334 
Fishery Bulletin 118(4) 
South Carolina 
Bulls Bay 
North Edisto River, s 
Saint Helena Sound 
Port Royal ° 
Sound 
Alabama Georgia 
Atlantic Ocean 
. oaint Andrew Sound N 
° Saint Vincent Sound A 
Waccasassa Bay@® 
Anclote Anchorage Florida 
Gulf of Mexico 
Tampa Bay 
No. of bonnetheads 
recaptured 
e 1-5 
@ 6-10 
@ 11-25 
@ 26-50 
@ 51: 
Pine Island Sound 
Florida Baye 
82°W 
Figure 1 
Map showing the locations where bonnetheads (Sphyrna tiburo) were recap- 
tured in the northeastern Gulf of Mexico between 1993 and 2006 and in 
the estuarine waters of the Atlantic coast of the southeastern United States 
between 1998 and 2019. Circle size indicates the number of bonnetheads 
recaptured at a location. Data from these tag-recapture efforts were used in 
an age-independent model, GROTAG, to estimate growth rates of bonnet- 
heads by region. 
age-based models are 300 and 760 mm FL (males) and 340 
and 960 mm FL (females) (Lombardi-Carlson et al., 2003). 
Effects of tagging on growth 
A total of 22 recaptured individuals were aged by Frazier 
et al. (2014), and resulting data were available for analysis 
of the effect of tag type on growth (8 sharks were tagged 
with a nylon dart tag, and 14 sharks were tagged with a 
rototag). The changes in residuals from mean age at ini- 
tial capture (i.e., tagging) and estimated age at recapture 
were plotted against time at liberty (Fig. 2). The slope (6) 
of the line is not significantly different from zero for tag 
types combined or for individual tag types, and 95% CIs for 
Gullivan Baye Chokoloskee Bay 
slopes bound zero (all tags: b=0.005 [95% 
CI —0.033—0.043], P=0.615, df=20; nylon 
dart tag: b=0.037 [95% CI —0.109-0.184], 
P=0.554, df=6; rototag: b=0.005 [95% 
CI -0.042-0.052], P=0.805, df=12). There- 
fore, there is no evidence to indicate that 
tagging or tag type affected growth, and 
growth increments from tag-recapture 
data were considered suitable for model- 
ing growth in a population. 
Models based on tag-recapture data 
The GROTAG model (Francis, 1988b) 
produced biologically reasonable param- 
eter estimates for males from both 
regions (Table 3); however, 95% CIs are 
large, indicating that the sample size 
was insufficient to produce robust esti- 
mates of growth. The best-fit model for 
males in the GOM (model 3) included 
parameters for mean growth rates at 
reference lengths (g479, 8737), mean and 
SD of measurement error, and growth 
variability; the model failed to con- 
verge when the model 1 configuration 
was used (Table 4). The final model for 
males in the Atlantic region (model 2) 
included parameters for mean growth 
rates at reference lengths (8619, 765), 
SD of measurement error, and growth 
variability; model 3 failed to fit the data 
for this region (Table 4). Estimates of 
growth variability and seasonal growth 
for males are uninformative because the 
low sample size produced large 95% CIs 
for estimates, with upper and lower lim- 
its of 95% CIs above parameter bounds 
in the GOM model. Although confidence 
intervals are large and overlap, the SD of 
measurement error of the GOM model is 
3 times that of the Atlantic region model. 
Region-specific models for female bon- 
netheads converged for all models run. 
The final model for females in the GOM 
(model 2) included parameters for mean growth rates at 
reference lengths (g4¢5, 8915), 0D of measurement error, and 
growth variability (Table 4). The final model for females in 
the Atlantic region (model 4) is more complex, with the 
parameters for mean growth rates at reference lengths 
(555, £1000)) mean and SD of measurement error, growth 
variability, and seasonal variation included (Table 4). The 
estimates of growth variability for females are large from 
both the GOM model (v=0.63) and the Atlantic region 
model (v=0.56), indicating that individuals in the pop- 
ulation could be expected to grow 0.37—1.63 (GOM) or 
0.44-1.56 (Atlantic region) times the estimated average 
growth rate per length class. The model for females in the 
Atlantic region has a strong seasonal growth component 
