102 
Fishery Bulletin 110(1) 
Table 1 
An overview of sampling effort (CPUE) and loggerhead sea turtle ( Caretta caretta ) catch (no. of loggerheads) in the Charleston, 
South Carolina, shipping channel between 2004 and 2007. Fate of turtles relative to original capture numbers is indicated in 
parentheses as follows: five within-year recaptures denoted by a W, three between-year recaptures denoted by a B, one logger- 
head recaptured elsewhere denoted by RE, two loggerheads tagged elsewhere denoted by TE, and one loggerhead stranded near 
Charleston in a subsequent year denoted by an S. CI=confidence interval. 
No. of 
Year 
Start 
End 
events 
“A” 
“B” 
“D” 
“E” 
C. caretta 
Mean CPUE 
95% Cl 
2004 
05/11 
05/19 
48 
15 
13 
14 
6 
49 (1W, 1TE, IS) 
1.55 
0.66 
2005 
05/09 
05/20 
70 
30 
20 
20 
36 (1TE) 
0.54 
0.13 
2006 
05/15 
05/26 
69 
29 
20 
20 
43 (2W, IB, IRE) 
0.63 
0.12 
2007 
05/21 
05/22 
16 
0 
1 
15 
7 
0.40 
0.10 
2004 
06/14 
06/25 
71 
31 
20 
20 
55 (2W, IB) 
0.74 
0.18 
2004 
08/23 
09/01 
43 
14 
15 
14 
16 (IB) 
0.36 
0.10 
2005 
08/08 
08/19 
92 
39 
26 
27 
11 
0.13 
0.03 
2007 
07/31 
08/01 
23 
0 
8 
15 
7 
0.33 
0.08 
Total 
432 
158 
123 
145 
6 
224 
Island, GA, in June 2008 and an 81.2-cm SCLnt log- 
gerhead (sex not determined) stranded approximately 
25 km north of the channel in May 2005, 372 days after 
being tagged and released. 
Forty percent (158 events) of trawling events dur- 
ing 2004-06 were completed in the “A” block (which 
included three stations) compared to 29% of sampling 
Table 2 
The importance of model terms (ordered by P-value) on log- 
gerhead sea turtle ( Caretta caretta ) catch in the Charles- 
ton, South Carolina, shipping channel, 2004-07. Seven 
variables and three interaction terms (see Materials and 
methods section for description) were removed from the 
final model. The Akaike information criterion (AIC) score 
and the percentage of model variance accounted for are 
also included. 
Model terms retained 
P-value 
Sampling period 
<0.001 
Sampling block 
<0.001 
Barometric pressure (mb) 
0.004 
Miscellaneous crabs (count) 
0.009 
Vessel speed (kn) 
0.020 
Horseshoe crabs (count) 
0.056 
Wind direction (degrees) 
0.077 
Cloud cover (%) 
0.088 
Tide stage (ebb, flood) 
0.126 
Miscellaneous jellyfish (count) 
0.135 
Surface temperature (°C) 
0.171 
AIC score 
675.2 
Null model deviance 
523.8 
Final model deviance 
288.4 
% of model deviance explained 
44.9 
effort in the “B” (114 events) and “D” (115 events) blocks 
which had two stations apiece; however, these differ- 
ences were not statistically significant among years 
(j 2 = 0.785, df=4, P=0.940). Trawling in 2007 was con- 
ducted only in the “D” (93% in May, 65% in August) 
and “B” (all others) blocks. 
Significant influences on loggerhead catch included 
sampling period, sampling block, barometric pressure 
(mean ±95%CI=1015.6 ±0.5 mb), vessel towing speed 
(2.8 ±0.02 kn), and miscellaneous crabs (586 speci- 
mens); 17 variables and factors were deemed nonsig- 
nificant or were dropped from the final model (Table 
2). High adjusted loggerhead sea turtle catch in the 
“D” (and to a lesser extent the “B”) block in May 2004 
(mean ±95%CI=1.55 ±0.66 turtles per 30.5 m net-hour) 
and June 2004 (0.74 ±0.18 turtles per 30.5 m net-hour) 
contributed greatly to significant results (Fig. 2). Baro- 
metric pressure in May 2004 (median=1026 mb) was 
significantly greater (77=296.2, df=7, PcO.001) than all 
other sampling periods except May 2007 (median=1020 
mb); however, miscellaneous crab counts (Fig. 3) in 
May 2004 (12.0 ±87.8 crabs/event) were not statistically 
different from other sampling periods (<1.8 ±28 crabs/ 
event). The GLM accounted for 45% of the model devi- 
ance in adjusted loggerhead catch. 
The GLM (AIC = 872.4) explained 17% of model devi- 
ance in adjusted loggerhead catch between the 1991- 
92 and 2004-07 study periods and both model terms 
(month and year) were significant (P<0.001). Greatest 
catch rates occurred in May (Fig. 4); however, catch 
rates in May 1991 and 1992 represented just one log- 
gerhead in 12 trawling events and five loggerheads in 
27 trawling events, respectively. Confidence intervals 
around mean adjusted loggerhead catch did not overlap 
between study periods for any month, and the greatest 
catch rate during 1991-92 only exceeded that of August 
2005 in the present study (Fig. 4). 
