Behaviour of leatherback sea turtles, Dermochelys coriacea, during the migratory cycle

Michael C James, Ransom A Myers, C. Andrea Ottensmeyer

Abstract

Leatherback sea turtles, Dermochelys coriacea, undertake broad oceanic movements. While satellite telemetry has been used to investigate the post-nesting behaviour of female turtles tagged on tropical nesting beaches, long-term behavioural patterns of turtles of different sexes and sizes have not been described. Here we investigate behaviour for 25 subadult and adult male and female turtles satellite-tagged in temperate waters off Nova Scotia, Canada. Although sex and reproductive condition contributed to variation in migratory patterns, the migratory cycle of all turtles included movement between temperate and tropical waters. Marked changes in rates of travel, and diving and surfacing behaviour, accompanied southward movement away from northern foraging areas. As turtles approached higher latitudes the following spring and summer, they assumed behaviours consistent with regular foraging activity and eventually settled in coastal areas off Canada and the northeastern USA. Behavioural patterns corresponding to various phases of the migratory cycle were consistent across multiple animals and were repeated within individuals that completed return movements to northern waters. We consider the potential biological significance of these patterns, including how turtle behaviour relates to predator avoidance, thermoregulation and prey distribution.

Keywords:

1. Introduction

Satellite telemetry is now widely used to study the migrations of many marine vertebrates (Le Boeuf et al. 2000; Block et al. 2001; Boustany et al. 2002); however, persistent challenges surrounding long-term instrument attachment and performance normally prevent collection of behavioural data throughout complete migratory cycles. Marine turtles have become popular candidates for satellite tracking studies (Papi et al. 1997; Hays et al.1999; Polovina et al. 2000). Yet, as many species are difficult to find and humanely capture in their oceanic habitat, much of what is known about the large-scale movements of these animals is limited to post-nesting behaviour of mature females tagged on nesting beaches. This is true of the leatherback turtle (Dermochelys coriacea), the largest of all sea turtle species, now globally endangered and facing possible extinction in the Pacific (Spotila et al. 2000). Shelf and slope waters in temperate and boreal regions of the Atlantic support enhanced zooplankton productivity in the summer and fall (Myers et al. 1994; McLaren et al. 2001), including large cnidarian species (e.g. Cyanea capillata and Aurelia aurita) that are prey for leatherbacks (Bleakney 1965; den Hartog & van Nierop 1984; Holland et al. 1990; James & Herman 2001). Seasonal aggregations of leatherbacks in these areas have been verified by aerial surveys (Shoop & Kenny 1992) and fisheries observer data (Witzell 1999). Satellite telemetry suggests that waters off eastern Canada and the northeastern USA constitute high-use habitat for these animals (James et al. 2005).

Recent tracking studies of nesting female leatherbacks tagged in the Caribbean and South America show that many turtles travel northward after nesting (Eckert 1998; Ferraroli et al. 2004; Hays et al. 2004a), presumably to take advantage of high seasonal concentrations of prey. While the diving behaviour of leatherbacks has been described as they displace from equatorial nesting areas (Hughes et al. 1998; Hays et al. 2004b), longer-term movement data have not been available, particularly for those turtles that use northern waters, to enable comparison of behaviour at northern latitudes with return travel to tropical waters. Here we consider the movement, diving and surface behaviour of 25 leatherbacks equipped with satellite transmitters off Nova Scotia, Canada, including 10 that were tracked during round-trip migrations between temperate and tropical waters.

2. Methods

Turtles were captured at the surface in waters off Nova Scotia, Canada, using a breakaway hoop net operated from a 10.5 m commercial fishing boat. Each turtle was guided up a stern ramp on to a raised platform, where curved carapace length (CCL) and curved carapace width were measured, a microchip (AVID brand) was implanted in the right shoulder muscle, and monel tags (style no. 49; National Band and Tag Company, Newport, Kentucky) were applied to the rear flippers. Satellite-linked transmitters integrating time–depth recorders (SLTDRs: models SSC3 and SDR-T16, Wildlife Computers, Redmond, WA, USA) and surface time sensors (KiwiSat 101, Sirtrack Ltd., Havelock North, NZ) were attached to the carapace using a custom-fitted harness made of nylon webbing and polyvinyl tubing, integrating corrodable links to ensure release (Eckert 2002). Turtles were repeatedly doused with buckets of sea water while aboard, and were normally released within 30 min of capture. All procedures were approved by the Dalhousie University Committee on Animal Care and licensed by Fisheries and Oceans Canada.

SLTDRs collected and relayed data on time at depth, time at temperature, maximum dive depth and dive duration (each binned within 14 user-defined data ranges) over 6 h collection periods. Time at depth reflected all time when SLTDRs were submerged, whereas dives were registered only when turtles descended below 4 m (n=12 tags) or 6 m (n=3 tags). While SLTDRs simultaneously record data from different channels (e.g. depth, duration and temperature), data are transmitted in histogram format to increase ease of transfer via the limited bandwidth of the Argos satellite system (Fedak et al. 2002). This decreases the resolution of the data and restricts the types of analyses which can be performed, as the relationship between dive depth, duration and temperature of individual dives is lost. However, patterns of depth use and dive duration can be readily identified and related to the spatial and temporal characteristics of horizontal movements. As our purpose was to identify broad behavioural patterns during the migratory cycle, SLTDR data were considered at the resolution of 24 h rather than 6 h periods.

Satellite transmitters were located with the Argos system (http://www.argosinc.com). Argos assigns location class, an index of positional accuracy, to all derived locations. The analyses presented here used all positions with location classes 3, 2 and 1 (categorized to lie within 150 m, 150–350 m or 350–1000 m, respectively, of the tag's true position). Except where otherwise noted, location classes A, B and 0 (categorized to lie 1000 m or more from the tag's true position) were also used if they yielded rates of travel less than or equal to 5 km h−1, consistent with 99% of rates of travel calculated for this species (James et al. 2005). Positions of location class Z were omitted. From these filtered locations, median daily locations for each turtle were calculated, interpolating positions, assuming constant speed and direction, for days in which no positions were obtained for a given turtle. Rates of travel were calculated between positions of location class 3, 2 and 1 at least 2 h apart.

To evaluate surface behaviour, we considered data from KiwiSat satellite transmitters, which transmit the fraction of each 24 h period that the saltwater switch is dry. These values were matched to median daily locations for each turtle and the median surface time was found for each hexagonal area bin. Medians were chosen so that non-normality of the data would not unduly influence the estimate of the centre of each distribution.

3. Results

Fifteen turtles were equipped with SLTDRs and 10 with KiwiSat satellite tags during summer, 2001–2003; 13 off mainland Nova Scotia (44°N, 64°W) and 12 off Cape Breton Island (47°N, 60°W). Of the 15 equipped with SLTDRs, there were three mature males, nine mature females and three subadults (CCL<140 cm; figure 1). In total, we received 33 171 positions (location class 3: 4.4%, 2: 12.2%, 1: 17.7%, 0: 14.9%, A: 21.7% and B: 29.1%) and kept 77% of the total after filtering. SLTDRs on six turtles transmitted long enough to show round-trip migrations to northern foraging areas. During the migratory cycle, turtles were seasonally resident in northern waters and swam a loop of 6000–12 000 km before returning to forage in continental shelf waters off Canada and/or the northeastern USA.

Figure 1

Tracks of 15 leatherback turtles equipped with satellite-linked time–depth recorders off Nova Scotia, Canada. (a) Mature males, n=3, (b) mature females, n=9 and (c) subadults, n=3. Thin dashed line: 1000 m depth contour; bold dashed line: portion of track when location data not received; bordered dashed line: subadult that entered the Caribbean Sea; x-axis, degrees longitude; y-axis, degrees latitude.

We found consistent patterns of behaviour among all turtles in our sample, which can be used to delineate distinct ‘phases’ of the migratory cycle. Often, changes in multiple measures delineated a shift between phases. To illustrate these phases, we present representative dive data and tracks from two leatherback turtles: turtle A, a mature female (CCL=155.5 cm) tagged in an inter-nesting year, and turtle B, a subadult (CCL=125.5 cm; figures 2 and 3). We present dive data for an additional subadult (turtle C: CCL=134.0 cm) and a mature male (turtle D: CCL=168.5 cm) in the Electronic Appendix (figures S1 and S2).

Figure 2

Tracks throughout the migratory cycle for two leatherback turtles tagged in coastal waters off Nova Scotia, Canada. Phase of the migratory cycle is indicated in top left corner of each panel. Bold line: movements during each phase; thin line: movements from previous phases; dashed line: 1000 m depth contour. Start month and day of each phase are indicated in bottom right corner of each panel. (a) Turtle A: mature female in inter-nesting year. Data to 18 September 2004. (b) Turtle B: subadult. Data to 22 October 2004. x-axis, degrees longitude; y-axis, degrees latitude.

Figure 3

Ambient water temperature and the diving behaviour and latitudinal movements of two leatherback turtles tagged in coastal waters off Nova Scotia, Canada, spanning time from tagging until the day before second migration southward. Left column: turtle A; right column: turtle B. Ticks on time axis represent 14 day intervals. (a), (b) Rate of travel (km h−1). (c), (d) Proportion of time (per 6 h sample) spent in different depth ranges. (e), (f) Proportion of maximum dive depths (per 6 h sample) in different depth ranges. (g), (h) Proportion of dives (per 6 h sample) of varying durations. (i), (j) Proportion of time (per 6 h sample) spent in different temperature ranges. (k), (l) Latitudinal movement. Capital letters indicate phases of the migratory cycle. Vertical lines indicate transitions between phases.

Phase A encompassed movements of turtles in northern shelf and slope waters (principally north of 38°N; figure 2). This phase was characterized by relatively low rates of travel, shallow diving (typically less than 50 m) and short dive durations (typically less than 12 min; figure 3a–h). Shelf waters in this region are generally less than 200 m deep. The slope waters grade from the shelf down to the abyssal plain at 4000 m and deeper.

The onset of phase B was delineated by increased rates of travel, the start of more consistent southward movements and large changes in diving behaviour. After an initial peak associated with departure from northern foraging areas, rates of travel decreased, but were typically higher and, in many turtles, less variable, than they had been during phase A (figure 3a,b). As turtles moved southward, dive depth and dive duration increased and the depths sampled by turtles became bimodally distributed (figure 3c–h). The maximum depths of the majority of dives were less than 6 m or fell within a range that shifted from 4–78 m to 78–252 m (figure 3e,f), which revealed specific intermediate depth ranges that were not targeted by turtles. Occasional very deep dives, exceeding the user-defined depth ranges of the tags (greater than 400 m, n=12 tags; greater than 450 m, n=3 tags), were also recorded during this phase. This increasing bimodality in maximum dive depth with decreasing latitude was also readily apparent in dive duration (figure 3g,h).

Consistent northward movement marked the onset of phase C (figure 2). Rates of travel were similar to those during phase B (figure 3a,b), while the bimodality in maximum dive depth and dive duration decreased with northward movements (figure 3e–h). Therefore, the relationship between diving behaviour and latitude was similar to that in the previous phase.

In phase D, turtle movements generally continued northward towards shelf waters off Canada or the northeastern USA (figure 2); however, there was a drop in average rate of travel and a dramatic change in diving behaviour (figure 3). During this phase, turtles arrived in waters corresponding to the continental slope (figure 2). Maximum dive depth no longer showed a bimodal distribution and was instead relatively uniform between 4 and 154 m (figure 3e,f). Dive duration showed an abrupt decrease and was generally less than 24 min (figure 3g,h).

Phase E encompassed movements primarily on the continental shelf and was marked by even shallower and shorter diving than turtles showed in phase D (less than 50 m, less than 12 min; figure 3c–h). Patterns of movement and diving behaviour for turtles in this northern phase were very similar to those recorded when animals were in phase A, indicating the completion of one migratory cycle and the initiation of a second.

These phases of the migratory cycle were typical both for female leatherbacks in their inter-nesting years and subadults, all of which spent phases B and C in pelagic waters (figure 1b,c) except for one subadult turtle, which entered and exited the Caribbean Sea but did not stop in near shore areas (figure 1c), and one mature female, which was resident in waters off southeastern USA during the first winter post-tagging (KiwiSat transmitter, track not shown). An additional behavioural phase, occurring between B and C, was observed in four mature males in waters adjacent to nesting beaches (see Electronic Appendix; figure S2, turtle D). The dates of transition between phases and the durations of the phases were variable between turtles (figure 4); however, the behavioural patterns within phases (e.g. figures 3, S1 and S2) were similar across turtles.

Figure 4

Timelines of migratory phases for four leatherback turtles equipped with satellite-linked time–depth recorders that completed round-trip migrations to northern foraging areas. Arrows indicate 1 January 2004. Capital letters indicate phase designations for turtle A; sequence is identical for turtles B–D. Turtle A: mature female; turtles B, C: subadults and turtle D: mature male. Turtle D showed additional breeding phase within phase C. Ticks on x-axis, 31 days.

For the 10 turtles equipped with transmitters integrating surface time counters (KiwiSat: one subadult, seven mature females and two mature males), a maximum of 10% of the day was spent at the surface in most of the areas they used (figure 5), with the exception of waters north of 38°N, principally corresponding to phases A, D and E of the migratory cycle, where surface times were highest (maximum 41%). Surface times declined as turtles travelled to lower latitudes (phase B), which is consistent with the increasing dive durations recorded during this part of the migratory cycle.

Figure 5

Time (% of 24 h period) spent at the surface by leatherback turtles equipped with KiwiSat transmitters (n=10). See §2 for calculation details.

4. Discussion

Leatherback turtles tagged on tropical beaches have been recovered thousands of kilometres away (Pritchard 1976; James 2004; Troëng et al. 2004), attesting to their ability to range across vast areas of ocean. By gathering information on the movements and diving behaviour of many individuals of varyied sex and reproductive status, we can begin to understand the biological relevance of these remarkable movements. Turtles in this study have shown movements from the shelf and slope waters of the northwestern Atlantic southward through pelagic waters to tropical waters and back to the north all within one year. Movement and diving behaviour show clear differences between legs of this round-trip journey. However, the biological motivations for these changes in behaviour are not always clear, particularly since many of the animals we tracked were not in their breeding or nesting years.

(a) Northern foraging

We expect that one of the primary determinants of the movements and behaviour of leatherback turtles is the spatial and temporal distribution of their primary prey, the gelatinous plankton, largely of phylum Cnidaria (Bleakney 1965; den Hartog & van Nierop 1984). Unfortunately, only limited information exists on these planktonic species in the areas frequented by the turtles in this study. Instead, most research on the biology of these organisms comes from studies in coastal bays and fjords, while data from pelagic areas are scarce.

The determinants of the timing and size of aggregations of medusae, the familiar free-swimming life stage of jellyfish, are poorly understood but there is general consensus that aggregations can be the result of two main factors: reproduction and physical oceanographic processes (Graham et al. 2001). For scyphomedusae like C. capillata and A. aurita, two common prey species of the leatherback turtle (den Hartog & van Nierop 1984), medusae develop after budding from the benthic sessile life stage over the winter or in the early spring. Through spring and summer, the medusae grow, feeding on zooplankton, and persist for four to eight months before spawning and dying (reviewed by Lucas 2001). Cyanea capillata medusae have been recorded annually in the Niantic River Estuary, Connecticut (USA), from March to late June or early July (Brewer 1989; Brewer & Feingold 1991). However, we have consistently observed leatherbacks feeding on large C. capillata off Cape Breton Island, Nova Scotia, until at least late September. This persistence of C. capillata into the fall is consistent with observations of this species and A. aurita through August and September in fjords in Denmark, Sweden and Japan (Gröndahl 1988; Olesen et al. 1994; Omori et al. 1995).

In general, medusa abundance is lower in pelagic versus coastal areas (Möller 1980; Mills 1995; Lucas 2001), which may reflect lower nutrient availability and greater distances from the coastal benthic life stages, although data in oceanic systems are sparse. However, physical transport of medusae can create local aggregations in pelagic waters, particularly at physical discontinuities such as shelf-breaks and upwelling zones (Graham et al. 2001).

Despite the lack of direct distributional data on gelatinous plankton in areas frequented by turtles in this study, many lines of evidence lead us to suggest that the leatherbacks we tracked use northern shelf and slope waters primarily for foraging. Low rates of travel, previously linked to foraging in other areas (Ferraroli et al. 2004), were observed in phases A, D and E. Moreover, leatherbacks sighted off Atlantic Canada (corresponding to areas frequented in phases A and E) are regularly observed handling jellyfish (Cyanea sp.) in their mouths at the surface (James & Herman 2001). Such prey handling normally involves repeated elevation of the head, which appears to facilitate swallowing (Eisenberg & Frazier 1983; James & Herman 2001). We frequently observed this behaviour, preceded by turtles biting large jellyfish into more manageable pieces (M. C. James, personal observation). While the occurrence of leatherbacks in potential foraging areas may be positively correlated with abundance of jellyfish at the surface (Grant et al. 1996), fieldwork off Nova Scotia has revealed that jellyfish are often not visible at the surface in the vicinity of turtles when prey handling is observed (James & Mrosovsky 2004). Therefore, leatherbacks foraging in shelf waters off Canada and the northeastern USA appear to search for and capture much of their prey at depth (figure 3), before returning to the surface to consume it (James & Mrosovsky 2004).

This pattern of foraging behaviour is consistent with the high proportion of time spent at the surface in northern waters (figure 5). Increased surface time at northern latitudes may also reflect basking, as we have routinely observed turtles resting at the surface during the middle part of the day and evening with both front and rear flippers extended and their heads lowered in the water. This posture, combined with the leatherback's dark dorsal colouring, may maximize absorption of solar radiation, facilitating both digestion and maintenance of body temperature in northern foraging areas where both cold ambient temperatures and consumption of large volumes of cold prey (Davenport 1998) may present thermal challenges. The surface time analysis presented here suggests that northern foraging areas may offer the best opportunities for estimating leatherback abundance from aerial surveys, due to the relatively large proportion of time spent at the surface in these areas.

Leatherback movements during phase D also appear to indicate foraging. Rate of travel dropped markedly from that shown in phases B and C, becoming consistent with rates of travel shown in phases A and E (figure 3a,b). Dive durations decreased and maximum dive depths lost the bimodality so distinctive of phases B and C (figure 3e–h). If indeed these behaviours represent foraging, we suggest that gelatinous prey in these pelagic and slope waters may be distributed at greater mean depth, and perhaps in a greater range of depths, than in the shelf areas further north.

(b) Southern movements

As leatherbacks left northern waters, they showed consistent changes in patterns of depth use, dive duration, rate of travel and time spent at the surface. What cues the onset of southward movements (phase B) is unclear; the departure date is variable among turtles (figure 4). However, in most turtles, it was marked by a rapid increase in rate of travel over the first few days to weeks. As average rates of travel during phases B and C are well above those associated with time spent in northern foraging areas, we expect that turtles are primarily transiting during these phases. While other sea turtles mainly conduct short and shallow dives during open ocean movements (Papi et al. 1997; Hays et al. 1999; Godley et al. 2003), the leatherbacks we tracked spent extended periods both in the uppermost depth bin (0–6 m) and at depths greater than 24 m, undertaking dives among the longest recorded during the migratory cycle (greater than 52 min). The gradual changes in dive duration and dive depth did not appear to be related to water depth, as both continued to increase even after turtles had moved far south of the continental slope and were travelling through areas characterized by relatively uniform bathymetry. Below, we consider alternative hypotheses to explain these changes in diving behaviour.

(i) Predator avoidance

Regular, long, deep diving in migrating green turtles may decrease susceptibility to visual predators such as large sharks by reducing silhouetting against the surface (Hays et al. 2001). Adult leatherbacks are believed to have few natural marine predators and the turtles we studied were all relatively large (125.5–168.5 cm CCL). Rare documentation of predation of leatherbacks by killer whales (Orcinus orca) (Caldwell & Caldwell 1969; Pitman & Dutton 2004) may suggest that this threat influences diving behaviour. We expect that such predation is normally directed at younger, smaller turtles. While the extent of natural predation on adults and subadults is unknown, if predation on these size classes is low, there must be alternative advantages to spending extended periods at depth during migration.

(ii) Foraging

Given the lack of distributional data on leatherbacks' primary prey in open ocean areas, it is difficult to predict how prey distribution might be influencing turtle diving behaviour through phases B and C. Post-nesting female leatherbacks in tropical pelagic waters show diurnal changes in diving behaviour consistent with a response to the distribution of gelatinous prey, which suggests foraging behaviour (Hays et al. 2004b). While moving between temperate and tropical waters, the turtles in this study showed a bimodal distribution of dive depths and durations somewhat similar to that reported by Hays et al. (2004b) and diel dive patterns that may correspond to the diel vertical migrations of their prey (M. C. James, C. A. Ottensmeyer, S. A. Eckert and R. A. Myers, unpublished data). However, our study and that of Hays et al. (2004b) are not strictly comparable due to differences in temporal resolution of the data and geographical zone considered. We also expect that there are large differences in body condition between female turtles that may not have eaten during a two-month nesting period (e.g. Hays et al. 2004b) and turtles that have foraged in northern areas for several months (this study). Indeed, leatherbacks utilizing foraging areas off eastern Canada are 33% heavier than nesting turtles of the same carapace length (James et al. 2005). Therefore, while some opportunistic foraging may occur among turtles departing northern foraging areas, feeding may not be their primary focus at that time. Average rates of travel much higher than those on the foraging grounds suggest that the focus of movements during phase B are primarily related to migration. However, in the southernmost portion of the migratory cycle, reduced rates of travel suggest that some foraging may occur, which is consistent with the interpretation of tropical foraging by Hays et al. (2004b). Moreover, some of the turtles we tracked travelled longitudinally for up to several hundred kilometres before turning north (e.g. figure 1b), which may also indicate foraging in tropical waters. After this brief period, northward travel during phase C revealed similar patterns to behaviour in phase B.

(iii) Seasonal buoyancy changes

Leatherbacks experience dramatic seasonal increases in adipose stores akin to those recorded in many marine mammals. In northern waters, we observe that increases in body fat are most apparent externally at the neck and around the rear flippers and tail, although thickening of the fibrous adipose layer underlying the shell (Goff & Stenson 1988) must certainly also occur. Adipose tissue contributes to buoyancy (Webb et al. 1998; Beck et al. 2000; Biuw et al. 2003); therefore, leatherbacks inhabiting foraging areas in temperate waters will be more buoyant than they are at other times of year and as adipose reserves are depleted during migration (Prange 1976), buoyancy will probably also be reduced.

Other sea turtles can modify their inspired lung volume, an important oxygen store, to adjust their buoyancy during dives (Milsom 1975; Minamikawa et al. 1997; Hays et al. 2000; Hays et al. 2004c) or select specific depths to maintain neutral buoyancy (Minamikawa et al. 2000). In these cases, changes in body condition may influence patterns of dive duration and depth, as has been reported in marine mammals (Webb et al. 1998; Beck et al. 2000; Biuw et al. 2003). In contrast to the hard-shelled turtles, the primary oxygen stores in leatherbacks are in the blood and tissues rather than the lungs (Lutcavage et al. 1992) and little information is available on their buoyancy control. Buoyancy control has been studied in other species of sea turtle (e.g. Minamikawa et al. 2000; Hochscheid et al. 2003; Hays et al. 2004c) and marine mammals (e.g. Webb et al. 1998). Novel approaches will be required before the relationships between body fat, buoyancy, lung volume and diving behaviour can be clarified for leatherbacks.

(iv) Thermoregulation

Increases in dive depth and length during migratory phases may assist with thermoregulation. Among sea turtles, the leatherback has extraordinary lower thermal tolerance limits, conferred by various anatomical and physiological adaptations which function to maintain body temperature while in cold water (Paladino et al. 1990; James & Mrosovsky 2004). In contrast, leatherbacks may face a different thermal challenge in tropical seas: overheating (James & Mrosovsky 2004). While warm core temperatures may increase the capacity for leatherbacks to undertake rapid migrations by enhancing the power output of their muscles, as shown in tuna (Altringham & Block 1997), intense muscle activity combined with relatively high ambient temperatures may require the use of not only physiological mechanisms, including changes in metabolism and blood flow (Paladino et al. 1990), but also behavioural mechanisms to dissipate heat during migration. Therefore, just as ascent to warmer waters following deep dives below the thermocline may serve to warm the core temperatures of some large pelagic fishes (Holland et al. 1992; Cartamil & Lowe 2004), behavioural thermoregulation in leatherbacks may include diving to deeper waters to cool body temperature during periods of elevated activity, such as during migration. The targeted depth would be expected to increase as the water temperature increased with decreasing latitude, as seen in this study. Simultaneous recording of dive depth, ambient temperature and body temperature during both foraging and migration would greatly increase our understanding of potential behavioural thermoregulatory mechanisms used by this species.

(c) Migratory cycle

Satellite telemetry has recently revealed high-use habitat for leatherbacks in waters off eastern Canada and the northeastern USA (James et al. 2005). This investigation into leatherback turtle movements and diving behaviour provides additional evidence that temperate shelf and slope waters of the northwest Atlantic support extensive foraging by adult male and female turtles, as well as subadults.

Leatherback turtles in this study showed round-trip migrations between temperate feeding areas and tropical waters. While leatherbacks have not previously been tracked through round-trip migrations to feeding areas, findings from other studies are consistent with the pattern shown here. Specifically, most leatherbacks tracked from tropical nesting beaches in the western Atlantic swim to temperate latitudes (Eckert 1998; Ferraroli et al. 2004; Hays et al. 2004a), with longer track-lines revealing subsequent movements southward. Other telemetry studies reveal that not all leatherbacks are destined for the northern areas used by the turtles in our study. Post-nesting, some turtles travel eastward, northeast or southward (Eckert 1998; Ferraroli et al. 2004; Hays et al. 2004a,b) to other foraging zones. Regardless of their location, individual fidelity to general foraging areas may be a common phenomenon among Atlantic leatherbacks.

Mature male leatherbacks tagged off Nova Scotia complete round-trip migrations from northern foraging areas to southern, often coastal, breeding destinations, where they can remain resident for up to several months (James et al. in press). A similar pattern is presumably true for females in their nesting years, as animals tagged in Canadian waters have been observed nesting the following spring (M. C. James, unpublished data) and turtles have been captured in Canadian waters within six months of nesting (Goff et al. 1994). Therefore, the movements of mature male leatherbacks and females in their nesting years is consistent with a migratory cycle involving travel between disparate feeding and nesting sites observed in other species of sea turtle (Luschi et al. 1998; Godley et al. 2002).

Our results also illustrate that, with few exceptions, mature females in their inter-nesting years and subadults remain largely in pelagic waters far from shore during the southern portion of their migration. This pattern is particularly intriguing since there is not an obvious reproductive benefit for extensive southward movements for these individuals, in contrast to mature males and females in their nesting years.

One possibility is that this strategy maximizes foraging efficiency. Tropical waters appear to offer some foraging opportunities, consistent with Hays et al. (2004b). An additional profitable zone for northern-foraging leatherbacks may be the pelagic and slope waters traversed by turtles in phase D. In this phase, behaviour consistent with more regular foraging was observed in all tracked turtles. We speculate that in waters off the shelf, blooms of gelatinous plankton may be more ephemeral and more patchily distributed than in shelf waters, but may appear earlier. There is some indication in Europe that A. aurita blooms may appear earlier in more southerly latitudes due to higher ambient temperatures (Lucas 2001). If this is the case in the northwestern Atlantic for this and other jellyfish species, subadults and inter-nesting female leatherbacks may swim southward post-foraging in part to position themselves to exploit emerging prey resources on the way north. Swimming northwards in the spring may allow turtles to opportunistically forage on temperate spring blooms of jellyfish en route to more predictable and abundant prey resources in slope waters and, later, in shelf waters off Canada and the northeastern USA.

For leatherbacks that use northern foraging areas, following this long-distance migratory pattern every year may be a simpler behavioural strategy than modifying the pattern greatly in years when reaching a southern destination is not necessary for reproduction. While the energetic costs associated with northward migration are probably large, our data suggest that these are compensated for by a lengthy, productive foraging period in northern waters. We urge further research into the spatial and temporal distribution of the gelatinous plankton and the diet of leatherbacks so that we may more clearly identify the determinants and constraints of leatherback turtle movements and diving behaviour.

5. Conclusion

Many adult and subadult leatherbacks migrate long distances to temperate waters, where foraging efficiency is enhanced by exploiting prey at readily locatable oceanographic features such as the continental shelf and slope off eastern Canada and the northeastern USA. While mature females return every 2–3 years to tropical nesting beaches and mature males may return annually to breed in the vicinity of these areas, for subadult turtles and females in their inter-nesting years, a return to pelagic habitats in southern waters offers some foraging opportunities and may also serve to position turtles for opportunistic seasonal feeding en route to northern foraging areas. By integrating diving behaviour, horizontal movements and field observations it is possible to identify how turtles use pelagic and coastal areas. The long period of time spent by turtles in foraging areas may make them especially vulnerable to incidental capture in fisheries.

Acknowledgments

We thank S. Eckert (WIDECAST/Duke University) for sharing his expertise in satellite telemetry and instrument attachment, without which such long tracks might not have been obtained, and for his mentorship throughout this study. This research could not have been completed without the dedication of, and logistical support provided by, B. Fricker, H. Fricker and B. Mitchell in the field. We are grateful to C. Harvey-Clark for veterinary advice. Thanks also to K. Martin, J. McMillan, D. Bowen, C. Ryder, R. Merrick, B. Schroeder and two very helpful anonymous reviewers. This research was supported by the National Marine Fisheries Service (USA), Fisheries and Oceans Canada, World Wildlife Fund Canada, Canadian Wildlife Federation, George Cedric Metcalf Charitable Foundation, the Sloan Census of Marine Life and the Natural Sciences and Engineering Research Council of Canada (scholarship to M.C.J. and grants to R.A.M.).

Footnotes

References

View Abstract