Long-term satellite tracking reveals region-specific movements of a large pelagic predator, the shortfin mako shark, in the western North Atlantic Ocean

Summary

1. As upper level predators, sharks serve an important role in marine ecosystems, but are often at risk from fisheries. Successful management of these species will require detailed information about their movements and distributions.
2. Using satellite telemetry, we investigated the long-term horizontal movements and seasonal distributions of shortfin mako sharks Isurus oxyrinchus in the western North Atlantic Ocean.
3. Twenty-six sharks (14 USA, 12 Mexico) were tracked for durations of 78–527 days. Ten sharks were tracked for >1 year. Sharks displayed region-specific movements, with little distributional overlap between the Gulf of Mexico and Caribbean Sea, and the western North Atlantic. Sharks tagged off the USA ranged over a larger area, including shelf and pelagic habitats. Their core distribution was largely over the continental shelf and fluctuated seasonally, ranging from South Carolina, USA, in the winter to Nova Scotia, Canada, in the autumn, and appeared to follow seasonal productivity peaks while favouring warmer waters. Sharks tagged off Mexico displayed more restricted movements, largely confined to shelf habitats, with core activity centred year-round on the eastern Campeche Bank, Mexico.
4. Sharks moved across the jurisdictional management boundaries of 17 nations, and the proportion of tracked sharks harvested (22%) was twice that obtained from previous fisheries-dependent, conventional tagging studies.
5. Sharks also displayed considerable variability in movements, with seven sharks tagged off the USA making long-distance, highly directional southern excursions into unproductive subtropical/tropical waters before returning north.
6. Policy implications. The large-scale and region-specific movements of shortfin mako sharks underscore the need for close cooperation amongst western North Atlantic nations and implementation of regionally and seasonally specific management strategies. The movement patterns also provide baseline information, which could be used in spatially explicit stock assessment models. Identification of high-use areas by shortfin mako sharks provides focal areas for quantifying interactions with fisheries. The high harvest rate observed in our fisheries-independent tracking study raises questions about the true rate of fisheries mortality experienced by shortfin mako sharks, calling for a cautionary interpretation of past stock assessments used to determine management policy for this highly migratory species of conservation concern.

Introduction

Many large shark species have experienced substantial declines as a result of fisheries (Dulvy et al. 2014), raising concerns about broader ecosystem consequences and sustainable management of their populations. Knowledge of a species' spatial dynamics is key to understanding its ecological role, improving population assessment and management, and predicting potential impacts of climate change on movements and distribution (Hayes et al. 2016). In the case of widely distributed, highly mobile sharks, information on long-term movements and habitat use also reveals behavioural aspects of direct management and conservation relevance, including population boundaries, essential habitat, spatiotemporal interactions with various fisheries and stakeholder jurisdictions responsible for fisheries management. However, such information is largely lacking for oceanic sharks, which are captured in large numbers world-wide (Queiroz et al. 2016).

Shortfin mako sharks Isurus oxyrinchus Rafinesque (hereafter, mako sharks) are large, highly mobile, pelagic predators that inhabit tropical and temperate waters circumglobally and are prized in both recreational and commercial fisheries (Campana, Marks & Joyce 2005). Their wide distribution and habitat use patterns (Casey & Kohler 1992; Rogers et al. 2015; Vaudo et al. 2016b) are similar to those of commercially important fishes (e.g. billfish and tunas), leading to high encounter rates with fisheries (ICCAT 2013; Queiroz et al. 2016). As a result, mako shark is one of the most common species in the global fin trade (Clarke et al. 2006). Thus, fishery exploitation is a major source of mortality for mako shark populations, which, because of their life-history characteristics, have a high risk of overexploitation (Cortés et al. 2010). Despite this risk, mako shark management is limited as there is a great deal of uncertainty in population estimates because of sparse biological information on the species, including its movement ecology (E. Cortés, pers. comm). Integrating knowledge of fish movements into spatially explicit population dynamics models is being urged for improving stock assessments and management (Sippel et al. 2015; Braccini, Aires-da-Silva & Taylor 2016).

In the western North Atlantic, current understanding of mako shark movements and distribution, and much of the basis for its management is based primarily on fishery catch data and conventional tag-and-recapture records (Casey & Kohler 1992; Kohler et al. 2002). Although such fisheries-dependent studies have provided valuable information, they have limitations. These include low tag recovery rates, sampling efforts biased by the spatiotemporal distribution of fisheries effort, and lack of information about paths travelled by individuals between tagging and recapture and thus characteristics of habitat use, resulting in an incomplete understanding of their movement patterns. More recently, satellite tracking of marine species has provided fisheries-independent observations of mako shark movements in the eastern North Pacific, southeastern Indian and eastern North Atlantic Oceans (Block et al. 2011; Rogers et al. 2015; Queiroz et al. 2016). Knowledge of mako shark movement ecology, however, is still severely limited, and minimal information exists for the western North Atlantic Ocean (Vaudo et al. 2016b). Such baseline ecological and behavioural information is necessary to inform stock assessments and management planning.

Conventional tag – recapture data suggest seasonal movement patterns for mako sharks in the western North Atlantic Ocean characterized by occupation of continental shelf waters of the US mid-Atlantic states during the spring, northward movement along the continental shelf to waters off the northeast USA and the Canadian Grand Banks during the summer and early autumn, followed by movement into the Gulf Stream and Sargasso Sea to overwinter (termed the ‘Sargasso Sea’ hypothesis; Casey & Kohler 1992). Casey & Kohler (1992) hypothesized that this migration pattern was primarily driven by water temperature, because mako sharks have typically been caught in waters with sea surface temperatures (SST) of 17–22 °C. Thus, this migration pattern would allow mako sharks to remain in waters within this temperature range. Recent work in the western North Atlantic suggests that temperature also influences the vertical movements of mako sharks, which have a minimum temperature threshold of ~12 °C (Vaudo et al. 2016b). Collectively, these studies suggest temperature is likely a major factor influencing movements of mako sharks in regions with seasonal changes in water temperature.

Dramatic seasonal temperature changes, however, do not typify all regions occupied by mako sharks in the western North Atlantic. The Gulf of Mexico and Caribbean Sea (hereafter, GMC) are subtropical/tropical waters characterized by relatively stable temperatures and productivity compared to the temperate waters of the western North Atlantic (Longhurst 2007). Such differences in the variability in environmental conditions between regions may result in differing movement patterns. For example, animal populations experiencing seasonally predictable variations in resource availability commonly engage in seasonal migrations, whereas those experiencing stable annual conditions are more likely to exhibit more restricted ranges (Jonzén et al. 2011).

We used satellite telemetry to provide the first high-resolution, long-term study of the seasonal movement ecology of the shortfin mako shark in the western North Atlantic. We describe and compare movement patterns of this species in the temperate and subtropical/tropical regions of the western North Atlantic Ocean and show that mako sharks exhibit regionally variable movement patterns and space use dynamics. We demonstrate that the ‘Sargasso Sea’ hypothesis does not adequately describe mako shark movements in the western North Atlantic, and outline relevance of the findings for management and conservation of this internationally exploited species.

Materials and methods

During 2013 and 2014 we captured mako sharks via rod and reel off the Yucatan Peninsula in the vicinity of Isla Mujeres, Mexico (~21·29°N, 86·29°W) and in the vicinity of Ocean City, Maryland, USA (~38·10°N, 74·50°W). Sharks were either secured alongside the vessel or brought on board for tagging, in which case a saltwater hose was placed in the mouth to irrigate the gills. We measured and sexed the sharks and attached Smart Position Only Transmitters (SPOT; Wildlife Computers, Redmond, WA, USA) to each shark's dorsal fin. The tags directly communicated with the Argos tracking system (www.argos-system.org) when a shark's dorsal fin broke the sea surface, providing an estimated position (latitude and longitude).

To account for variable temporal intervals and spatial accuracy between Argos locations, we fit a continuous-time correlated random walk model (CTCRW) via a Kalman filter within a state-space framework to each shark's track (Johnson et al. 2008). The Argos system assigned one of 6 location classes to each positional estimate, representing the location's accuracy, which were used as covariates in an error model based on reported error distributions (Vincent et al. 2002). To account for bouts of directed travel or movement drift as a result of large-scale ocean currents, we fit a modified version of the CTCRW that included a random drift component for each shark sec. Johnson et al. (2008). We used the fitted models to predict daily location estimates at regular 24-h intervals. CTCRW models were fit using the package crawl (Johnson 2013) in the R statistical computing environment (http://cran.r-project.org).

To improve the accuracy of these models (Patterson et al. 2010), we first applied a speed-distance-angle algorithm to filter out improbable Argos locations (Freitas et al. 2008). We filtered out locations requiring travel speeds >4·5 m s⁻¹ (provided these locations were >5 km from the previous location) and locations requiring turn angles of ≥165° and ≥155° if the distance between successive locations was >5 and 8 km respectively. Because prediction uncertainty of the CTCRW increases with increasing time between data points, we did not include daily location predictions associated with gaps in the raw Argos data of ≥5 days in further analyses.

For each shark, we calculated the distance between each daily location and the tagging location. We then plotted dispersion from tagging location as a function of time to compare movements between tagging locations. For each daily location, we calculated SST by taking the mean of SST values within the 95% CI surrounding each daily location prediction. SST values were obtained by referencing the date of each location with the daily Multi-scale Ultra-high Resolution (MUR) SST data set (http://mur.jpl.nasa.gov/), which provides daily SST at a spatial resolution of 0·01°.

We quantified the mako shark seasonal distributions by constructing utilization distributions (UD). Seasonal distributions were based on meteorological seasons: summer (June–August), autumn (September–November), winter (December–February) and spring (March–May). A UD is a spatial probability distribution describing the probability of an animal occurring in a specific location during a given time period. As our goal was to describe population-level distributions, we used fixed kernel density analysis to construct UDs using the combined daily locations of all sharks within a given season. To avoid possible biases from the addition of short tracks, we excluded locations from sharks that were tracked for ≤45 days within a specific season and year. To mitigate tagging location bias, we censored the first 10 days of each track to allow individuals to disperse from the tagging location. To mitigate individual sharks from biasing the results, we created a 0·5° × 0·5° spatial grid and counted the total number of daily locations from all sharks within each cell for each season and multiplied that value by the number of unique sharks within the cell. The resulting grid was used as a weighting factor. Thus, UDs were weighted more heavily towards areas that were frequently visited by many individuals. To achieve the final UD for each season, we set all portions that overlapped land to 0 and normalized all values so that the summation of all cells equalled 1.

To highlight seasonal distributional shifts, we plotted the seasonal 95%, 75%, and 50% UD isopleths. The 50% isopleth is commonly used to define core-use areas within animal home ranges (e.g., Byrne & Chamberlain 2011), and here can be interpreted as the core distribution area.

To provide ecological context to seasonal distributions, we plotted UDs over mean seasonal measures of SST and primary productivity (PP; mg C m⁻² day⁻¹). Seasonal SST values were derived by averaging daily MUR SST within each season (spring 2013–summer 2015). Seasonal PP was estimated by averaging monthly PP values within each season during the study period [monthly composite PP estimates (0·08° resolution) from: http://www.science.oregonstate.edu/ocean.productivity].

Because movement paths revealed several sharks engaged in long distance, highly directed, round-trip excursions into southern regions (see Results), we quantified several aspects of this distinctive behaviour. We considered southward and northward travelling portions of each trip to begin when the latitudinal displacements of daily locations exceeded 0·3° day⁻¹ for >3 consecutive days. Residence period at the southern terminus of each excursion was characterized by daily latitude displacements <0·3° day⁻¹; using this displacement threshold, we determined the starting date of each excursion, the duration of residence at the southern terminus of each excursion and the total trip duration. Mean turning angles for the southward and northward movement portions of each excursion were used to quantify directionality. A turning angle of 0° represents no change in direction and a turning angle of 180° represents a complete reversal of direction. Thus, as the mean turn angle of a path approaches 0° with low variance, the path becomes straighter and more directionally persistent.

Results

We tagged 32 mako sharks with SPOT tags (Table 1). Six sharks did not report or had prohibitively short duration tracks (≤11 days) and were excluded from further analyses. Tracks for the 26 sharks that we analysed occurred between 23 March 2013 and 31 August 2015 with track durations of 78–527 days. Fourteen sharks were continuously tracked for periods >300 days. We constructed tracks of daily location estimates from 12 sharks tagged off Mexico (five female, seven male), and 14 sharks tagged off the USA (3 female, 11 male). Females were 148–252 cm and males 128–198 cm fork length (FL). On the basis of size, all females and most males were immature, however, six males (Mexico = 4, USA = 2) were larger than the size at 50% maturity (185 cm; Natanson et al. 2006), suggesting some males were mature or nearing maturity during the study period. The majority of tags (19) stopped transmitting data for unknown reasons, however, at least seven sharks were harvested by fishers based on tag returns and observations of tags continually reporting from a static location on land (Fig. 1). Sharks in the western North Atlantic outside of the GMC (hereafter, WNA; note: the entire western North Atlantic will be referred to in full) reported more frequently (4·47 ± 0·50 Argos locations day⁻¹; mean ± SD) than GMC sharks (2·53 ± 0·29 Argos locations day⁻¹).

Table 1. Tagging information for shortfin mako sharks in the western North Atlantic. Tracking duration and total Argos detections through 31 August 2015

Shark #	Tagging location	Sex	Fork length (cm)	Deployment date	Tracking duration (days)	Total Argos detections
1	Isla Mujeres, Mexico	F	180	23 March 2013	411	1408
2	Isla Mujeres, Mexico	F	175	25 March 2013	297	608
3a^a Harvested by fishers.	Isla Mujeres, Mexico	F	250	8 April 2013	293	581
4	Isla Mujeres, Mexico	M	175	8 April 2013	211	120
5a^a Harvested by fishers.	Isla Mujeres, Mexico	M	186	21 March 2014	125	519
6	Isla Mujeres, Mexico	F	148	22 March 2014	527	963
7	Isla Mujeres, Mexico	M	193	23 March 2014	296	1287
8	Isla Mujeres, Mexico	M	195	24 March 2014	138	357
9	Isla Mujeres, Mexico	M	179	24 March 2014	520	1447
10	Isla Mujeres, Mexico	M	193	25 March 2014	523	979
11a^a Harvested by fishers.	Isla Mujeres, Mexico	M	178	25 March 2014	494	1324
12	Isla Mujeres, Mexico	F	252	29 March 2014	203	560
13a^a Harvested by fishers.	Ocean City, USA	M	168	27 May 2013	98	670
14	Ocean City, USA	F	175	28 May 2013	317	2022
15	Ocean City, USA	M	172	28 May 2013	251	1444
16a^a Harvested by fishers.	Ocean City, USA	M	198	28 May 2013	89	579
17	Ocean City, USA	M	155	31 May 2013	404	1912
18a^a Harvested by fishers.	Ocean City, USA	M	153	17 May 2014	357	1590
19	Ocean City, USA	M	155	19 May 2014	468	1468
20	Ocean City, USA	F	178	19 May 2014	468	1693
21	Ocean City, USA	F	188	20 May 2014	379	1106
22	Ocean City, USA	M	179	20 May 2014	438	907
23	Ocean City, USA	M	150	20 May 2014	357	538
24a^a Harvested by fishers.	Ocean City, USA	M	180	21 May 2014	112	357
25	Ocean City, USA	M	192	22 May 2014	78	281
26	Ocean City, USA	M	128	11 September 2014	98	670

• ^a Harvested by fishers.

Figure 1

Seasonal population-level utilization distributions (UD) for shortfin mako sharks tracked in the western North Atlantic Ocean during 2013–2015 and plotted in relation to primary productivity (left column) and sea surface temperature (right column). Seasons are summer (June–August), autumn (September–November), winter (December–February) and spring (March–May). Red circles indicate harvest locations. [Colour figure can be viewed at wileyonlinelibrary.com]

Collectively, mako sharks ranged over large portions of the western North Atlantic Ocean, spanning a geographic area from 9·7°–51·9°N latitude and 35·6°–96·6°W longitude (Fig. 1). Overall, there was little spatial overlap in movements of sharks from each tagging location. However, one male tagged off Mexico (#7) exited the Caribbean between Cuba and Hispaniola 38 days post-tagging, travelled north and spent the remainder of its track (238 days) along the northeastern coast of the USA and Canada. Two male sharks (#22 and #18) tagged off the USA entered the Caribbean, making short duration trips to the south side of Cuba and the coast of Venezuela respectively (Fig. 2).

Figure 2

Daily locations across seasons of 26 shortfin mako sharks tagged off Mexico (yellow) and the USA (red). Seasons are summer (June–August), autumn (September–November), winter (December–February) and spring (March–May). Black diamonds indicate tagging locations. The arrows from north to south indicate the Grand Banks, Cape Hatteras and the Campeche Bank. [Colour figure can be viewed at wileyonlinelibrary.com]

Sharks in the NWA travelled greater distances, often by several orders of magnitude, than those in the GMC (Fig. 3). With one exception, all sharks tagged off Mexico remained within the GMC region (Fig. 2), with only two travelling >1000 km from their tagging location even after >1 year (Fig. 3). GMC sharks made extensive use of the eastern edge of the Campeche Bank, Mexico, often remaining along the bank edge for several consecutive months. GMC sharks also used the shelf waters south of the Mississippi River Delta in the northern Gulf of Mexico, the southernmost coast of the Gulf of Mexico and the shelf waters of the western portion of the northern Caribbean Sea, north of Honduras. NWA sharks travelled widely throughout the NWA, occupying continental shelf waters off eastern North America from South Carolina, USA to the Grand Banks, Canada, the Gulf Stream, the Sargasso Sea, the Caribbean Sea and the North Equatorial Current. Although these sharks exhibited considerable individual variability in their movements, all used the continental shelf and slope waters as well as the northern edge of the Gulf Stream (Appendix S1, Supporting Information).

Figure 3

Distance of daily locations from tagging location as a function of track duration for 26 shortfin mako sharks tagged off Mexico and the USA. [Colour figure can be viewed at wileyonlinelibrary.com]

Population level UD analysis further demonstrated the low spatial overlap between sharks in the NWA and GMC (Fig. 1). Seasonal UD estimates also indicated two areas of year round concentrated use (e.g. areas within the 50% UD isopleth): the eastern edge of the Campeche Bank off the Yucatan Peninsula, and the US Mid-Atlantic Bight, encompassing the continental shelf, continental slope and northern edge of the Gulf Stream from North Carolina to New Jersey, USA (Fig. 1).

Overall, seasonal variation in mako shark UDs was much more pronounced in the NWA than GMC (Fig. 1). Although the continental shelf in the NWA was occupied year round, core distribution areas (50% UD) shifted north over the course of the year extending from the southeastern USA during the winter to Nova Scotia, Canada in the autumn. Mako shark distribution (95% UD) in the NWA was most restricted during summer and autumn, when sharks were primarily concentrated north of the Gulf Stream from the Carolinas, USA to Newfoundland, Canada. Notably, summer and autumn distributions showed low overlap of the Sargasso Sea, where PP is low compared to northern shelf regions where PP is at its annual peak (Fig. 1, left column). During winter and spring, the distribution expanded considerably southwards, shifting away from the northern shelf regions as sharks travelled more frequently through the southern boundary of the Gulf Stream and the Sargasso Sea (Fig. 2).

Latitudinal range of mako sharks in the NWA peaked in winter, with sharks observed from 49·5°N to 12·2°N. The northern limit of mako shark distribution during winter and spring corresponded with the southward intrusion of cold waters (≤10 ^oC) of the Labrador Current (Fig. 3, right column). Concomitant with their large latitudinal distribution and the various oceanic regions they traversed, NWA sharks occupied a wide range of temperatures (8–30 °C; Fig. 4a). In the winter, for example, individuals north of the Gulf Stream regularly encountered SSTs <14 °C, whereas individuals that moved south through the Sargasso Sea into tropical waters regularly encountered SSTs >26 °C (Fig. 4a).

Figure 4

Daily latitude estimates and associated sea surface temperatures (from satellite remote-sensing data) for shortfin mako sharks tagged off the (a) USA and (b) Mexico. Vertical lines indicate the start of each meteorological season: summer (June–August), autumn (September–November), winter (December–February) and spring (March–May). [Colour figure can be viewed at wileyonlinelibrary.com]

In contrast, mako shark distributions within the GMC showed considerably less variation with no clear seasonal patterns. Similarly, there was little spatiotemporal variability in PP and SST within the GMC compared to the NWA (Fig. 1). Excluding shark #7, which exited the Caribbean, GMC sharks remained in more temporally stable and considerably warmer water throughout the duration of their tracks. As such, GMC sharks experienced a much narrower range of SSTs (21–31 °C), with little seasonal variation (Fig. 4b).

Seven (three female, four male) NWA sharks undertook long-distance, behaviourally distinct, round-trip southern excursions (Fig. 5). These individuals reached latitudes as far south as 9·7°N, with southern journeys terminating across tropical and sub-tropical regions of the western North Atlantic (Fig. 5). These excursions were typically characterized by directionally persistent movements, with relatively short-term residency periods (median = 5 days) at the southern terminus of each excursion (Table S1). Total excursion durations lasted 48–128 days (median = 80·5 days). Of the eight excursions observed, three were initiated in the autumn, three in the winter and one each in spring and summer (Table S1). Travel direction during excursions was largely consistent, with relatively low mean turn angles of 8·8–24°, and 8·9–31·7° for southward and northward trips respectively.

Figure 5

Isolated tracks (daily locations) of round-trip southern excursions by western North Atlantic shortfin mako sharks. [Colour figure can be viewed at wileyonlinelibrary.com]

Discussion

Mako sharks are surface oriented (Vaudo et al. 2016b) making them an ideal shark species to study using fin-mounted satellite tags, which only transmit when exposed to the air. The large number of locations obtained via satellite combined with the longevity of the tracks has resulted in some of the highest quality, long-term tracks of any shark species to date. This data set, therefore, allows the examination of mako shark movements in the North Atlantic at a scale and resolution not described previously.

Although the sharks tagged in this study were primarily juveniles, they represent the typical sizes of mako sharks observed and captured in fisheries in the western North Atlantic (Casey & Kohler 1992). Furthermore, changes in juvenile mako shark survival are predicted to have large effects on population growth rates (Cortés 2002). Consequently, this segment of the population is of key management concern, and elucidating their movement patterns is necessary to inform management applications in this region. Whether the movements of juveniles are representative of the movements of mature mako sharks in this region remains unknown because adults are not often captured and, therefore, there is little information about their movements.

Conventional tag and recapture data suggested that mako shark movements in the western North Atlantic Ocean could be explained by the ‘Sargasso Sea’ hypothesis (i.e., summer residence off the northeastern USA north to the Grand Banks and winter residence in the Sargasso Sea; Casey & Kohler 1992). In addition, this hypothesis was used to partially explain the presence of mako sharks in the GMC, with sharks presumably following Sargasso Sea waters into the GMC (Casey & Kohler 1992). In our study, however, there was little spatial overlap between movements of NWA and GMC sharks; with few exceptions, GMC sharks remained within the GMC and NWA sharks remained within the NWA year round. These observations suggest that although mako sharks occasionally transit between the GMC and NWA, their movements are largely spatially independent. The minimal mixing observed between these two regions suggests that consideration should be given to managing mako sharks in these regions with regionally specific management strategies.

Seasonal movements of NWA sharks were consistent with some, but not all aspects of the ‘Sargasso Sea’ hypothesis. During summer and autumn, NWA sharks were concentrated along the northern portion of the Gulf Stream and along the continental shelf between Cape Hatteras, USA and the Grand Banks. In late autumn and winter, however, most mako sharks abandoned the shelf waters between Cape Cod, USA and the Grand Banks, moving south into warmer waters, occupying a high-use area centred around Cape Hatteras, also noted by Casey & Kohler (1992). Overall, the US Mid-Atlantic Bight, from the mid-shelf to the northern and western edge of the Gulf Stream represented the core area of distribution for sharks tagged in the NWA. In contrast to the ‘Sargasso Sea’ hypothesis, long-term tracking indicated a high degree of fidelity to this region throughout the year.

Despite the large-scale trends in seasonal distributions seen in the NWA, individual movements were highly variable and there was no evidence of coordinated seasonal movements as suggested by the ‘Sargasso Sea’ hypothesis. This individual variability is similar to that seen in studies of juvenile mako sharks in other locations. For example despite undertaking occasional long-distance excursions, mako sharks in temperate regions of the eastern North Pacific and Great Australian Bight largely showed fidelity to regions of high productivity: the California Current large marine ecosystem and Australian shelf habitats respectively (Block et al. 2011; Rogers et al. 2015).

Casey & Kohler (1992) suggested that mako sharks left cooler northern waters to overwinter in more favourable thermal conditions (i.e., ~18 °C) of the Sargasso Sea. However, we found that while mako sharks may traverse through the Sargasso Sea, they do not linger. Lack of residency to the Sargasso Sea may be related to low productivity in these waters, which decreases with increasing distance from the major marginal boundary currents (Longhurst 2007). Thus, while the Sargasso Sea may be thermally suitable for mako sharks, foraging resources may be scarce resulting in a relatively unsuitable habitat for long-term occupation. In addition, given that currents in the Sargasso Sea are relatively weak, mako sharks may be able to travel through the region efficiently using it as a movement corridor.

Overall, NWA sharks exhibited more pronounced seasonal movements and ranged over a much larger geographic area than sharks in the GMC, although GMC sharks that left the eastern edge of the Campeche Bank often returned for the winter and early spring, suggesting possible seasonal cues. Because mako sharks occasionally move between these regions (this study; Casey & Kohler 1992; Kohler et al. 2002), the marked difference we observed in spatial scale of movements between the NWA and GMC is unlikely the result of geographic restriction. More likely, this difference represents behavioural responses to differing environmental variability, a phenomenon also observed in other taxa (e.g., Bonte et al. 2006; Avgar et al. 2013). In the NWA, large-scale physical factors such as climate, irradiance levels, and the Gulf Stream and its associated eddy front system result in a high degree of spatiotemporal variability in environmental conditions, such as water temperature and productivity (Longhurst 2007). This environmental variability leads to resources that are patchy and ephemeral in nature requiring mako sharks to range over a large area to exploit shifting prey distributions. Magnitude of seasonal variation in these abiotic factors generally decreases with latitude, resulting in more stable ecological conditions in tropical and sub-tropical regions, such as the GMC, allowing sharks to stay in profitable areas that are relatively stable in space and time. One such area is the eastern Campeche Bank. As the Yucatan Current flows north, it is forced over the bank creating a persistent and productive frontal zone (Belkin, Cornilon & Sherman 2009) concomitant with the major distributional centre we observed for mako sharks in the GMC. Given that pelagic waters in the GMC are characterized by low PP, the unique physical characteristics of the eastern Campeche Bank area likely create an oasis of consistently concentrated prey in an otherwise unproductive region, as suggested by the large number of fishes that congregate here (e.g., de la Parra Venegas et al. 2011; Cox 2012).

Temperature has been hypothesized as the primary driver of seasonal mako shark movements. Casey & Kohler (1992) suggested water temperature was the key driver of mako shark distribution and seasonal movements in the western North Atlantic, citing a preferred temperature range of 17–22 °C and studies from other temperate regions have supported this preferred range (Abascal et al. 2011; Block et al. 2011; Rogers et al. 2015). This temperature range, however, appears to underestimate the upper and overestimate the lower limits of mako shark thermal preference (Vaudo et al. 2016b). Our data are consistent with a higher upper thermal limit; sharks in the GMC consistently occupied waters with SSTs 22–31 °C. In addition, NWA sharks also made excursions into tropical regions with SST's well above 22 °C.

Despite being associated primarily with warm to temperate waters, some mako sharks spent considerable time in cooler waters during the winter months, with a few individuals observed moving into water with SST <10 °C. Like other lamnid sharks, mako sharks are regional endotherms capable of maintaining core temperatures ~6 to 8 °C above ambient water temperatures (Carey, Teal & Kanwisher 1981), an adaptation that provides physiological benefits to swimming and digestive enzyme performance (Newton, Wraith & Dickson 2015; Watanabe et al. 2015), but may also allow for thermal niche expansion (see Dickson & Graham 2004). An expanded thermal niche could explain the large range of thermal environments mako sharks used. Thus, the observed movements of mako sharks may be a response to the availability of prey resources, rather than temperature alone. In the eastern North Pacific, seasonal movements of mako sharks coincided not only with changing temperature, but also with chlorophyll-a concentrations, a commonly used index of productivity (Block et al. 2011). Similarly, movements of the closely related salmon shark Lamna ditropis in the North Pacific were more strongly correlated with chlorophyll-a concentration and PP than SST (Weng et al. 2008).

Still, thermal conditions will place some degree of physiological constraint on mako sharks. Even in lamnid sharks, lowered body temperatures resulting from entering cooler waters can substantially reduce muscle performance, despite muscles being warmer than ambient conditions (Bernal et al. 2005). Temperature, therefore, may limit access for mako sharks to productive regions like the Grand Banks during colder seasons. Warm temperatures in the GMC may also physiologically constrain mako shark movements. Vertical movement data from the Gulf of Mexico indicated that mako sharks avoided surface waters >27 °C (Vaudo et al. 2016b), which may explain why fewer average daily Argos locations were received from sharks in the GMC than the NWA.

Seven NWA sharks made large-scale round-trip southern excursions; the purpose of which is unclear. Excursions lacked a common destination and timing and were performed by juveniles, ruling out reproductive behaviours as the driver. The short residency at the southern terminus of each trip and termination of trips in areas of relatively low productivity also suggests that these excursions were not directed travel to foraging areas. The directionally consistent and relatively rapid movements associated with these excursions further suggest that mako sharks did not stop to forage while in transit. As such, these excursions may entail a considerable energetic cost, especially if sharks are not regularly foraging. Interestingly, similarly directional, round-trip excursions have been observed in mako sharks in the temperate eastern Pacific Ocean (Block et al. 2011). More work is required to elucidate the ecological significance of this behaviour.

A notable finding was that the mako sharks we tracked passed through the management jurisdictions of at least 17 nations and international waters. These extensive movements expose the sharks to fisheries of multiple nations, highlighting the critical need for closely coordinated, multinational fisheries management. We also found distinct areas of consistent, concentrated use by juvenile mako sharks in both the NWA and GMC, suggesting a specific behavioural attribute of this highly migratory species. In the NWA, core-use areas of the sharks tended to be concentrated in regions off both the US and Canada that are characterized by heavy commercial and recreational fishing (ICCAT 2013), further increasing the likelihood of interaction with fisheries and underscoring the need for internationally coordinated management. It is likely other areas of concentrated use also occur in the North Atlantic and throughout the world's oceans (Block et al. 2011; Rogers et al. 2015; Queiroz et al. 2016). Identifying these areas will be important because mortality of juvenile age classes can have significant population-level ramifications (Cortés 2002) and there is increased potential of overexploitation in these areas.

Capture risk combined with their life-history traits results in mako sharks having one of the highest risks of overexploitation for sharks caught by Atlantic pelagic longline fleets (Cortés et al. 2010). Notably, 22% of the sharks we tagged were harvested by fishers, twice the recapture rate (11·4%) reported by Kohler et al. (2002) for mako sharks in the western North Atlantic tagged with conventional tags, suggesting the stock assessments (ICCAT 2013) from which management decisions are based have underestimated mortality. This difference is likely because assessing capture via satellite telemetry is fisheries-independent, suggesting this technology may be useful for both obtaining more reliable fishing mortality estimates for stock assessments and identifying regions of particular concern for this species. Notably, five of the seven harvested sharks were captured in the high-use areas revealed by the telemetry tracking.

Concluding remarks related to management and conservation

In addition to providing the most comprehensive view thus far of the movement ecology of shortfin mako sharks in the western North Atlantic, our findings also have direct applications to improving management and conservation of this exploited species. First, the large-scale movements of these sharks across multinational jurisdictional boundaries underscore the need for improving the scale of international cooperation if management of this highly vagile species is to be effective. Second, the low level of spatial overlap in mako shark movements between the NWA and GMC suggests that management will be most effective when tailored specifically to these geographic regions. Third, the high-resolution movement data obtained, as well as identification of spatial and seasonal distributions and high-use areas lays the foundation for evaluating potential interactions of mako sharks with fisheries, especially as data from the Vessel Monitoring System becomes available (Queiroz et al. 2016), and will also be useful for incorporation into spatially explicit stock assessment models (Sippel et al. 2015) in the western North Atlantic. Finally, the high proportion of harvested individuals in our study (twice that obtained in a fisheries-dependent, conventional tagging study) raises questions about the true rate of fisheries mortality experienced by this species, calling for a cautionary interpretation of past stock assessments.

Authors' contributions

M.S.S., B.M.W., M.E.B., J.J.V. and G.M.H. conceived the ideas and designed methodology; M.S.S., B.M.W. and G.M.H. collected the data; M.E.B. and J.J.V. analysed the data; J.J.V., M.E.B., B.M.W. and M.S.S. wrote the manuscript. All authors gave final approval for publication.

Acknowledgements

This research was funded by Florida Sea Grant (award FDSP00010205), Swiss Shark Foundation, Guy Harvey Ocean Foundation, Save Our Seas Foundation (award SOSF157) and Virgin Unite. We thank D. Burkholder, G. Jacoski, A. Mendillo, M. Sampson, L. Sampson, C. Donilon and G. Schellenger for assistance during fieldwork.

Data accessibility

Tracking locations are available from Harvard Dataverse https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/MYA8SE (Vaudo et al. 2016a).