BUTLER ET AL.: ESTIMATING LOGGERHEAD TURTLES BY TRAWL SURVEY 



1) St. Mary's entrance to King's Bay (lat. 

 30°43'N, long. 80°20'W) is divided by the state 

 boundary between Georgia and Florida and in- 

 cludes Cumberland Sound through which the In- 

 tracoastal Waterway connects King's Bay with 

 the entrance channel. Mud predominates inside 

 of the jetties, and mud and rock bottom are found 

 in the channel offshore. 



2) Ponce de Leon Inlet (lat. 29°04'N, long. 

 80°53'W), on the northeast coast of Florida, is a 

 small inlet accessible only to small craft. A jetty 

 protects the inlet to the north; inside the inlet a 

 narrow channel leads to the Intracoastal Water- 

 way. The substrate is hard sand and silt with 

 scattered rubble. 



3) The Port Canaveral ship channel is located 

 on the central east coast of Florida (lat. 28°23'N, 

 long. 80°33'W). The ship channel allows naviga- 

 tion from offshore, through a manmade inlet, into 

 a protected harbor. A depth of 11 to 13 m is main- 

 tained by dredging. Soft mud and detritus bottom 

 is found in the channel and sand-clay in the sur- 

 rounding areas. 



4) Fort Pierce Inlet (lat. 27°28'N, long. 

 80°16'W) is located on the south-central east 

 coast of Florida. The channel allows navigation 

 from offshore, through the inlet that is protected 

 by jetties, into the Intracoastal Waterway. The 

 bottom is hard sand and rubble. 



5) St. Lucie Inlet, also on the south-central 

 east coast of Florida (lat. 27°09'N, long. 80°07'W), 

 is another small inlet with use limited to small 

 craft. A completed jetty protects the north side 

 of the inlet and a second jetty was under con- 

 struction to the south during the survey periods. 

 The substrate offshore is sloping hard sand and 

 silt. 



MATERIALS AND METHODS 



Quarterly trawl surveys of the navigational 

 channels were conducted from November 1981 

 through August 1982. During each survey, the 

 Port Canaveral ship channel was sampled twice 

 and the remaining four sites (St. Mary's entrance. 

 Ponce de Leon Inlet, Fort Pierce Inlet, and St. 

 Lucie Inlet) were sampled once. A standard 18 m 

 "mongoose" fish trawl, spread by 3 m x 1 m trawl 

 doors and equipped with mudrollers, was used 

 throughout the study period. 



Prior to the surveys, the boundaries of each 

 channel were located using National Ocean Sur- 

 veys charts and subdivided by a grid pattern for 



systematic sampling. Lengthwise, each channel 

 was separated into 1,483 m stations which 

 were divided into 30 m wide substations (Fig. 2). 

 The number of substations in each station was 

 dependent on channel width. 



A systematic sampling scheme was devised to 

 sample each channel substation: every other sta- 

 tion was sampled in leapfrog fashion in one direc- 

 tion, and then the direction was reversed. The 

 substation sampled within each station was de- 

 termined by random drawing without replace- 

 ment and sampling continued until all substa- 

 tions were occupied. This approach avoided the 

 "edge effect", but allowed samples to be statisti- 

 cally treated as random (Milne 1959). Control sta- 

 tions outside the channel were sampled at all 

 sites during each survey period. 



In addition to standard survey procedures, ex- 

 periments designed to estimate gear efficiency 

 were conducted in the Port Canaveral ship chan- 

 nel. Following each survey, a substation with 

 abundant loggerhead turtles was selected and a 

 series of repetitive tows performed. All logger- 

 heads captured during these experiments were 

 tagged and released on station prior to the next 

 tow. As this was essentially a "removal" method, 

 any recaptures of loggerhead turtles tagged dur- 

 ing the experiment were not considered as part of 

 the catch and were excluded from analysis. Tows 

 were continued in rapid order until two consecu- 

 tive samples yielded zero catches or the working 

 day ended. 



ANALYTICAL PROCEDURES 



The efficiency of the sampling gear was estab- 

 lished before population estimates were com- 

 puted. The probability of loggerhead turtle cap- 

 ture iP) was estimated for each repetitive towing 

 experiment using the formula: 



P = C^/No 



where Ci - catch on the first tow in the substa- 

 tion 

 Nq = estimated number of loggerhead tur- 

 tles in the substation. 



A regression of cumulative loggerhead turtle 

 catch (Y) on catch per sample (X), expressed as 

 y = 6o + biX, was used to estimate (A^o* based on 

 the relationship: Nq = bo. The estimated variance 

 of A^o was calculated according to procedures of 

 Kleinbaum and Kupper (1978): 



449 



