Ellis and DeMartmi. Video camera sampling of Pristipomoides filamentosus abundance 



71 



1)]. The number of stations where each species was 

 caught or seen was also tallied for each gear type. 

 For nonzero mean data collected in 1993, the best 

 form of the video index (LM\ see Results section), 

 was calculated as follows: 



(4) 



where x - individual datum for MAXNO and 

 rc=number of deployments per station. 



A matrix of Pearson's correlation coefficients was 

 calculated for 1992 data (SAS, 1987) with the log- 

 transformed variables to detect interrelationships 

 among all the video and longline indices. Spearman's 

 rank correlations were also calculated and compared. 

 Multiple linear regression (SAS, 1987) was used to 

 estimate the effect of competition between opakapaka 

 and puffers for longline hooks on the basis of the fol- 

 lowing model: 



Y = p 1 X 1 + p 2 X 2 +e, 



(5) 



where F=ln (opakapaka video MAXNO), Xj=ln (no. 

 hooks lost + no. puffers caught), andX 2 =ln (number 

 of opakapaka caught). The model was run as a for- 

 ward regression without an intercept and with an 

 entry level for significance equal to P<0.10. The pre- 

 cision (repeatability) of video and longline was de- 

 scribed by the coefficient of variation (V, Sokal and 

 Rohlf, 1981; Zar, 1984): 



V = f-^-|xl 

 I mean , 



(6) 



where SD is the standard deviation. 



Longshore station and relative depth effects for 

 1993 data were analyzed by using standard para- 

 metric and nonparametric procedures (SAS, 1987). 



Sample size and power analysis 



We evaluated video and longline data in a power 

 analysis for the £-test of means. Specifically, we esti- 

 mated the sample sizes required to detect a twofold 

 change in abundance by using either sampling 

 method. Skalski and McKenzie (1982) set a prece- 

 dent for use of the criterion of twofold change in en- 

 vironmental monitoring studies; annual variations 

 much larger than this are typical for marine fishes 

 (Hennemuth et al., 1980; Francis, 1993). The effect 

 size (ES) was calculated as follows: 



ES = 



0.693 

 SD 



(7) 



where 0.693= I ± twofold difference in x I for the natu- 

 ral log ( x ) and SD is the standard deviation. Cohen 

 (Tables 2.3.4 and 2.4.1, 1988) was consulted for the 

 requisite sample sizes. The ES for each gear was 

 evaluated at 13=0.20, power (1-J3)=0.8, and a 2 =0.05. 

 For the 1993 data, ES was also evaluated at a 2 =0.1. 



Results and discussion 



Sample composition 



The mean time to first appearance (TFAP) of 

 opakapaka for 1992 video tapes with opakapaka 

 present (all islands included) was 203 ±165 (SD) sec- 

 onds. The total time (TOTTM) of opakapaka during 

 a deployment averaged 122 ± 133 seconds. The maxi- 

 mum number (MAXNO) of opakapaka appeared on 

 tape at approximately 354 (±153) s, based on the nine 

 video sequences for which the weighted average 

 MAXNO ( X w ) was calculated. These data confirm 

 our initial choice of a 10-min bottom time. 



In 1992, only windward Oahu data were used for 

 comparisons and statistical analyses, because the 

 opakapaka measures from Maui and Kauai included 

 large percentages (92% and 67%) of "double-zeros" 

 (zero longline catch, zero fish recorded). Catches of 

 P. filamentosus also were greatest for the windward 

 Oahu site; 54 of the 58 juvenile opakapaka were 

 longlined off windward Oahu. Puffers were preva- 

 lent at windward Oahu and at Maui. Both longlined 

 and video-recorded opakapaka were juvenile size (13 

 to 21 cm FL, and 15 to 25 cm FL, respectively; 

 Kikkawa, 1984; Moffitt and Parrish 4 ). 



Frequency of occurrence data and total number of 

 species differed between longline catches and video 

 records (Fig. 3). Puffers ranked first in abundance 

 and opakapaka second in both the longline and video 

 data. Video cameras recorded the presence of 

 opakapaka and puffers more often than did the 

 longlines (Fig. 3). Video tapes also recorded a greater 

 diversity of species (Table 1), suggesting greater ac- 

 curacy of the video system. Fish that were not caught 

 by the longline but were seen on video included reef- 

 associated species (e.g. pennant butterflyfish, 

 Heniochus diphreutes, and whitesaddle goatfish, 

 Parupeneus porphyreus, sharks {Carcharhinus sp.), 

 and rays (Dasyatis sp.). Longlines also undersampled 

 the lizardfish, Trachinocephalus myops (Fig. 3), a 

 major component of this deep-water, soft-bottom fish 

 assemblage. 5 No major differences in species compo- 

 sition occurred in video surveys from 1992 and 1993. 



1 992 video-longline relations 



The MAXNO index for opakapaka and puffers was 

 highly correlated with the total duration on film 



