 1992. 



Table 1 



Dates of remotely operated vehicle (ROV) and trawl sam- 

 pling cruises at 200-m, 400-m, and 600-m depth strata off 

 central California. The number in parentheses is the num- 

 ber of transects or trawls occurring during that month. 



Depth stratum (m) 



ROV 



Trawl 



200 



400 



600 



Oct 1991 (3) 



Mar 1991 (1) 

 Oct 1991 (1) 

 Oct 1992 (2) 



Jul 1991 (1) 

 Sep 1991 (2) 



Apr 1991 (2) 

 Sep 1991 (3) 

 Jan 1992 (3) 



Apr 1991 (1) 

 Sep 1991 (3) 

 Jan 1992 (4) 



Apr 1991 (2) 

 Sep 1991 (4) 

 Jan 1992 (3) 



