94 



Fishery Bulletin 89(1), 1991 



using photographic and video records made during each 

 dive. 



Occupancy of "active" burrows was estimated by 

 observing tilefish diving into burrows and by observ- 

 ing "smoking" (sediment plumes extruded) burrows 

 during each of three other dives. These "positive" 

 sightings provided a minimum estimate of occupancy 

 of the total number of active-looking burrows seen on 

 each dive. Occupancy dives were not conducted in the 

 yellowedge grouper site, because grouper were usual- 

 ly seen outside their burrows and only moved out of 

 sight into burrows when the submersible approached. 



Bottom longline fishing Intensive fishing activities 

 were conducted with bottom longlines in a portion of 

 each study area (Figs. 1, 2). In the tilefish area longline 

 sets consisted of 100 No. 7 circle hooks on 4.6-cm 

 gangions, attached to a 366-m long ground line with 

 halibut line snaps. Weights were used at both ends of 

 the longline to prevent movement of the line along the 

 bottom. Longlines were baited with squid and fished 

 during daylight only. Two lines were fished on a 

 rotating basis, with sets being soaked for approximate- 

 ly 2 hours. A maximum of eight sets (800 hooks) were 

 made per day. In the yellowedge grouper area "Kali" 

 poles were used to reduce gear loss caused by hanging 

 on large boulders. This gear consisted of 40 2.4-m long 

 PVC pipes weighted at one end and buoyed at the 

 other, with 5 hooks 20.3cm apart on each pole 

 separated vertically by about 0.5m. One pole was clip- 

 ped to a floating mainline every 9 m with a halibut line 

 snap. The "Kali" lines had only the lower end of the 

 PVC pipe and anchors touching bottom. The lines were 

 set at randomly selected locations for approximately 

 2 hours, using squid for bait. 



Upon retrieval of all longlines, number of hooks 

 returned, number of baits returned, and catch by 

 species were recorded. Catch-per-unit-effort (CPUE) 

 was calculated for each set using number of hooks 

 returned. 



Data analysis 



Population densities and sizes were estimated using the 

 Leslie method modified by Braaten (Ricker 1975, p. 

 151) from longline catches of Cuban dogfish Squalus 

 cubensis, gulf hake Urophycis cirrata, southern hake 

 U. floridana, and tilefish in the longline-fished portion 

 of the tilefish area; and barrelfish Hyperoglyphe per- 

 ciformis, longspine scorpionfish Pontinus longispinis, 

 southern hake, and yellowedge grouper in the longline- 

 fished portion of the grouper area. A regression of 

 CPUE (in number of fish/ 100 hook-hours) for each 

 longline on adjusted (50% of each day's total catch) 

 cumulative daily catch was calculated using least 



squares regression weighted by the inverse of variance 

 in daily CPUE (SAS 1982). The cumulative catch was 

 adjusted to reduce bias that can result from using the 

 cumulative catch at the start or end of each fishing 

 period (Braaten 1969). The X-axis intercept was the 

 population estimate (N) for the area fished. Associated 

 95% confidence intervals were calculated following 

 Sokal and Rohlf (1981, p. 498). Catchability coefficients 

 (for species caught on longlines with significant regres- 

 sions) were equal to the slopes of the regressions 

 (Ricker 1975, p. 150). The assumptions of this technique 

 include constant catchability, geographically closed 

 (within the study area) population (i.e. no recruitment 

 or emigration), no natural mortality, and constant 

 fishing effort (Ricker 1975). 



Data from the submersible were used to estimate 

 tilefish and yellowedge grouper populations (N) within 

 the area fished with longlines (fished) and the remain- 

 ing portion of the study area (unfished). There were 

 two transects made in the fished portion, and three 

 transects in the unfished portion of the tilefish study 

 area. There was one transect made completely in the 

 fished portion, and four transects in the unfished por- 

 tion of the yellowedge grouper study area. These 

 populations were estimated as the product of the mean 

 number of burrows or fish per km 2 , the percent bur- 

 row occupancy (for tilefish only), and the total km 2 in 

 the study area. The mean number of burrows or fish 

 per km 2 (± 1 SE) was estimated for each transect and 

 each area using the mean and variance equations for 

 the delta-distribution (Pennington 1983, 1986) after 

 transforming the density data along each leg of each 

 transect to natural logarithm. Differences in mean den- 

 sities were tested using the t-test (Sokal and Rohlf 

 1981). Occupancy was estimated as the mean percent 

 (± 1 SE) from three dives using the "smoking" bur- 

 row data discussed previously using the ratio estimator 

 (Cochran 1977). The variance of the population esti- 

 mates (N) based on burrow counts was calculated as 

 the variance of a product (burrows/km 2 x percent oc- 

 cupancy; Goodman 1960). Differences in the population 

 estimates resulting from the longlines and submersibles 

 were tested using the f-test (Cochran 1977, Sokal and 

 Rohlf 1981) and variances associated with the popula- 

 tion estimates. 



Results 



Fish data from longlines and submersibles provided 

 significantly different estimates of tilefish populations. 

 Population estimates for yellowedge grouper could not 

 be made from the longline data because it did not yield 

 a significant regression (Table 1). The number of tilefish 

 (with 95% confidence intervals in parentheses) in the 



