Dapp et al.: Immediate mortality of Prionace glauca and Galeocerdo cuvier caught by pelagic longlines 
37 
has since been reduced or removed by the shift to circle 
hooks (Campana et al., 2009). Despite the decreases in 
rates of immediate mortality noted in this study, catch 
rates for species of sharks may increase when circle 
hooks are used and this increase may have contributed 
to a greater number of animals being captured (see 
meta-analysis by Gilman et al., 2016). 
The reported effect of SST on immediate mortality 
of blue shark varies among studies. Clarke et al. (2013) 
and Gallagher et al. (2014) found a positive relation 
between the immediate mortality rate of blue sharks 
and SST. In contrast, Campana et al. (2009), Epperly et 
al. (2012), Serafy et al. (2012), and Coelho et al. (2013) 
reported that temperature had no effect on the rate 
of immediate mortality rate for blue sharks. Although 
we identified that the influence of SST on immediate 
mortality as statistically significant in blue sharks, im- 
mediate mortality rates were lower by only 0.5% at an 
SST of 30°C than at an SST of 10°C, indicating that 
other factors exert more influence on mortality of blue 
sharks. Statistical significance is more likely to be de- 
tected in studies with a very large sample size, as in 
our study, even when the effect that sample size has 
on mortality is minor (for more information on how 
sample sizes can influence P-values, see Sullivan and 
Feinn, 2012). Sea-surface temperature had no effect on 
immediate mortality of tiger shark (Fig. 3) — a finding 
that is consistent with the results of Gallagher et al. 
(2014). Although SST has been reported to have little 
influence on the immediate mortality rates of species 
we examined, higher SSTs are thought to increase 
physiological stress during fishery capture (Cicia et al., 
2012; Hoffmayer et al., 2012; Guida et al., 2016) and 
the effect of SST on mortality is likely to be species 
specific. 
Blue sharks caught on longline sets that targeted 
swordfish were more likely to be discarded dead than 
those caught on longline sets that targeted tunas or 
sharks (Fig. 2). However, the majority (83%) of blue 
shark caught on longline sets that targeted swordfish 
were caught in the NED zone, and blue sharks caught 
in the NED zone had higher mean rates of immediate 
mortality than those caught in other geographic zones 
(Fig. 2). It is unknown why blue shark caught in the 
NED zone had higher immediate mortality rates than 
those caught in other geographic zones. Tiger sharks 
exhibited the opposite trend; individuals caught on 
longline sets that targeted tunas were more likely to 
be discarded dead than those caught on longline sets 
that targeted swordfish. Several gear and operational 
strategies differ between longline sets that target tu- 
nas and those that target swordfish. For example, long- 
line sets that target swordfish are typically left fishing 
nearer the water’s surface and overnight, but longline 
sets that target tunas are typically set deeper and left 
from sunrise to sunset (NMFS^). Despite the identifica- 
tion of some operational differences, the exact causes 
of increased immediate mortality rates of tiger sharks 
during capture on longline sets that target tunas com- 
pared with mortality rates during sets that target 
swordfish remain unclear and will require additional 
investigation. 
Deviance in the data explained by our models was 
low (1.6% for blue sharks and 4.5% for tiger sharks), 
indicating that important factors that contributed to 
immediate mortality may not have been recorded in the 
USPL data set or included in our analysis. Factors that 
potentially contributed to immediate mortality during 
capture and that were not investigated include capture 
duration, body size, sex, differences between SST and 
air temperature, capture depth, degree of movement re- 
striction, behavior during capture, hook size, individual 
crew handling practices, and bait used (Campana et al., 
2009; Cicia et al., 2012; Serafy et al., 2012; Dapp et 
al., 2016a). These factors were not considered in our 
analysis because they were not recorded in the USPL. 
Despite the potential effect of unknown factors, models 
for immediate mortality of sharks typically explain a 
low percentage of the deviance in the data (e.g., 7% 
in Campana et al., 2009; 2-5% in Coelho et al., 2012; 
3-6% in Clarke et al., 2013), and more research is nec- 
essary to evaluate the variety and extent of factors af- 
fecting immediate mortality of sharks. 
One caveat concerning our analysis is that tempo- 
ral differences in immediate mortality were attributed 
to regulatory measures, but explicit recordings of hook 
type and number of animals finned were not avail- 
able. We assumed that before August 2004 longline 
fisheries had used J-hooks because J-hooks were the 
predominant hook type used by the U.S. Atlantic pe- 
lagic longline fishery during this time period (Hoey and 
Moore^; Watson et al., 2005). Before the requirement to 
use circle hooks was established in August 2004, circle 
hooks were not used with some vessels, but because 
hook type was not recorded within the USPL, we could 
not identify the vessels with which they were used. 
Despite this limitation, time period has been previ- 
ously used as a proxy for hook type in the U.S. Atlantic 
pelagic longline fishery (Serafy et al., 2012). Addition- 
ally, we postulated that the number of dead discards 
of tiger sharks decreased after March 1993 because of 
the establishment of fin-to-carcass ratios that reduced 
the economic viability of retaining shark fins. Because 
the number of sharks that had been finned was not 
explicitly recorded, it is possible that other unknown 
variables could have decreased immediate mortality of 
tiger sharks after this time period. Despite this caveat, 
there were clear trends of decreasing immediate mor- 
tality after regulatory measures, and these manage- 
ment initiatives are likely to have positively affected 
the survival of blue sharks and tiger sharks. 
One statistical limitation in our study was that tar- 
get catch was examined by using a univariate model 
because of the need to use a reduced data set to exam- 
ine this factor. A univariate approach ignores the effect 
of other potentially important factors and such effects 
should be considered when interpreting our results. 
Another caveat concerning our study is that the 
USPL data set used is reported by fishermen. Inac- 
curacies may arise in the USPL data set as a result 
