Moffirt and Parrish: Assessment of exploitable biomass of Heterocarpus laevigatas in Hawaiian Is , Part 2 



477 



trapping technique and site selection. The mean of 

 observed densities recorded during submersible dives 

 is regressed against yields of a trap set at the dive sites 

 to obtain an estimate of q, and this value is compared 

 with that reported by Ralston and Tagami (1992). An 

 accurate estimate of catchability is important in order 

 to better estimate exploitable biomass for management 

 purposes. 



Methods 



A total of five submersible dives were conducted in the 

 main Hawaiian Islands, at two sites off leeward Oahu 

 in February 1988 and three sites off the Kona coast 

 of the Island of Hawaii in August 1988 (Table 1). The 

 Oahu sites were selected for their proximity to port and 

 because of previously-observed concentrations in the 

 area of unidentified red shrimp at appropriate depths 

 (400-900 m) for Heterocarpus. The three Kona sites 

 were selected as extremes in H. laetngatus yield for the 

 area during a trapping survey using pyramid traps con- 

 ducted in March 1988 (see Tagami and Barrows (1988) 

 and Ralston and Tagami (1992) for trap description and 

 trapping methods). Catches of H. laevigatus were 

 lowest for the Kona area (< 1 kg/trap-night) at one of 

 the sites and highest (> 10 kg/trap-night) at the remain- 

 ing two sites. 



Visual censuses 



All dives used the Pisces V, a three-man submersible 

 that allowed simultaneous observations by two re- 

 searchers through separate view ports with non- 

 overlapping fields of view directed diagonally forward 

 and down. A video camera continuously recorded the 

 bottom throughout each dive as well. The same two 

 researchers estimated shrimp abundance on all dives 

 and independently reviewed the dive video tapes as a 

 check on observer bias. On each survey, the submers- 

 ible descended to depths of 480-920 m, then traveled 

 to an arbitrary starting location at ~600-750m depth. 

 At this point, a baited container was placed on the 

 bottom and observations of shrimp behavior in the 

 presence of bait were recorded. After observing shrimp 

 behavior for ~15min, the submersible traveled a hap- 

 hazard, rectangular track at a speed of ~2 knots along 

 the contours within the zone of maximum shrimp abun- 

 dance (defined below), returning to the baited container 

 for retrieval at the end of the dive. At preselected time- 

 intervals (5 or 10 min), the submersible settled to the 

 bottom and counts of shrimp were taken by each 

 observer in an independent quadrant filling the field 

 of view. The estimated area of each quadrant was 

 10 m-, which was calibrated by underwater observa- 



tion of known dimensions with the submersible at- 

 tached to its launch-and-retrieval vehicle. The minimum 

 distance between observation sites was ~100m. The 

 number of observations of shrimp density varied be- 

 tween dives, for bottom time was dependent on bat- 

 tery power. 



Bottom depth, temperature, and substrate type were 

 recorded with each shrimp count. The substrate within 

 each quadrant was categorized by composition and par- 

 ticle size of the major component. Substrate composi- 

 tion included coralline, volcanic, and mixed; particle 

 size included sand, rock both small (~<15cm diameter) 

 and large (>15cm diameter), and pavement. 



A x^ goodness-of-fit test using a Poisson distribution 

 for the expected frequencies was conducted to deter- 

 mine whether H. laeingatus were concentrated or even- 

 ly distributed over the bottom at each dive site. Mean 

 H. laevigatus density and 95% CI was calculated for 

 each dive site based on a Poisson distribution, 



CL = D,i, + (1.96) 



f(D(i, 



where D(i) and n(,) are the mean density and number 

 of observations for each dive site. Expected density 

 values (De) for each site were calculated using trap 

 landings for the site and the normalized catchability 

 coefficient (q) reported by Ralston and Tagami (1992), 

 using the following formula: 



De(i, = 



CPUE,i) 



Confidence limits for the expected density values for 

 each site could not be calculated, since variance can- 

 not be computed for CPUE, based as it is on the catch 

 of a single trap. 



An analysis of variance (ANOVA) was performed to 

 determine whether higher mean densities of H. laevi- 



