Whitney et al.: Mortality of Carcharhinus limbatus caught in the Florida recreational fishery 
535 
Figure 1 
Photograph of a tag float package, which includes an acceleration data logger (ADL) and a 
very high frequency (VHF) transmitter with a 1-d galvanic release, attached to the dorsal 
fin of blacktip shark (Carcharhinus limbatus) S16, which was 112 cm in precaudal length 
and was 1 of 31 blacktip sharks caught and released off Florida between September 2011 
and April 2013. The ADL was embedded on the opposite side of the tag float abutting the 
fin (not visible). The top of the float is painted orange for identification and recovery of 
the tag package at sea (Whitmore et al., 2016). 
Therefore, we also divided data into descent, ascent, 
and level phases before statistical analysis (Whitney 
et al., 2016). 
To determine possible recovery period, we took hour¬ 
ly means of each metric, and built asymptotic nonlin¬ 
ear mixed models using the nlme package in the open- 
source statistical software R, vers. 3.1.0 (R Core Team, 
2014). Recovery period was defined as the amount of 
postrelease time it took for the metric value to gain 
80% of the difference between the initial postrelease 
value and the fully recovered value, defined as the up¬ 
per asymptote in the logistic equation (Whitney et al., 
2016). Metrics shown to display a recovery period were 
then calculated for each individual (for a more detailed 
description of these analyses, see Whitney et al., 2016). 
Statistical analysis 
All statistical analyses were conducted in R, and all 
results were reported as means with SDs unless oth¬ 
erwise stated. 
At-vessel capture metrics Chi-square tests were per¬ 
formed to test the effect of hook type on the location of 
hooking (jaw, mouth, gut), the presence of abrasions or 
bleeding, and the likelihood of the hook being removed 
by the fisherman (as opposed to the line being cut and 
the hook left attached to the shark). 
Generalized linear models (GLMs) were used to de¬ 
termine which at-vessel capture metrics (temperature, 
dissolved oxygen, hook type, hooking location, fight 
time, handling time) affected blood biomarkers (pH, 
pC0 2 , La - ), and an ordinal logistic regression (OLR) 
was used to determine which capture metrics impacted 
the BRCS. A full complement of all possible models 
(with the addition and removal of each term) was con¬ 
structed and compared by using the MuMIn package, 
vers. 1.15.6, in R. The model with the lowest Akaike’s 
information criterion (AIC) was considered the candi¬ 
date model and the significance of each term was de¬ 
termined by using the F-statistic from an analysis of 
variance (ANOVA). 
Postrelease outcome To investigate the ability of at- 
vessel metrics to determine postrelease outcome, the 
dimensionality of behavioral recovery periods was re¬ 
duced by using 2 methods in order to minimize po¬ 
tential type-I error. With the first method, an average 
