74 



Fishery Bulletin 93(1), 1995 



be limited in practice by the duration and availabil- 

 ity of a large research vessel (a maximum of five days 

 at one location) and by estimates of maximum effort 

 attainable for each gear. The latter considered the 

 time needed for deployment, retrieval, and process- 

 ing of specimens for longlines and the time required 

 for video camera battery and tape changes. The esti- 

 mated upper limit for sample size was 30 stations 

 (at 6 stations per day) for both the longline and the 

 video cameras; a video station consisted of three cam- 

 era deployments. All 1992 data from windward Oahu 

 (n=15 stations) were analyzed first. Sample sizes of 



o 

 z 

 x 



< 



c 



< 



2 



3.0 

 2.5 

 2.0 

 1.5 

 1.0 

 0.5 

 0.0 

 -0.5 



1.5 

 1 .0 

 0.5 - 

 0.0 



-i 1 1 r 



MAXNO = 0.32 + 

 p < 0.001 



r 2 = 0.62 

 n = 15 



I I 



0.69CPUE 



-0.5 



■1.0 



0.0 0.5 1.0 1.5 2.0 2.5 3.0 



Opakapaka longline InCPUE 



3.5 



Figure 4 



(A) Scatterplot and fitted regression line for the video index maxi- 

 mum number seen (InMAXNO) and longline InCPUE for 

 Pristipomoid.es filamentosus (opakapaka) for 1992 data. Four double- 

 zero observations were coincident. (B) Plot of residuals from the 

 video versus longline regression of Fig. 4A versus longline InCPUE for 

 1992 data; zero line is included. Four points were coincident as in 4A. 



26 stations for video and 33 stations for longlines 

 were estimated as necessary to detect a twofold 

 change in numbers of juvenile opakapaka. 



Power was next estimated by using only those 1992 

 windward Oahu stations with the full complement 

 of three deployments (n=10). The variability ( V=81%) 

 of the video MAXNO index at these 10 stations (3 

 deployments inclusive) did not differ from the total 

 Oahu data set. A sample size of 17 stations (51 de- 

 ployments) was estimated as necessary to detect a 

 twofold change in the MAXNO index, within the prac- 

 tical limit of 30 stations. The estimated sample size 

 for longline, based on data for these same 10 

 stations, was still 33 stations, slightly over 

 the limit of 30 stations. 



Power was next reexamined to determine 

 the effect of reducing effort to two video de- 

 ployments per station, rather than to three. 

 By using only the shallow and deep sets of 

 the 10 1992 stations with three complete de- 

 ployments, variability improved slightly 

 (V=76%), and only 18 stations (36 deploy- 

 ments) were estimated as necessary to de- 

 tect a twofold change for opakapaka. Video 

 MAXNO remained significantly correlated 

 with longline CPUE for opakapaka by using 

 two deployments per station (r=0.66, P=0.04, 

 n=10). 



The 18 pairs of 1993 video data were used 

 to reevaluate power and sample size for a 

 smaller, more homogeneous study area. The 

 LM' form of the video MAXNO index was 

 used to avoid possible bias introduced by 

 adding a constant to the data. The calcula- 

 tions used 17 pairs of deployments because 

 the mean of one data pair was zero, the log 

 of zero being undefined. Variability (V) for 

 the 1993 data improved to 49% with these 

 refinements. Twenty-two stations were esti- 

 mated as necessary to detect a twofold change 

 in abundance at a 2 =0.05. If a 2 were relaxed 

 to 0.1, 17 stations would be necessary to de- 

 tect a twofold change. In practice, slightly 

 greater numbers of samples would be neces- 

 sary, because deleting the single zero-mean 

 datum artificially inflated our power esti- 

 mate. Nonetheless, the 1993 samples provide 

 the best data for evaluating precision that 

 are presently available. 



Comparison with other baited camera 

 studies 



Previous baited camera studies of deep-sea 

 species (Priede et al., 1990; Armstrong et al., 



