536 
Fishery Bulletin 115(4) 
time to recovery was determined for each individual 
by averaging the recovery periods across all swimming 
metrics. GLMs were used to determine which at-vessel 
metrics (blood biomarkers, environmental and capture 
metrics) predicted average time to recovery. As with 
blood biomarker models, a full complement of models 
was constructed and compared by using AIC, with the 
significance of the terms in the candidate model deter¬ 
mined by an ANOVA. 
However, reducing the dimensionality of the recov¬ 
ery period into 1 average negates the possibility that 
there are multiple ways in which an individual can 
recover, and these could be affected by different cap¬ 
ture parameters. For the second method a principal 
components analysis (PCA) was conducted on the time 
to recovery for all swimming metrics. Using the envfit 
function in the vegan package, vers. 2.3-5, in R, we 
overlaid the at-vessel metrics on the recovery period 
ordination from the PCA. 
Results 
Between September 2011 and April 2013, 31 blacktip 
sharks were caught and tagged with ADLs (Cape Ca¬ 
naveral, n= 2; Charlotte Harbor, n- 29), providing a to¬ 
tal of 838 h of acceleration data. The durations of indi¬ 
vidual ADL records on surviving sharks lasted between 
7.1 and 71.7 h (mean: 30 h [SD 22]). Precaudal length 
of the tagged sharks ranged from 92 to 132 cm (mean: 
107.5 cm [SD 11.2]), and girth ranged from 48 to 81 cm 
(mean: 61.2 cm [SD 7.4]). Fight times lasted between 2 
and 16 min (mean: 7 min [SD 3]), and handling times 
lasted between 6 and 18 min (mean: 9.7 min [SD 2.9]) 
(Table 1). These times were largely consistent with 
those practiced by participating recreational captains 
during their typical charters to take photographs and 
remove fishing gear (Moore 4 ; Rapp 5 ). 
At-vessel capture metrics 
Individuals were captured on both circle (n=14) and J- 
hooks (n - 17), and hooking locations were jaw (n=22), 
mouth (n= 4), and gut (n= 3). Hook type did not affect 
where the shark was hooked (% 2 =0 32, df =2, P=0.62), 
how likely the hook was to be removed by the fish¬ 
erman (x 2 =1.01, df=l, P-0.32), severity of abrasions 
()' 2 =0,02, df=l, P=0.89), observed bleeding (x 2 =0.05, 
df=l, /-'=().82), or the BRCS (% 2 =3 34, df=3, P=0.36). 
Additionally, no capture-related variables (hook-type, 
fight time, La - , pC0 2 , and pH) significantly affected 
BRCS (OLR: P>0.08 for all predictors). 
There was interindividual variability in observed 
blood biochemical markers (Table 1), yet pH correlated 
4 Moore, R. 2012. Personal common. Florida Light Tackle 
Charters. 17044 Greenan Ave., Port Charlotte, FL 33948. 
5 Rapp, D. 2012. Personal commun. Sea Leveler Sport 
Fishing Charters. 505 Glen Cheek Dr., Cape Canaveral, FL 
32920. 
with pC0 2 (p= -0.44, t lt 29 = - 2 . 66 , P=0.01) and La - (p= 
-0.51, G ,29 = -3.17, P=0.004); however, La - was not 
correlated with pC0 2 (p= -0.31, f 1>2 9 = -1.76, P-0.09). 
Hook type, hooking location, and dissolved oxygen were 
not found to predict biochemical markers and, there¬ 
fore, were not included in any final predictive models. 
La - increased (coefficient of multiple determination 
[i? 2 ]=0.57) with increasing fight times (Pi i 2 ?/-' : ’i.2.i2, 
P=0.G02) and handling times (Fi,28=25.89, P<G.001) 
(Fig. 2, Table 2), whereas increasing handling time 
was found to significantly lower blood pH (J? 2 =Q.18, 
Pi, 29- 6 44, P=Q.02; Fig. 2, Table 2). Individuals cap¬ 
tured at higher temperatures were found to have in¬ 
creased pC0 2 levels (F 1)2 g=4.96, P=0.034; Fig. 2, Table 
2 ) . The final model for pC0 2 included a negative rela¬ 
tionship with fight time, however this was not signifi¬ 
cant (Px, 28 =0.63, P=0.43). 
Postrelease outcome 
Mortality All sharks swam away after capture and 
handling and only 1 individual (S28; BRCS=4; Table 
1) needed to be extensively revived (2-3 min until it 
swam under its own volition) before release. Three 
of the 31 tagged sharks died after being released as 
indicated from the acceleration and depth data (Fig. 
3) representing a postrelease mortality rate of 9.7%. 
All mortalities occurred within 2 h of release (58, 
76, and 103 min), all succumbing individuals were 
hooked in the jaw, and 2 of the 3 were caught on 
J-hooks. All 3 confirmed mortalities had a BRCS of 
“fair.” Two of the dead sharks appeared to have been 
scavenged after the animals sank to the sea floor and 
had ceased movement for over 30 minutes. The ADLs 
prematurely released during a series of high inten¬ 
sity movement, an indication of scavenging, and one 
of the packages displayed bite marks upon recovery 
(e.g., Lear and Whitney, 2016). Two of the mortalities 
occurred at high temperatures, and these sharks also 
had low blood pH and high La - . However, the third 
mortality occurred for an individual with blood stress 
values similar to sharks that survived (Fig. 4). 
Quantifying sublethal effects Based on the data col¬ 
lected from the ADLs, we determined that 19 of 58 
metrics of swimming behavior showed indications of a 
possible recovery period (for more detail, see Whitney 
et al., 2016, table 1). Overall, blacktip sharks recovered 
10.5 h (SD 3.8) after release. Larger sharks had a sig¬ 
nificantly shorter average recovery time than smaller 
sharks (F ljl4 =7.83, F--0.014, Fig. 5); our model also in¬ 
dicated that increasing pC0 2 decreases time to recov¬ 
ery, however this term was not significant (F 1> i 4 =3.58, 
P=Q.Q79). 
The PCA of recovery period showed that the first 2 
principal components (PC) accounted for 50.7% of the 
variance in the data. PCI was correlated with recovery 
periods determined from average ODBA, TBC, TBAA 
and average W, whereas PC2 was correlated with 
recovery periods determined from maximum ODBA, 
