Orbesen et al.: Mortality rates of Thunnus thynnus associated with pelagic longline gear in the Gulf of Mexico 
19 
Table 1 
Results of logistic regression examining the influence 
of hook type, target species, sea-surface temperature 
(SST), soak duration, straight fork length (SFL), and 
maximum hook depth on the probability of at-vessel 
mortality of 1498 bluefin tuna (Thunnus thynnus ) cap¬ 
tured in the pelagic longline fishery of the Gulf of Mex¬ 
ico during 1993-2017. 
Variable 
Estimate 
Standard 
error 
P- 
value 
Standard circle hook 
-0.464 
0.129 
<0.001 
J-hook 
-0.602 
0.224 
0.007 
Weak circle hook 
0.000 
0.000 
- 
Mixed target 
0.090 
0.131 
0.493 
Swordfish 
-0.354 
0.473 
0.454 
Yellowfin tuna 
0.000 
0.000 
- 
SST 
-0.021 
0.015 
0.149 
Soak duration 
-0.035 
0.040 
0.382 
SFL 
-0.001 
0.002 
0.374 
Maximum hook depth 
0.0046 
0.010 
0.711 
and they were tagged in the months of February-June 
at SSTs ranging from 21.8°C to 29.7°C. Representing 
fish that successfully survived the fishery interaction, 
29 PSATs remained attached for at least 30 d. Ten 
PSATs failed to reach the 30 d threshold, and 2 ad¬ 
ditional PSATs failed to transmit any data. 
Tagged fish at large for fewer than 30 days 
Ten PSATs began transmitting data less than 30 d af¬ 
ter tagging (1-18 d). Six of these tags were equipped 
with tethers that we identified as negatively buoyant. 
Four tags sunk at a rate greater than the rate of the 
reference tag (0.251 m/s); therefore, the fish tracked 
with those tags were put in the mortality category (Ta¬ 
ble 3). Three tags had sinking rates that were slower 
than the rate of the reference tag, indicating likely tag 
attachment failures and not observed mortalities. The 
remaining 3 tags were equipped with positively buoy¬ 
ant tethers, detached from the fish at a depth <1000 m, 
and floated to the surface, indicating tag attachment 
failures. 
Postrelease mortality 
An upper bound estimate of postrelease mortality was 
obtained by treating all fish with tags that either did 
not report (2 PSATs) or failed to attach (6 PSATs) as 
potential dead fish (12 of 41 fish tagged with PSATs), 
giving a maximum postrelease mortality estimate of 
29% (95% Cl: 18-44%). Assuming that these non-re¬ 
porting tags and tags that failed to remain attached 
were not associated with fish mortalities, and, there¬ 
fore, that the fish tagged with them were removed from 
the sample, we determined that the most likely esti¬ 
mate of postrelease mortality is 12% (95% Cl: 5-27%) 
(i.e., 4 of 33 PSATs associated with mortalities). 
Hook timers 
Twelve tagged fish were captured on leaders that in¬ 
cluded a hook timer, which measures the length of time 
a fish is on the line prior to crew engagement (Table 
2). Tag data indicates an apparent mortality for only 
1 fish captured on a line with a hook timer (8.3 h at¬ 
tached to a longline). Three additional tagged fish were 
associated with either attachment failures (2 PSATs) 
or their tag failed to report (1 PSAT). The remaining 8 
PSATs were deployed on surviving fish with an average 
time on the line of 7.4 h (2.2-14.4 h). 
Overall mortality 
We estimated the probability of a mortality of a bluefin 
tuna occurring as a result of an interaction with, and 
release from, PLL gear in the GOM, using Equation 
1 with the most likely estimate of postrelease mortal¬ 
ity (12%, 95% Cl: 5-27%) as P(R) and P(C) obtained 
from the logistic regression model predicted for fishing 
with weak hooks (54%, 95% Cl: 46-62%). The resulting 
overall estimate of the probability of capture-induced 
mortality of bluefin tuna in the GOM PLL fishery, op¬ 
erating as it currently does with weak hooks, is 59% 
(95% CL 47-71%). 
Discussion 
On the basis of the data presented in this study, we es¬ 
timated postrelease mortality of bluefin tuna from PLL 
fishery operations in the GOM at a range of 12-29%, 
depending on the treatment of the non-reporting or pre¬ 
mature release of tags. The highest estimates of mor¬ 
tality were obtained when all non-reporting tags were 
assumed to be associated to a mortality event. Howev¬ 
er, we considered that there is a sound basis for elimi¬ 
nating fish with non-reporting tags from the sample. 
Our non-reporting rate for this study is relatively low 
(5%) in comparison to the rates of other studies (Musyl 
et al., 2011b). Furthermore, with an RD-1800 device at¬ 
tached to its tether, a PSAT would detach from a sink¬ 
ing, dead fish before it reached the tag crush depth, 
making it unlikely that non-reporting was a result of 
such a mortality. Non-reporting, therefore, was the re¬ 
sult of either equipment failure or damage (perhaps, 
due to predation) (Musyl et al., 2011b), with neither 
cause being informative on mortality due to a capture 
event. Given that we can separate attachment failures 
from mortalities by examining sinking rates, we re¬ 
moved fish associated with both non-reporting tags and 
attachment failures from the sample to provide what 
we believe is a more accurate postrelease mortality es¬ 
timate of 12%. Under either assumption regarding the 
fate of malfunctioning tags, the postrelease mortality 
estimates are relatively low (12-29%), indicating that, 
