Satellite telemetry reveals higher fishing mortality rates than previously estimated, suggesting overfishing of an apex marine predator

Abstract

Overfishing is a primary cause of population declines for many shark species of conservation concern. However, means of obtaining information on fishery interactions and mortality, necessary for the development of successful conservation strategies, are often fisheries-dependent and of questionable quality for many species of commercially exploited pelagic sharks. We used satellite telemetry as a fisheries-independent tool to document fisheries interactions, and quantify fishing mortality of the highly migratory shortfin mako shark (Isurus oxyrinchus) in the western North Atlantic Ocean. Forty satellite-tagged shortfin mako sharks tracked over 3 years entered the Exclusive Economic Zones of 19 countries and were harvested in fisheries of five countries, with 30% of tagged sharks harvested. Our tagging-derived estimates of instantaneous fishing mortality rates (F = 0.19–0.56) were 10-fold higher than previous estimates from fisheries dependent data (approx. 0.015–0.024), suggesting data used in stock assessments may considerably underestimate fishing mortality. Additionally, our estimates of F were greater than those associated with maximum sustainable yield, suggesting a state of overfishing. This information has direct application to evaluations of stock status and for effective management of populations, and thus satellite tagging studies have potential to provide more accurate estimates of fishing mortality and survival than traditional fisheries-dependent methodology.

Full text

rspb.royalsocietypublishing.org
Research
Cite this article: Byrne ME, Corte
´s E, Vaudo
JJ, Harvey GCMN, Sampson M, Wetherbee BM,
Shivji M. 2017 Satellite telemetry reveals
higher fishing mortality rates than previously
estimated, suggesting overfishing of an apex
marine predator. Proc. R. Soc. B 284:
20170658.
http://dx.doi.org/10.1098/rspb.2017.0658
Received: 28 March 2017
Accepted: 23 June 2017
Subject Category:
Global change and conservation
Subject Areas:
ecology, environmental science
Keywords:
conservation, fisheries, Isurus oxyrinchus,
mortality, shortfin mako shark, stock
assessment
Authors for correspondence:
Michael E. Byrne
e-mail: byrneme@missouri.edu
Mahmood Shivji
e-mail: mahmood@nova.edu
†
Present address: School of Natural Resources,
University of Missouri, Columbia, MO 65201, USA.
Electronic supplementary material is available
online at https://dx.doi.org/10.6084/m9.
figshare.c.3825508.v1.
Satellite telemetry reveals higher fishing
mortality rates than previously estimated,
suggesting overfishing of an apex marine
predator
Michael E. Byrne1,†, Enric Corte
´s2, Jeremy J. Vaudo1, Guy C. McN. Harvey1,
Mark Sampson3, Bradley M. Wetherbee1,4 and Mahmood Shivji1
1
Guy Harvey Research Institute, Nova Southeastern University, Dania Beach, FL 33004, USA
2
National Marine Fisheries Service, National Oceanic and Atmospheric Administration, Panama City,
FL 32408, USA
3
Fish Finder Adventures, Ocean City, MD 21842, USA
4
Department of Biological Sciences, University of Rhode Island, Kingston, RI 02881, USA
MEB, 0000-0002-9464-9534
Overfishing is a primary cause of population declines for many shark
species of conservation concern. However, means of obtaining information
on fishery interactions and mortality, necessary for the development of suc-
cessful conservation strategies, are often fisheries-dependent and of
questionable quality for many species of commercially exploited pelagic
sharks. We used satellite telemetry as a fisheries-independent tool to docu-
ment fisheries interactions, and quantify fishing mortality of the highly
migratory shortfin mako shark (Isurus oxyrinchus) in the western North
Atlantic Ocean. Forty satellite-tagged shortfin mako sharks tracked over
3 years entered the Exclusive Economic Zones of 19 countries and were
harvested in fisheries of five countries, with 30% of tagged sharks harvested.
Our tagging-derived estimates of instantaneous fishing mortality rates
(F¼0.19–0.56) were 10-fold higher than previous estimates from fisheries-
dependent data (approx. 0.015– 0.024), suggesting data used in stock
assessments may considerably underestimate fishing mortality. Addition-
ally, our estimates of Fwere greater than those associated with maximum
sustainable yield, suggesting a state of overfishing. This information has
direct application to evaluations of stock status and for effective manage-
ment of populations, and thus satellite tagging studies have potential to
provide more accurate estimates of fishing mortality and survival than
traditional fisheries-dependent methodology.
1. Background
Worldwide, populations of many shark species have experienced significant
declines, primarily attributed to increased fisheries exploitation [1]. Although
some shark species are targeted commercially, many are captured incidentally
as bycatch, which is often poorly quantified. A suite of generally k-selected life-
history traits (i.e. slow growth, late age at maturity, low fecundity) typical of
larger-bodied sharks yields slow population growth rates, making rebounding
from even moderate levels of exploitation difficult, and rendering many species
especially vulnerable to over-exploitation [2]. As upper trophic level predators,
sharks can exert considerable top-down functional forces in marine ecosystems,
leading to concerns about the potential impacts their removal may trigger on
the stability of marine communities [3 –6].
Successful fisheries management and conservation are dependent on accu-
rate estimates of population parameters, including survival from total mortality
and the specific effect of fishing mortality. Because of their wide distributions,
propensity to travel long distances and use of remote offshore environments,
&2017 The Author(s) Published by the Royal Society. All rights reserved.
on August 2, 2017http://rspb.royalsocietypublishing.org/Downloaded from

these parameters are particularly difficult to measure for
highly mobile pelagic shark species. Conventional tagging
studies provide a possible means to quantify survival of pela-
gic species via mark–recapture methods. Agency-run
tagging programmes, such as the NOAA NMFS Cooperative
Shark Tagging Program, have deployed thousands of tags
over several decades. Although these programmes have
provided important information on movements and distri-
butions of sharks, only recently have these data been used
to quantitatively estimate survival for the few species with
sufficient tag returns [7,8]. A major limitation of using con-
ventional mark–recapture studies to estimate mortality and
survival is that such programmes rely on voluntary
cooperation of fishers to report captures of tagged animals.
Several studies have demonstrated that substantial numbers
of tagged fishes of various species captured in commercial
fisheries are not reported, and that tags may be shed prior
to recapture, resulting in underestimated mortality of
tagged fish if not properly accounted for [9,10]. Additionally,
it may take many years to accumulate a sufficient sample size
of tag reports to support such analyses, further limiting the
applicability of such studies. Owing to this and other
limitations, stock assessments of pelagic sharks conducted
by organizations such as the International Commission for
the Conservation of Atlantic Tunas (ICCAT) have not
incorporated tag –recapture data, and instead rely on
catch or effort information from the different fleets to
estimate fishing mortality.
In studies of terrestrial wildlife, survival rates and sources
of mortality are often estimated using telemetry [11], where
the fates of individuals within a population are determined
by frequent and regular monitoring over time. The appli-
cation of telemetry-based methods to estimate these critical
population parameters for exploited shark populations
would be advantageous by overcoming many of the limit-
ations associated with fisheries-dependent data collection.
Although acoustic telemetry has been used for estimating
natural and fishing mortality of juvenile coastal sharks
[12–14], the primary use of telemetry in survival studies
of sharks has been for estimating post-release survival
[15–18]. Satellite telemetry is now a standard tool for study-
ing behaviour and ecology of pelagic sharks, with potential
applications for estimates of mortality and survival. Two
types of tags are commonly used in satellite telemetry studies
on sharks, which could be expected to provide useful data for
survival and cause-specific mortality studies: pop-up satellite
archival tags (PSATs) are programmed to release from the
animal after a given time period and relay archived light,
depth and temperature data, whereas satellite-linked radio
tags (SLRTs) communicate with satellite systems and triangu-
late a positional estimate each time the tag is exposed to air.
The SLRT-style tags are often attached to a shark’s dorsal fin
and are particularly effective on species that frequent the sur-
face. Advances in these technologies have resulted in tracking
of individuals for durations exceeding 1 year [19– 21].
Shortfin mako sharks (Isurus oxyrinchus; hereafter, mako
sharks) are among the pelagic shark species considered
most vulnerable to exploitation [22]. As a long-lived species
with low fecundity and late age at maturity, population
recovery times are slow. Although mako sharks are only
occasionally targeted in commercial fisheries, mako shark
habitat use overlaps that of other commercially targeted
taxa such as billfish and tunas, resulting in frequent capture
as bycatch in these fisheries [23]. Mako sharks captured as
bycatch are often retained due to the high market value of
their meat [24], and even when released, mako sharks cap-
tured on pelagic longline gear exhibit low survival [18].
Given their vulnerable life-history traits and concerns about
declining populations, mako sharks are categorized as Vul-
nerable globally by the International Union for the
Conservation of Nature (IUCN) Red List [25], as Critically
Endangered in the Mediterranean Sea [26], and are also
listed under Appendix II (Unfavorable Conservation Status)
of the United Nations Convention on the Conservation of
Migratory Species of Wild Animals (CMS; http://www.
cms.int/en/species/isurus-oxyrinchus). In the North Atlan-
tic, the most recent stock assessment by ICCAT concluded
that the population was not overfished and the probability
of overfishing was low [27]. However, ICCAT also recognized
discrepancies and high model uncertainty resulting from data
deficiencies, and recommended a precautionary management
approach until more reliable stock status estimates could be
obtained. Among the key shortcomings identified in data
availability were reliable estimates of fishing mortality.
Thus robust fisheries-independent estimates of fishing mor-
tality would represent a major contribution towards
obtaining more reliable characterizations of stock status for
mako shark populations.
As part of a separate study of mako shark movement ecol-
ogy in the western North Atlantic Ocean using SLRTs, we
found that in addition to tracking shark movements,
we were able to identify fishing mortality events [28].
Here, we use an expanded dataset of SLRT-tracked mako
sharks to describe fisheries interactions and provide the
first fisheries-independent estimate of fishing mortality in
the western North Atlantic. We compare our fisheries-
independent estimates of fishing mortality to previous esti-
mates for this population derived from stock assessments
based on fisheries-dependent data sources, and discuss
potential future applications of satellite telemetry to study
survival of pelagic shark species.
2. Material and methods
(a) Tagging and movement analysis
We tagged mako sharks with SLRTs (SPOT5; Wildlife Compu-
ters, Redmond WA) during 2013–2015 at two locations: the
Yucata
´n Peninsula in the vicinity of Isla Mujeres, Mexico
(approx. 21.298N, 86.298W), and the northeast coast of the US,
primarily in the vicinity of Ocean City, Maryland (approx.
38.108N, 74.508W). Tagging off the Yucata
´n Peninsula took
place during March and April, and tagging off the US coast pri-
marily took place in May, with the exception of three sharks
tagged in June, August and September. All sharks were captured
by rod and reel, and were either secured along the side of the
boat or brought onboard for processing. When sharks were
brought onboard the vessel, we covered their eyes with a wet
towel to reduce stress, and placed a saltwater hose in the
mouth to irrigate the gills. All sharks were sexed and fork
length (FL) measured. We used FL and sex-specific growth
curves for mako sharks in the North Atlantic [29] to estimate
the age of each shark.
The SLRTs allow communication with the Argos satellite
system when exposed to air, and were attached to the dorsal
fin of each shark. Provided sufficient satellite communication is
achieved, a location estimate of the shark is triangulated and
rspb.royalsocietypublishing.org Proc. R. Soc. B 284: 20170658
2
on August 2, 2017http://rspb.royalsocietypublishing.org/Downloaded from

remotely relayed to the researcher. Mako sharks spend consider-
able amounts of time at the surface [30], and such fin-mounted
tags have proven to be effective tracking tools for this species
[19,21,28]. We released all mako sharks immediately following
processing, and all sharks swam away under their own power.
Mako sharks which failed to report any locations, or which
stopped reporting within 14 days of tagging with no evidence
of harvest, were considered to represent either tag malfunction
or tagging-related mortality and were excluded from analysis.
(b) Management jurisdictions
We used a continuous-time correlated random walk model
(CTCRW) fitted using the package ‘crawl’ [31] in the R statistical
computing environment [32] to account for location measure-
ment error and obtain regular daily location estimates of each
shark from the temporally irregular Argos locations. We inter-
sected the resulting daily location estimates with a shapefile
of global Exclusive Economic Zones (EEZs) in ArcMap (ESRI,
Redlands, CA). This allowed us to identify the internatio-
nal jurisdictional boundaries mako sharks traversed during
the study.
(c) Harvest detection and fisheries interactions
Fishing mortality was identified directly from Argos data using
similar clues used to detect fishing mortality of satellite-tracked
marine turtles [33]; harvests were identified when a tag began
consistently reporting from a static location on land, or when
the transmitter began continuously tracking towards a coastal
port, indicating that the SLRT was onboard a vessel. In several
instances, contact with the fishers allowed us to confirm the fish-
ery (gear type) in which the shark was captured. When it was not
possible to contact the fishers to confirm gear type, we provision-
ally assigned the mortality to the most likely gear type based on
our knowledge of fishery activity in the vicinity of capture, and
the locations on land where the transmitter reported (e.g. com-
mercial fishing port or fish processing centre). We defined a
capture location as the last Argos location received before a tag
began transmitting from land, or when a captured tag reported
from onboard a fishing vessel, as the last location received
before the vessel began directly tracking towards port
(figure 1). Thus, for each fishing mortality event we were able
to ascertain the approximate date and location of capture, the
country of origin of the vessel, and the fishery, or most likely
fishery, in which the animal was harvested.
(d) Survival analysis
We used known-fate models in MARK [34] to estimate annual
harvest-specific survival probabilities, which we denote as S
F
.
Our estimates are harvest-specific because technological limit-
ations of the SLRTs did not allow us to detect natural
mortalities and distinguish them from instances of tag malfunc-
tion, which would be necessary to estimate total survival (S).
Known-fate models are binomial models of survival through dis-
crete time intervals. Tracked individuals known to be alive at the
beginning of a sample interval either survive to the end of the
interval or die during the interval. If contact is lost (e.g. in our
study in the case of transmitter malfunction or undetected natu-
ral mortality) and fate is thus unknown at the end of the interval,
the individual is censored from that interval. The model is for-
mulated as a generalized linear model (GLM), which allows
modelling survival as a function of individual-level covariates
(e.g. size, age, sex). Estimates of survival across time intervals
(for example, annual survival estimated from monthly sample
intervals) can be calculated as the product of the survival likeli-
hoods of each interval. Detailed information regarding model
formulation and implementation can be found in the MARK
software manual available at http://www.phidot.org/soft-
ware/mark/docs/book/.
We estimated S
F
over three-month intervals corresponding to
the four meteorological seasons: spring (March– May), summer
(June–August), autumn (September–November) and winter
(December–February). Because of unbalanced sample sizes
among years, and to ensure adequate sample sizes and mortality
events in all sample intervals, we assumed S
F
was constant
among years, and individuals tracked for more than 1 year re-
entered the analysis in the following year. We constructed four
candidate models representing hypotheses regarding the poten-
tial effects of season and tagging location on S
F
. The candidate
models (i) held S
F
constant across all seasons and tagging
locations, (ii) allowed S
F
to vary across seasons, (iii) allowed S
F
to vary based on tagging location or (iv) allowed S
F
to vary
based on season and tagging locations. We ranked models
using Akaike’s information criterion adjusted for small sample
sizes (AICc), and selected the most parsimonious model based
on DAICc and AICc weights (w
i
) [35]. We used the most parsimo-
nious model to derive an estimate and associated 95% confidence
interval (CI) of annual S
F
. Models were run by calling MARK
through the package ‘RMark’ [36] in R.
(e) Fishing mortality rate
Once S
F
was estimated, the instantaneous fishing mortality rate
(F) was calculated simply as F¼2ln(S
F
). We calculated a
range of Fbased on the 95% CI of our estimate of annual S
F
.
Empirically derived ratios of the fishing mortality rate associated
with maximum sustainable yield (F
msy
) and instantaneous natu-
ral mortality (M) have been postulated to range from
approximately 0.41 to 0.50 for elasmobranchs [37,38]. We used
these values and a recent estimate of Mfor mako sharks in the
North Atlantic (0.075) [39] to calculate a range of approximate
F
msy
values of 0.031–0.038. We note that these values are
approximations as they do not account for gear selectivity. We
compared these F
msy
values with our tagging-derived estimate
of Fto examine whether there is evidence that the stock is
undergoing overfishing (i.e. F.F
msy
) and compared our results
with those from the most recent ICCAT mako shark stock
assessment [27].
We illustrate the influence of Fon mako shark population
size with a simple exponential population model of the form
N
t
¼N
0
e
rt
projected forward for one generation (approx. 26
years [22]), where N
0
is initial population size, tis years and r
is the intrinsic rate of population increase. We calculated r
through a Leslie matrix based on current life-history inputs
[40]. Detailed descriptions of life-history inputs used are pro-
vided as the electronic supplementary material. We compared
results with Fcorresponding to estimates derived from this
study to recent stock assessment [27], tag-recapture data [7]
and the case of no fishing mortality (F¼0).
3. Results
We tagged 46 mako sharks, of which we received sufficient
data from 40 sharks tracked between March 2013 and May
2016 for our analyses; 14 sharks were tagged off the Yucata
´n
Peninsula (8M, 6F) and 26 off the northeast coast of the USA
(19M, 7F). Male sharks ranged in size from 117 to 198 cm FL,
with estimated ages of 2.3 – 9.5 years [29]. Four males tagged
off the Yucata
´n and three off the US east coast exceeded the
size of approximately 50% maturity for males (185 cm FL)
[29]; as such, it is likely that some of these individuals were
mature or reaching maturity at the time of the study.
Female sharks ranged from 122 to 252 cm FL, with estimated
rspb.royalsocietypublishing.org Proc. R. Soc. B 284: 20170658
3
on August 2, 2017http://rspb.royalsocietypublishing.org/Downloaded from

ages of 3–15.4 years. All females were immature based on
age and length at maturity estimates [29].
The SLRTs were successful in providing long-duration
tracks. Track durations averaged 358 days (range: 81– 754)
for mako sharks that ceased reporting prior to the end of
the study and were not harvested, with 11 tracks of non-
harvested makos exceeding 1 year. Four mako sharks were
still actively reporting locations as of the end of the study
(31 May 2016), with tracking durations of 374, 375, 379 and
415 days, respectively. Mako sharks ranged widely throughout
the western North Atlantic Ocean, although there was
little spatial overlap between sharks from each respective tag-
ging location (figure 2). Sharks traversed EEZs under the
sovereignty of 19 countries during the study period (figure 2).
Overall, we identified 12 mako sharks (30%) harvested by
fishers: four (28.6%) sharks tagged off the Yucata
´n and eight
(30.8%) sharks tagged off the US east coast. Mako sharks
were harvested by vessels from five countries, namely
Canada (4), the USA (3), Mexico (3), Spain (1) and Cuba (1)
(figure 2). We estimate that 10 sharks (83.3%) were harvested
by pelagic longliners based on contact with the fishers after
capture (six harvests), or our knowledge of fishery activity
in the vicinity of where the shark was captured and the
locations on land where the transmitter reported (four har-
vests). One shark was harvested by a sport fisher off
New York, USA, and one in a bottom trawler off Maryland,
USA. Harvests occurred during all seasons in both tagging
regions.
The model that held survival constant across seasons and
tagging regions was the most supported, with the lowest
AICc value and 56% of the combined model weight (w
i
;
table 1). Confidence limits for all season and tagging region
parameters in the remaining models crossed 0, suggesting
they were uninformative, and confidence intervals of derived
estimates of S
F
overlapped considerably as well (see elec-
tronic supplementary material, table S2 and figure S1).
Therefore, we used the model that held survival constant
across time and region to derive an estimate of annual S
F
¼
0.72 (95% CI: 0.57–0.83). This can be interpreted as a mako
shark in the western North Atlantic having an approximately
72% probability of surviving a year and not being harvested
by a fisher. The 95% CI estimates of S
F
yielded a range of
F¼0.19– 0.56, which is 5– 18 times greater than estimates of
F
msy
(0.031– 0.038). As fishing mortality estimates exceeded
those associated with maximum sustainable yield (F.F
msy
),
this suggests a state of overfishing of mako sharks in the
western North Atlantic.
In the absence of fishing mortality (F¼0; applied across
all age groups), the population is expected to grow at an
instantaneous rate of 0.037 yr
21
, whereas population
growth decreases to 0.017 yr
21
when F0.02 as estimated
in the 2012 assessment [27]. The population decreases when
40° N
5 Sep 2013
2 Sep 2013
tagging location
daily movements
capture location
vessel track
27 May 2013
30° N
70° W 60° W
Figure 1. Example of location data from a harvested shortfin mako shark illustrating the movements of the shark, the date and approximate capture location, and
the track of the vessel back to a port in Nova Scotia, Canada. (Online version in colour.)
rspb.royalsocietypublishing.org Proc. R. Soc. B 284: 20170658
4
on August 2, 2017http://rspb.royalsocietypublishing.org/Downloaded from

F¼0.10 as reported in Wood et al.[7](r¼20.069) based on
tag–recapture data, and declines precipitously when F¼0.328
(r¼20.361), corresponding to S
F
¼0.72 estimated in our
study (figure 3).
4. Discussion
Using satellite telemetry, we were able to document harvest
events and quantify fisheries interactions of mako sharks in
the western North Atlantic ocean. This included the
number of individuals harvested, the spatial distribution of
harvest events, the countries responsible for harvests and
the relative contribution of different gear types to total har-
vest. Furthermore, to our knowledge this study represents
the first use of satellite telemetry to quantify fishing mortality
(F) of an exploited shark in a fisheries-independent manner.
As expected for the species, the majority of harvests were
attributed to longline fisheries; however, we also observed
harvests in bottom trawl and sport fisheries. Mako sharks
are a popular sport fish and are occasionally recorded in
bottom trawls [41,42], although the relative mortality associ-
ated with these fisheries is assumed to be dwarfed by that of
longlines [41]. The ability to quantify the contribution of indi-
vidual fisheries independent of reporting by fishers is an
appealing aspect of satellite telemetry. Similarly, with 40 indi-
viduals we recorded harvests by fishers from five countries
from both the western and eastern North Atlantic, and
recorded mako sharks travelling through the EEZs of 19
sovereign nations as well as international waters, underscor-
ing the critical need for coordinated international
management efforts.
We discovered high levels of fishing mortality, with 30%
of our sample known to have been harvested by fishers.
Notably, estimates of Fderived from survival models in
our study (0.19–0.56) were considerably higher than esti-
mates available for mako sharks in the North Atlantic
based on stock assessments that rely heavily on fisheries
data. The 2012 ICCAT stock assessment estimated Fon the
order of approximately 0.015–0.024 with Bayesian surplus
and catch-free age-structured production models. Based
on 27 years of conventional tag-return data in which
5813 tags were deployed and 654 (11%) tags were sub-
sequently recovered, Wood et al.[7]estimatedFat
approximately 0.10. Wood et al. [7] computed Fby sub-
tracting Mfrom the total instantaneous mortality rate
50° N capture location
port location
EEZ boundaries
40° N
30° N
20° N
10° N
100° N 10° W20° W30° W40° W50° W60° W70° W80° W90° W
Figure 2. Daily locations of shortfin mako sharks tagged off the US coast (blue dots) and Yucata
´n Peninsula (grey dots) during March 2013–May 2016. Capture and
landing port locations of harvested sharks are indicated, as are boundaries of Exclusive Economic Zones (EEZs). (Online version in colour.)
Table 1. Model selection results of candidate models used to model
survival from fishing mortality (S
F
) of satellite-tagged shortfin mako sharks
in the western North Atlantic Ocean, March 2013 –May 2016.
model
a
K AICc DAICc w
i
S
F
(.) 1 85.8 0.00 0.56
S
F
(location) 2 87.8 2.02 0.20
S
F
(season) 4 88.2 2.36 0.17
S
F
(location þseason) 5 90.3 4.46 0.07
a
Models allow S
F
to remain constant (.), or vary by season or tagging
location (Yucata
´n Peninsula or US east coast). We report the number of
estimable parameters (K), Akaike’s information criterion adjusted for small
sample size (AICc), difference in AICc relative to smallest value (DAICc) and
Akaike weights (w
i
).
3.0 F=0
F= 0.02
F= 0.10
F= 0.33
0
0
0.5
1.0
1.5
2.0
2.5
relative population size
252015
pro
j
ection
y
ear
105
Figure 3. Effect of assuming different levels of F(F¼0; F¼0.02, 2012
ICCAT stock assessment; F¼0.10, tag – recapture data; and F¼0.33, pre-
sent study) when projecting a hypothetical shortfin mako population forward
one generation (approx. 26 years) using a simple exponential population
model. (Online version in colour.)
rspb.royalsocietypublishing.org Proc. R. Soc. B 284: 20170658
5
on August 2, 2017http://rspb.royalsocietypublishing.org/Downloaded from

(Z). As Mwas obtained based on life-history invariant esti-
mators which likely overestimated M[39], Fwas also thus
underestimated.
Our study is unique in that we were able to estimate F
directly from observations of harvest events within a popu-
lation of continuously monitored individuals. Such direct
estimates should be more accurate than those derived from
stock assessments, which estimate Fbased on often incom-
plete or unreliable catch or effort information from the
different fleets. Assuming our sample is truly representative
(i.e. individuals in this study were not more susceptible to
harvest than unmarked sharks of the same age/size class),
our results suggest that mako sharks in the western North
Atlantic are experiencing greater fishing mortality than pre-
viously inferred from fisheries-dependent data sources.
These observations broadly coincide with other studies
suggesting that available fisheries data underestimate true
human-induced mortality of shark populations [43].
Our fishing mortality estimates were 5–18 times greater
than those associated with maximum sustainable yield
(F
msy
), implying the North Atlantic stock of mako sharks is
currently experiencing overfishing (i.e. F.F
msy
). Thus, if
the level of fishing mortality we observed is representative
of the western North Atlantic population, it is likely to be
unsustainable (figure 3). Furthermore, the fact that our har-
vested sample consisted primarily of immature individuals
is concerning because populations of slow-growing, low-
productivity shark species are most negatively impacted by
high mortality in juvenile age classes, as reduced recruitment
into the adult population slows population growth rates [2].
Additionally, despite the high levels we observed, it is poss-
ible that we are still underestimating fishing mortality to
some degree because some fishers may have destroyed or dis-
carded tags at sea and never reported the capture. In such
cases, we would not have been able to detect the harvest
event. We refrain from making a definitive claim regarding
the status of the mako shark stock in the western North
Atlantic, and caution that interpretation of our results must
consider that our estimates of F
msy
did not account for gear
selectivity, and that we tracked sharks within a relatively
narrow size/age class. Still, the discrepancy between our esti-
mates of Fand those derived previously from tag– recapture
[7] or stock assessment [27] highlights the need for further
assessments of mortality in this population.
Given the technological limitations of the tags we used,
which only reported locations when exposed to air, we
were not able to detect natural mortalities. This was sufficient
for our goal of quantifying F; however, we were unable to
estimate total survival because the survival probabilities we
report (S
F
) refer to survival from fishing mortality only. It
is difficult to speculate how prevalent natural mortality was
in our study, but we believe it is unlikely total survival,
S¼e
2(MþF)
, would be much lower than the S
F
we estimated
for several reasons. Evidence suggests natural mortalities
were rare as tracking durations of non-harvested sharks in
our study were often as long as, or longer than, the predicted
manufacturer tag longevity based on battery size and tag pro-
gramming, indicating battery drain may have been the
primary cause of lost contact rather than shark mortality.
Interestingly, our estimate of annual S
F
(0.72; 95% CI: 0.57–
0.83), which only accounts for survival from harvest, is
lower than S(0.79; 95% CI: 0.71– 0.87) reported by Wood
et al. [7], which accounted for all forms of mortality.
We found no strong evidence for an effect of season or
tagging region on S
F
. Simulations suggest our sample size
should have been sufficient to detect regional differences
when effect sizes were moderate to large (i.e. 25– 50%
regional difference in S
F
); however, it is possible that smaller
effects may have gone undetected (see the electronic
supplementary material). Fishing effort is not evenly distrib-
uted at large scales in the north Atlantic [23], and it is likely
that some degree of spatial variation in harvest mortality
also exists. Larger sample sizes and carefully designed
studies will help elucidate such effects in the future, and
have important implications if there exists strong spatial
structuring of the Atlantic mako shark population. For
example, if immature makos show strong fidelity to high-
productivity coastal regions, as recent evidence suggests
[19,21,28], then this segment of the population may have
lower S
F
than mature cohorts as a result of greater exposure
to fishing pressure. This would limit recruitment into the
breeding cohort, subsequently limiting population growth
and recovery potential. The stock structure of mako sharks
in the Atlantic is currently not well resolved, and future fish-
eries-independent studies of mako shark movements and
fishing mortality will be important in developing focused
management actions.
(a) Future studies
The known-fate modelling approach we used has been
widely adopted in studies of terrestrial wildlife monitored
via telemetry [44– 46]; however, to our knowledge appli-
cation to pelagic vertebrates has been limited to sea turtles
[47]. Like many contemporary approaches to modelling sur-
vival of wildlife [11], known-fate models are appealing
because of the ability to model survival probability as a func-
tion of individual covariates (e.g. location, time of year, sex or
age-class), and to employ information-theoretic approaches
to compare relative support of hypotheses regarding the
influence of covariates on survival. Furthermore, because
annual survival probabilities are the products of the survival
likelihoods of each sample interval, it is possible to derive
annual survival estimates from tags with deployment lengths
less than 1 year, provided enough individuals are tracked
within each respective sample interval. This may be particu-
larly useful for studies using PSATs, which often do not last
as long as fin-mounted SLRTs. Known-fate models quantify
survival in discrete time, and we note that similarly powerful
methods of modelling wildlife survival in continuous time
that incorporate individual covariates and allow for infor-
mation-theoretic model selection are available, such as
proportional hazard models [11]. The appropriate statistical
approach will depend on specific study objectives and data
availability.
For management and conservation purposes, knowledge
of the relative contributions of both natural and human-
induced causes to total mortality is important. At present,
we are not aware of any studies that provide direct measu-
res of natural mortality for pelagic shark populations. For
pelagic and other sharks, Mis generally calculated through
life-history-invariant methods based on life-history character-
istics, such as maximum age and parameters of size-at-age
curves (e.g. [48,49]). While life-history-invariant methods
are efficient in that they do not require large amounts of
data, the estimates they provide are of unknown precision
rspb.royalsocietypublishing.org Proc. R. Soc. B 284: 20170658
6
on August 2, 2017http://rspb.royalsocietypublishing.org/Downloaded from

[50]. As empirical management benchmarks such as F
msy
can
be generated relative to these estimates of M, it would be
advantageous to have some independent measure of natural
mortality. Incorporation of PSATs into future studies may
prove fruitful in this regard. PSATs archive time-series data
of depth and temperature, and thus mortality can be inferred
from records that indicate constant depth for extended
periods of time (in most tags this scenario will trigger the
pop-off mechanism). PSATs have been used to measure
post-release survival of shark species [15– 18] with the
focus on survival over relatively short time periods (days to
weeks). However, with slight modifications to study design
and analysis it is feasible to use these tags to also generate
annual survival estimates. An additional benefit of PSATs
is that the depth data from the tags could be combined
with information on depth and spatial distribution of
fishing effort to derive selectivity in stock assessments, as
demonstrated by Carvalho et al. [51].
The benefit of telemetry-based studies in constructing
effective conservation and management plans will be most
realized in their ability to contribute to, or be directly incor-
porated into, stock assessments. In the most basic scenario,
independent, telemetry-derived parameter estimates (e.g. F
or M) may be used to improve stock assessment models,
for example by generating informative priors for Bayesian
models. Beyond this, harvest and survival data, along with
the associated movement data, could be incorporated directly
into integrated spatially explicit assessment models, and used
to test hypotheses about population dynamics and structure
[52]. Although electronic tagging data have been used
for capture– recapture and population dynamics models
[53,54], estimates of Fand Mderived directly from satellite
telemetry data are still lacking for sharks and other pelagic
fishes. Given the widespread use of satellite telemetry
to study the ecology of sharks, we suggest attention be
given to using these data to explicitly address fundamental
questions of survival and mortality.
5. Conclusion
Our study illustrates how the use of satellite telemetry to
track individual sharks in a population is a potentially
time-efficient way to gather useful fisheries interaction data.
By tracking 40 mako sharks over a 3-year period we were
able to capture a wealth of information regarding the distri-
bution of harvests and the associated fisheries and
countries involved. Significantly, we were able to quantify F
in a fisheries-independent manner, a metric with direct appli-
cation to stock assessment and stock status evaluation.
Importantly, the fishing mortality rates we observed were
well above those previously reported for mako sharks in
the North Atlantic, calling into question the sustainability
of current fishing pressure on this population. This, combined
with movements across multiple international management
jurisdictions and the documentation of harvest by multiple
countries, underscores the importance of coordinated inter-
national management to ensure the long-term sustainability
of mako sharks in the North Atlantic.
Ethics. This study was conducted under NSU IACUC control # 064-
398-15-0203.
Data accessibility. Mako shark tagging and harvest data are available
from the Dryad Digital Repository (http://dx.doi.org/10.5061/
dryad.h9f3c) [55].
Authors’ contributions. M.E.B, E.C., J.J.V, B.M.W. and M.Sh. conceived the
study. M.E.B and E.C. developed and carried out analyses. M.E.B,
J.J.V, G.C.M.H., M.Sa. B.M.W. and M.Sh. conducted fieldwork.
M.E.B wrote the paper with extensive contributions from E.C.,
J.J.V., B.M.W. and M.Sh. All authors provided input on the paper.
Competing interests. We have no competing interests.
Funding. This research was supported by Florida Sea Grant (award
UFDSP00010205), Swiss Shark Foundation/Hai Stiftung, Guy
Harvey Ocean Foundation and Virgin Unite.
Acknowledgements. We thank D. Burkholder, G. Jacoski, A. Mendillo,
L. Sampson, C. Donilon, R. de la Parra and G. Schellenger for field-
work assistance.
References
1. Dulvy NK et al. 2014 Extinction risk and
conservation of the world’s sharks and rays. eLife 3,
e00590. (doi:10.7554/eLife.00590)
2. Corte
´s E. 2002 Incorporating uncertainty into
demographic modeling: application to shark
populations and their conservation. Conserv. Biol.
16, 1048–1062. (doi:10.1046/j.1523-1739.2002.
00423.x)
3. Frid A, Baker GG, Dill LM. 2008 Do shark declines
create fear-released systems? Oikos 117, 191– 201.
(doi:10.1111/j.2007.0030-1299.16134.x)
4. Heithaus MR, Frid A, Wirsing AJ, Worm B. 2008
Predicting ecological consequences of marine top
predator declines. Trends. Ecol. Evol. 23, 202– 210.
(doi: 10.1016/j.tree.2008.01.003)
5. Ferretti F, Worm B, Britten GL, Heithaus MR, Lotze
HK. 2010 Patterns and ecosystem consequences of
shark declines in the ocean. Ecol. Lett. 13, 1055–
1071. (doi: 10.1111/j.1461-0248.2010.01489.x)
6. Britten GL, Dowd M, Minto C, Ferretti F, Boero F,
Lotze HK. 2014 Predator decline leads to decreased
stability in a coastal fish community. Ecol. Lett. 17,
1518–1525. (doi:10.1111/ele.12354)
7. Wood AD, Collie JS, Kohler NE. 2007 Estimating
survival of the shortfin mako Isurus oxyrinchus
(Rafinesque) in the north-west Atlantic from tag-
recapture data. J. Fish. Biol. 71, 1679– 1695. (doi:
10.1111/j.1095-8649.2007.01634.x)
8. Aires-da-Silva AM, Maunder MN, Gallucci VF, Kohler
NE, Hoey JJ. 2009 A spatially structured tagging
model to estimate movement and fishing mortality
rates for the blue shark (Prionace glauca) in the
North Atlantic Ocean. Mar. Freshw. Res. 60, 1029–
1043. (doi: 10.1071/MF08235)
9. Pine WE, Pollock KH, Hightower JE, Kwak TJ, Rice
JA. 2003 A review of tagging methods for
estimating fish population size and components of
mortality. Fisheries 28, 10– 23. (doi: 10.1577/1548-
8446(2003)28[10:AROTMF]2.0.CO;2)
10. Bjo
¨rnsson B, Karlsson H, Thorsteinsson V,
Solmundsson J. 2011 Should all fish in mark –
recapture experiments be double-tagged? Lessons
learned from tagging coastal cod (Gadus morhua).
ICES. J. Mar. Sci. 68, 603 – 610. (doi:10.1093/
icesjms/fsq187)
11. Murray DL. 2006 On improving telemetry-based
survival estimation. J. Wildl. Manage. 70, 1530–
1543. (doi: 10.2193/0022-541X(2006)70[1530:
OITSE]2.0.CO;2)
12. Heupel MR, Simpfendorfer CA. 2002 Estimation of
mortality of juvenile blacktip sharks, Carcharhinus
limbatus, within a nursery area using telemetry
data. Can. J. Fish. Aquat. Sci. 59, 624– 632. (doi:
10.1139/f02-036)
13. Heupel MR, Simpfendorfer CA. 2011 Estuarine
nursery areas provide a low-mortality environment
for young bull sharks, Carcharhinus leucas.Mar.
Ecol-Prog. Ser. 433, 237– 244. (doi: 10.3354/
meps09191)
14. Knip DM, Heupel MR, Simpfendorfer CA. 2012
Mortality rates for two shark species occupying a
shared coastal environment. Fish. Res. 125– 126,
184–189. (doi: 10.1016/j.fishres.2012.02.023)
rspb.royalsocietypublishing.org Proc. R. Soc. B 284: 20170658
7
on August 2, 2017http://rspb.royalsocietypublishing.org/Downloaded from

15. Musyl MK, Brill RW, Curran DS, Fragoso NM,
McNaughton LM, Nielsen A, Kikkawa BS, Moyes CD.
2011 Post-release survival, vertical and horizontal
movements, and thermal habitats of five species of
pelagic sharks in the central Pacific Ocean. Fish. B-
NOAA. 109, 341– 368.
16. Hutchinson MR, Itano DG, Muir JA, Holland KN. 2015
Post-release survival of juvenile silky sharks captured
in a tropical tuna purse seine fishery. Mar. Ecol-Prog.
Ser. 521, 143–154. (doi:10.3354/meps11073)
17. Marshall H, Skomal G, Ross PG, Bernal D. 2015 At-
vessel and post-release mortality of the dusky
(Carcharhinus obscurus) and sandbar (C. plumbeus)
sharks after longline capture. Fish. Res. 172, 373–
384. (doi:10.1016/j.fishres.2015.07.011)
18. Campana SE, Joyce W, Fowler M, Showell M. 2016
Discards, hooking, and post-release mortality of
porbeagle (Lamna nasus), shortfin mako (Isurus
oxyrinchus), and blue shark (Prionace glauca) in the
Canadian pelagic longline fishery. ICES. J. Ma. Sci.
73, 520–528. (doi:10.1093/icesjms/fsv234)
19. Block BA et al. 2011 Tracking apex marine predator
movements in a dynamic ocean. Nature 47, 86– 90.
(doi:10.1038/nature10082)
20. Lea JSE et al. 2015 Repeated, long-distance
migrations by a philopatric predator targeting
highly contrasting ecosystems. Sci. Rep.5, 11202.
(doi:10.1038/srep11202)
21. Rogers PR, Huveneers C, Page B, Goldsworthy SD,
Coyne M, Lowther AD, Mitchell JG, Seuront L. 2015
Living on the continental shelf edge: habitat use of
juvenile shortfin makos Isurus oxyrinchus in the
Great Australian Bight, southern Australia. Fish.
Oceanogr. 24, 205–218. (doi:10.1111/fog.12103)
22. Corte
´sEet al. 2010 Ecological risk assessment of
pelagic sharks caught in Atlantic pelagic longline
fisheries. Aquat. Living. Resour. 23, 25– 34. (doi:10.
1051/alr/2009044)
23. Queiroz N et al. 2016 Ocean-wide tracking of
pelagic sharks reveals extent of overlap with
longline fishing hotspots. Proc. Natl Acad. Sci. USA
113, 1582–1587. (doi:10.1073/pnas.1510090113)
24. Campana SE, Marks L, Joyce W. 2005 The biology
and fishery of shortfin mako sharks (Isurus
oxyrinchus) in Atlantic Canadian waters. Fish. Res.
73, 341–352. (doi:10.1016/j.fishres.2005.01.009)
25. Cailliet GM et al. 2015 Isurus oxyrinchus. The IUCN
Red List of Threatened Species 2009,
e.T39341A10207466. (doi:10.2305/IUCN.UK.2009-2.
RLTS.T39341A10207466.en)
26. Walls RHL, Soldo A. 2016 Isurus oxyrinchus. The
IUCN Red List of Threatened Species 2016,
e.T39341A16527941. See http://www.iucnredlist.
org/details/summary/39341/3.
27. ICCAT. 2012 shortfin mako stock assessment and
ecological risk assessment meeting. See https://
www.iccat.int/Documents/Meetings/Docs/2012_
SHK_ASS_ENG.pdf.
28. Vaudo JJ, Byrne ME, Wetherbee BM, Harvey GM,
Shivji MS. In press. Long-term satellite tracking
reveals region-specific movements of a large pelagic
predator, the shortfin mako shark, in the western
North Atlantic Ocean. J. Appl. Ecol. (doi:10.1111/
1365-2664.12852)
29. Natanson LJ, Kohler NE, Ardizzone D, Cailliet GM,
Wintner SP, Mollet HF. 2006 Validated age and
growth estimates for the shortfin mako shark, Isurus
oxyrinchus, in the North Atlantic Ocean. Environ. Biol.
Fish. 77, 367– 383. (doi:10.1007/s10641-006-9127-z)
30. Vaudo JJ, Wetherbee BM, Wood AD, Weng K,
Howey-Jordan LA, Harvey GM, Shivji MS. 2016
Vertical movements the shortfin mako sharks (Isurus
oxyrinchus) in the western North Atlantic Ocean are
strongly influenced by temperature. Mar. Ecol-Prog.
Ser. 547, 163–175. (doi:10.3354/meps11646)
31. Johnson DS. 2015 Crawl: fit continuous-time
correlated random walk models to animal
movement data. R package version 1.5. See http://
CRAN.R-project.org/package=crawl.
32. R Core Team. 2015 R: a language and environment
for statistical computing. Vienna, Austria: R
Foundation for Statistical Computing.
33. Hays GC, Broderick AC, Godley BJ, Luschi P, Nichols WJ.
2003 Satellite telemetry suggest high levels of
fishing-induced mortality in marine turtles. Mar. Ecol-
Prog. Ser. 262, 305 –309. (doi:10.3354/meps262305)
34. White GC, Burnham KP. 1999 Program MARK:
survival estimation from populations of marked
animals. Bird Study 46, S120–S138. (doi:10.1080/
00063659909477239)
35. Burnham KP, Anderson DR. 2002 Model selection
and multimodel inference, 2nd edn. New York, NY:
Springer.
36. Laake JL. 2013 RMark: An R Interface for analysis of
capture-recapture data with MARK. College Station,
TX: Institute of Renewable Natural Resources, Texas
A&M University.
37. Au DW, Smith SE, Show C. 2008 Shark productivity
and reproductive protection, and a comparison with
teleosts. In Sharks of the open ocean: biology,
fisheries and conservation (eds MD Camhi, EK
Pikitch, EA Babcock), pp. 298–308. London, UK:
Blackwell.
38. Zhou S, Yin S, Thorson JT, Smith ADM., Fuller M.
2012 Linking fishing mortality reference points to
life history traits: an empirical study. Can. J. Fish.
Aquat. Sci. 69, 1292–1301. (doi:10.1139/f2012-
060)
39. Corte
´s E. 2016 Perspectives on the intrinsic rate of
population growth. Methods Ecol. Evol. 7, 1136–
1145. (doi:10.1111/2041-210X.12592)
40. ICCAT. 2017 Report of the 2017 ICCAT shortfin mako
data preparatory meeting. See https:// www.iccat.
int/Documents/Meetings/Docs/2017_SMA_DATA_
PREP_ENG.pdf.
41. Casey JG, Kohler NE. 1992 Tagging studies on the
Shortfin Mako Shark (Isurus oxyrinchus) in the
Western North Atlantic. Aust. J. Mar. Fresh. Res. 43,
45–60. (doi:10.1071/MF9920045)
42. Beutel D, Skrobe L, Castro K, Ruhle Sr P, O’Grady
J, Knight J. 2008 Bycatch reduction in the
Northeast USA directed haddock bottom trawl
fishery. Fish. Res. 94, 190– 198. (doi:10.1016/j.
fishres.2008.08.008)
43. Clarke SC, McAllister MK, Milner-Gulland EJ,
Kirkwood GP, Michielsens CGJ, Agnew DJ, Pikitch
EK, Nakano H, Shivji MS. 2006 Global estimates of
shark catches using trade records from commercial
markets. Ecol. Lett. 9, 1115–1126. (doi:10.1111/j.
1461-0248.2006.00968.x)
44. Moynahan BJ, Lindberg MS, Thomas JW.
2006 Factors contributing to process variance
in annual survival of female greater sage-grouse
in Montana. Ecol. Appl. 16, 1529–1538.
(doi:10.1890/1051-0761(2006)016[1529:FCTPVI]2.0.
CO;2)
45. Schwartz CC, Haroldson MA, White GC. 2010
Hazards affecting grizzly bear survival in the greater
Yellowstone ecosystem. J. Wildl. Manage. 74, 654–
667. (doi:10.2193/2009-206)
46. Ackerman JT, Herzog MP, Hartman CA, Herring G.
2014 Forster’s tern chick survival in response to a
managed relocation of predatory California gulls.
J. Wildl. Manage. 78, 818–829. (doi:10.1002/jwmg.
728)
47. Sasso CR, Epperly SP. 2007 Survival of pelagic juvenile
loggerhead turtles in the open ocean. J. Wildl.
Manage. 71, 1830–1835. (doi:10.2193/2006-448)
48. Hoenig JM. 1983 Empirical use of longevity data to
estimate mortality rates. Fish. B-NOAA. 81, 898–903.
49. Jensen AL. 1996 Beverton and Holt life history
invariants result from optimal trade-off of
reproduction and survival. Can. J. Fish. Aquat. Sci.
53, 820–822. (doi:10.1139/f95-233)
50. Pascual MA, Iribarne OO. 1993 How good are
empirical predictions of natural mortality? Fish. Res.
16, 17–24. (doi:10.1016/0165-7836(93)90107-I)
51. Carvalho F, Ahrens R, Murie D, Bigelow K, Aires-Da-
Silva A, Maunder MN, Hazin F. 2015 Using pop-up
satellite archival tags to inform selectivity in
fisheries stock assessment models: a case study for
the blue shark in the South Atlantic Ocean.
ICES. J. Mar. Sci. 72, 1715– 1730. (doi:10.1093/
icesjms/fvs026)
52. Sippel T et al. 2015 Using movement data from
electronic tags in fisheries stock assessment: a
review of models, technology and experimental
design. Fish. Res. 163, 152–160. (doi:10.1016/j.
fishres.2014.04.006)
53. Eveson JP, Basson M, Hobday AJ. 2012 Using
electronic tag data to improve mortality and
movement estimates in a tag-based spatial fisheries
assessment model. Can. J. Fish. Aquat. Sci. 69,
869–883. (doi:10.1139/f2012-026)
54. Whitlock RE, McAllister MK, Block BA. 2012
Estimating fishing and natural mortality rates for
Pacific Bluefin tuna (Thunnus orientalis) using
electronic tagging data. Fish. Res. 119– 120, 115 –
127. (doi:10.1016/j.fishres.2011.12.015)
55. Byrne ME, Corte
´s E, Vaudo JJ, Harvey GC, Sampson
M, Wetherbee BM, Shivji M. 2017 Data from:
Satellite telemetry reveals higher fishing mortality
rates than previously estimated, suggesting
overfishing of an apex marine predator.
Dryad Digital Repository. (http://dx.doi.org/10.5061/
dryad.h9f3c)
rspb.royalsocietypublishing.org Proc. R. Soc. B 284: 20170658
8
on August 2, 2017http://rspb.royalsocietypublishing.org/Downloaded from