Introduction

The loggerhead sea turtle (Caretta caretta) is a highly migratory species that inhabits subtropical and temperate waters of all oceans (Plotkin 2003). After leaving their natal beach, hatchlings move offshore and exploit oceanic habitats, feeding on crabs, molluscs, jellyfish and Sargassum (Dodd 1988; Bjorndal 1997; Bolten 2003a). Following a pelagic stage that may last for a decade or more (Bjorndal et al. 2000), juveniles enter a benthic feeding stage into estuaries, lagoons and other coastal regions where they consume hard-shelled invertebrates (Dodd 1988; Bolten 2003b; Hopkins-Murphy et al. 2003). Some individuals, however, remain in oceanic waters or shuttle between oceanic and neritic areas (Witzell 2002). When sexually mature, turtles move from subadult to adult neritic foraging habitats (Plotkin 2003). A generally accepted life-history model for Atlantic loggerheads proposes that every 2–5 years, female loggerheads make reproductive migrations to breed and nest in the vicinity of their natal beach, and then travel back to their foraging sites, to which they show a high level of fidelity (Limpus et al. 1992; Bolten 2003a; Schroeder et al. 2003; Broderick et al. 2007). These sites can be relatively nearby along a coast or hundreds to thousands kilometres away from each other. However, two studies have reported distinct adult foraging strategies linked to body size, with the larger females foraging in coastal waters and the smaller into oceanic areas (Hatase et al. 2002; Hawkes et al. 2006). In all cases, turtles experience environmental conditions that may influence their migration tracks and habitat use. Moreover, they often face threats (e.g. fisheries bycatch, collisions with boats, red tides) that may dramatically impact their populations. To highlight priorities for conservation, a major need is to know the adults’ migration routes and destinations after leaving their breeding site. While flipper tags (Schroeder et al. 2003) and genetic markers (e.g. Bowen et al. 2004, 2005) provide significant information on population structures and demographic parameters, migration routes can only be assessed through satellite telemetry. Moreover, interpretation of turtles’ tracks can be greatly improved using remotely sensed oceanographic data (e.g. loggerheads: Polovina et al. 2004; Bentivegna et al. 2007; Revelles et al. 2007; green turtles: Girard et al. 2006; Seminoff et al. 2008; leatherbacks: Luschi et al. 2003; Gaspar et al. 2006; Lambardi et al. 2008).

Genetic marker studies revealed that the western North Atlantic is inhabited by multiple genetically distinct loggerhead nesting populations (Encalada et al. 1998; Shamblin 2007). Three occur in the Gulf of Mexico (GOM), including the South Florida Nesting Subpopulation, that ranges from Sarasota County (Florida, USA) to 29°N along the East coast of USA. It is the largest nesting assemblage in the Atlantic Ocean and the second largest in the world (SWOT 2006). With an average of 2,707 loggerhead nests per year (mean value for the 1990–2006 period obtained from the Statewide Nesting Beach Survey program coordinated by the Fish and Wildlife Research Institute; see http://research.myfwc.com/), Sarasota County hosts the most important rookery for loggerheads in the GOM. However, the long-term monitoring of the beaches shows a continuing decline of the number of nests (Witherington et al. 2009). Although nesting beaches are subject to monitoring and conservation effort, adult sea turtles spend most of their time in other areas. Little is known about the migration routes and foraging grounds of loggerheads nesting along the Florida’s Gulf coast. Based upon the returns of tags, foraging areas of adult loggerheads have been reported along the Florida West coast, Cuba and Mexico (Schroeder et al. 2003). Addison et al. (2002) tracked post-nesting movements of loggerheads with sonic and radio telemetry on the southwest coast of Florida, but for a limited period of time (max. 8.25 h). Finally, Timko and Kolz (1982) and Renaud and Carpenter (1994) satellite tracked the movements of adult loggerheads in the GOM, but across feeding grounds in its central and western parts.

The GOM is a complex environment with a large continental shelf (bathymetry < 200 m) that occupies the eastern half of the basin (Gore 1992). In deeper waters, the circulation is characterized by major oceanographic features (Fig. 1). The Loop Current, formed by the northward flow of the Yucatán Current and the eastward flow of the Florida Current, governs the circulation of the eastern Gulf. Its position varies seasonally between an almost direct path to the Florida Current (Fig. 1a), and a long clockwise flow that penetrates deep into the eastern Gulf (Fig. 1b). In the latter case, flow instabilities generate large anti-cyclonic (warm-core) eddies that pinch off from the Loop Current and propagate westward, possibly associated with smaller cyclonic (cold-core) eddies (Coats 1992). Although the Loop Current’s intrusion is most likely to form in the spring, it also occurs during other seasons, for periods varying from 6 to 17 months (Molinari 1980). The water circulation of the eastern boundary of the GOM is driven by the strong Florida Current (Fig. 1), which receives water from the Loop Current and the Antilles Current, and flows through the Florida Straits before reaching Cape Hatteras, where it feeds the Gulf Stream. Knowing how these oceanographic conditions influence loggerheads’ migrations is necessary for understanding their habitat use and improving management and conservation issues.

Fig. 1
figure 1

Main currents in the Gulf of Mexico. a Situation where the Loop Current (LC) reaches the Florida Current (FC) in an almost direct path. b Case where the Loop Current penetrates deep into the eastern Gulf and generates anti-cyclonic eddies

Full size image

By tracking post-nesting movements of loggerhead turtles from Sarasota County, we aimed to provide essential information for the management of the South Florida Nesting Subpopulation. Our goals were (1) to document for the first time the migration routes of females nesting along the Florida’s Gulf coast, (2) to investigate the impact of the major circulation features of the GOM on the migrations using satellite-derived estimation of surface ocean currents.

Materials and methods

Satellite telemetry

Female loggerhead turtles (curved carapace length CCL between 82.2 and 112.0 cm; Table 1) were satellite tracked during the 2005 (n = 5), 2006 (n = 7) and 2007 (n = 15) nesting seasons on Casey Key beach (28.7°N, 82.3°W) in Sarasota County. In May 2006, one additional female (turtle FL06-1) was tagged on Manasota Key, ~15 km south of Casey Key (27.1°N, 82.5°W; Table 1). Loggerhead nesting densities range from 35 to 70 nests/km on Casey Key and 80–160 nests/km on Manasota Key. We approached females as they returned to the sea, corralled the turtle in a portable wooden box, and draped the head with a towel to calm her. Standard carapace measurements were recorded, and two Inconel flipper tags and a PIT tag were applied for each female. We cleaned the carapace of epibiota, abraded the dorsal scutes for a good mounting surface, and wiped with alternating washes of water and acetone to ensure dryness. Turtles were fitted either with a Sirtrack Kiwisat 101 or a Wildlife Computer SPOT5 PTT, which were similar in specification (300–360 g in weight, 2 × C cells, 0.5 W, 30 s repetition rate, saltwater switch). PTTs were adhered with Powerfast slow-curing epoxy and formed into a smooth hydrodynamic shape with the carapace. Ablative antifouling paint was brushed over the epoxy to minimize biofouling. The application process took 1–2 h to complete and turtles swam away immediately. Some of the turtles nested again before a final migration towards their foraging grounds. In the present study, however, we focused on the post-nesting migration only. All tracks were made publicly available and archived on the website http://www.seaturtle.org/tracking and processed with STAT (Coyne and Godley 2005).

Table 1 Main characteristics of the post-nesting movements of the 28 loggerheads turtles
Full size table

Estimating ocean currents

In oceanic areas outside the equatorial band, a large proportion of surface currents can be estimated with satellite-derived oceanographic data by computing the geostrophic and Ekman (wind-induced) components (Pascual et al. 2006). This estimation, however, is less reliable near coasts and in shallow waters across continental shelves, where other processes markedly influence the surface velocities. During their post-nesting migrations, loggerheads usually remain in the upper layer of the water column (Tucker, unpublished data). Therefore, when moving in oceanic areas far from the equator, the currents experienced by the turtles can be estimated as described in Gaspar et al. (2006) and Girard et al. (2006). In short, we computed surface velocity fields on a daily basis and with a resolution of 1/3° on a Mercator projection grid, by combining the geostrophic and Ekman (wind-induced) components. The geostrophic component was calculated as the sum of (1) the mean geostrophic surface currents derived from the mean dynamic topography (Rio and Hernandez 2004) and (2) the geostrophic current anomalies, derived from weekly sea level anomalies gridded fields (product issued from the merged process of all the altimeter data available for the turtle tracking period; see http://www.aviso.oceanobs.com). The Ekman component was derived from daily QuikSCAT scatterometer data of wind stress (see http://www.ifremer.fr/cersat) until February 2007 and from the analysis of the ECMWF weather forecast model afterwards (see http://www.ecmwf.int). Since geostrophic velocities measured between the 31 July 2007 and the 9 August 2007 showed inconsistencies, daily surface currents were not calculated for that period.

Path analysis

Turtles’ movements were tracked using the Argos satellite location system (http://www.clsamerica.com), which classifies fixes into six location classes (LC): 3, 2, 1, 0, A and B. Accuracy estimations are provided for LC 3–0 (<150 m for LC 3, 150–350 m for LC2, 350–1,000 m for LC 2 and >1,000 m for LC 0), but not for LC A and B. The turtles’ migration routes were reconstructed as follows. The apparent speed between successive fixes was computed, and locations leading to values higher than 2.8 m s−1 (10 km h−1) were discarded. To estimate the end of a post-nesting migration, we measured the beeline distance BD between the nesting beach and each turtle’s location. We considered that the migration was completed when the beeline distance reached an asymptote, and used the corresponding fix to define the foraging ground location. Then, the remaining locations were standardized as a fix every 12 h by linear time interpolation, and sea bottom depth was estimated along the track from a 1-min global topography grid derived from Smith and Sandwell (1997); available at http://topex.ucsd.edu/index.html). For each 12-h location, a ground-related track velocity was computed, corresponding to the turtle movement between the present and the following locations. For locations in oceanic waters (bathymetry > 200 m), two additional velocities were calculated: (1) a current velocity, calculated by bi-linear spatial interpolation of the daily gridded surface current fields (see above) at the turtle’s location and (2) a water-related heading velocity, computed by subtracting the current velocity from the track velocity (Gaspar et al. 2006; Girard et al. 2006). The latter represents the average swimming movement of the turtle during the next 12 h. For each turtle’s oceanic journey, a corresponding water-related ‘motor path’ was then reconstructed as a sequence of heading vectors derived from the heading velocities. This path represents the turtle’s locomotion movements only, without considering the drifting impact of the currents.

Many studies reported a strong fidelity of adult loggerheads towards a neritic foraging ground (see “Introduction”). Therefore, we hypothesized that the foraging ground reached by each turtle was actually her goal, and estimated the orientation efficiency of the whole post-nesting migration by calculating the mean cosine of directional errors at a high spatial frequency (Benhamou 2004). In practice, we rediscretised the tracks in a sequence of steps s with a constant length l = 1 km and an orientation θi and for each location (xi, yi), we calculated the cosine of the angular differences between the foraging ground direction γi and the orientation of the following step. The orientation efficiency of the migration track (Ot) was then computed as \( O = \sum\nolimits_{i = 1}^{s} {l\cos (\theta_{i} - \gamma_{i - 1} )/s}. \) To further investigate the impact of the ocean currents on the turtles’ routes and final destinations, the orientation efficiencies of the track (Oot) and motor paths (Oom) sections occurring in oceanic areas only were also calculated using the same procedure. Since turtle FL05-5 was not localized for more than 1 month during her migration, she was discarded from this analysis. Orientation efficiencies for the oceanic journeys of turtles FL07-1 and FL07-2 were not calculated since they occurred at the beginning of August 2007, when the surface currents could not be reliably estimated (see above).

Results

Surface currents

Figure 2 illustrates the main circulation patterns in the GOM at the beginning of the loggerheads’ post-nesting migrations in July 2005 (Fig. 2a), 2006 (Fig. 2b) and 2007 (Fig. 2c). In 2005 and 2006, post-nesting seasons were characterized by a relatively short intrusion of le Loop Current until 24–26°N, and by the presence of a large anti-cyclonic ring at ~25–28°N/87–91°W. In July and August 2007, however, the Loop Current was located slightly deeper into the GOM (until ~27°N) and its associated anti-cyclonic ring as high as 29°N.

Fig. 2
figure 2

Surface currents and sea surface height in the GOM at the beginning of the post-nesting migrations (end of July). a 25 July 2005, b 25 July 2006, c 25 July 2007

Full size image

Migratory patterns

Loggerheads turtles were tracked for a mean of 234 days (SD = 105 days, range: 94–393 days; Table 1). They all initiated their post-nesting movements by mid-August of each nesting season, and reached a neritic foraging ground at the latest in mid-October. Post-nesting migrations averaged 23 days (SD = 16 days, range: 3–68 days; Table 1). Twenty-two individuals remained in the GOM, whereas six moved towards Bahamas (Fig. 3). Five distinct movement patterns were observed, two of which faced the Loop Current or the Florida Current. One pattern consisted in local residence movements (n = 6 turtles; type A3 movements according to the nomenclature of Godley et al. 2008) while the other four were long-distance migrations ending on divergent foraging grounds (type A1 movements in Godley et al. 2008), either in the south-western part of Florida (n = 9 turtles), the Northeast GOM (n = 2 turtles), the South GOM (n = 5 turtles) or the Bahamas (n = 6 turtles). While moving to their foraging site, loggerheads spent 62% of time across the continental shelf of the East and Northeast GOM, 16% in oceanic waters, 12% across the continental shelf of the South GOM (Yucatán, Campeche Bay and Cuba) and 10% across the Bahamas plateau. Once arrived on their foraging site, they were tracked for a mean of 178 days (SD = 93 days, range: 34–332 days; Table 1).

Fig. 3
figure 3

Post-nesting migration routes and foraging areas of 28 loggerhead turtles satellite-tracked in the Gulf of Mexico between 2005 and 2008. a Resident movements, b Migration tracks towards South Florida, c Migration tracks towards the Northeast Gulf of Mexico, d Migration tracks towards the South Gulf of Mexico (Campeche Bay, Yucatán Shelf, Cuba) and Bahamas. Solid contour lines indicate the 60 and 200 m isobaths

Full size image

Local residence movements

Six turtles (CCL ranging between 84.5 and 108.0 cm; Table 1) did not migrate to distant foraging grounds (Fig. 3a). After their last nest was deposited, they moved on the continental shelf for 3–9 days towards foraging destinations relatively close to their nesting colony. Post-nesting movements were then very limited and sometimes difficult to distinguish from the movements at the foraging area (e.g. turtle FL06-6, Fig. 3a). For turtles FL07-6 and FL07-10, the migration consisted in movement offshore from Casey Key until the 60 m isobath, ~130 km from the nesting beach. While the former turtle stayed in a limited foraging area for 308 days, the latter moved along the 60 m isobath for 304 days, until the transmissions ceased. Turtles FL06-6 and FL07-15 moved to a large foraging area ~138 km north from Casey Key and remained there until the tracking ended, 87 and 221 days after arrival, respectively. Turtle FL06-6 was later found dead during a severe red tide. Turtle FL07-8 moved 95 km northward into Tampa Bay, where she was still localized when the transmissions ceased, 34 days after arrival. Finally, turtle FL06-8 moved along the coast 57 km south from her nesting beach and remained in this foraging site for 78 days, until the transmissions failed.

Movements to South Florida

Seven turtles (CCL ranging from 82.2 to 105.6 cm; Table 1) migrated to neritic foraging grounds around the Florida Keys, located ~270 km from the main nesting area of Casey Key (Fig. 3b). In addition, one individual (turtle FL06-7, CCL = 101.2 cm) moved offshore of the Florida Everglades, 211 km from Casey Key, and another (turtle FL07-7, CCL = 105.0 cm) 171 km south-eastward on the continental shelf. Migrations lasted for 7–40 days (Table 1). During their post-nesting movements, all the turtles except FL06-1 remained on the continental shelf and moved in rather straight lines. Surprisingly, turtle FL06-1 (the largest individual), first migrated towards oceanic waters of the central GOM, where she remained for 18 days and moved around an anti-cyclonic (warm-core) eddy originating from the Loop Current (ESM S1). Then, she returned offshore of Sarasota County and finally moved towards the Florida Keys that she reached 40 days after departure from the nesting site. Turtles were tracked on their foraging sites for 59 (turtle FL07-13) up to 309 days (turtle FL07-7) until the transmissions failed.

Movements to Northeast GOM

Two individuals (turtles FL05-4 and FL06-2; CCL equal to 103.0 and 96.2 cm, respectively; Table 1) moved to the Northeast GOM (Fig. 2c). They travelled for 20–22 days before reaching their foraging grounds on the continental shelf, either in Alabama, offshore of Mobile Bay (30.0°N, 88.0°W) for turtle FL05-4, or along the coast of Louisiana, close to the Mississippi River delta (29.0°N, 89.3°W) for turtle FL06-2. These foraging grounds are located 634 and 723 km from Casey Key. While turtle FL05-4 migrated in continental waters only, turtle FL06-2 briefly experienced oceanic conditions before reaching the Mississippi River delta. They were still on these foraging areas when the transmission ceased, 332 and 228 days after arrival, respectively.

Movements to South GOM

Five turtles (CCL ranging from 86.3 to 112.0 cm; Table 1) migrated towards foraging grounds located in the Southern part of the GOM (Fig. 3d). Among them, three individuals (turtles FL05-1, FL05-5, and FL07-1) travelled for 30–68 days and ended their migrations on a neritic area of Campeche Bay (Mexico). Turtle FL07-4 also moved southwest but ended on the Yucatán Shelf (Mexico), 30 days after departure for Casey Key (Fig. 2d). Then, they were tracked on their foraging grounds between 113 (turtle FL05-1) and 267 days (turtle FL07-1; Table 1). During their post-nesting migrations, these four turtles crossed the Loop Current from East to West and appeared to be drifted first by its southward flow, and second by its northward one. This pattern, however, is not visible on the track of turtle FL05-5 because her satellite transmitter stopped emitting just after her departure from Casey Key and resumed when the turtle was on the Yucatán Shelf. Surface currents experienced by the turtles in the Loop Current were >70 cm s−1 (Fig. 4a). The smallest individual of this group, turtle FL07-5 (CCL = 86.3 cm), first migrated in a similar direction as the other individuals. However, when she reached the Loop Current, her track curved south-eastward and finally ended on the northwest coast of Cuba, 13 days after departure (Fig. 2d). The following tracking data indicate that the transmitter was on land in a coastal town for the last 2 months of emission.

Fig. 4
figure 4

Surface current speeds estimated along eight turtle’s track sections in oceanic waters (bathymetry >200 m). a Migrations towards the South Gulf of Mexico; b Migrations towards Bahamas. The highest speed values estimated along each track correspond to the Loop Current and the Florida Current, respectively. The X axis indicates the time spent in oceanic waters

Full size image

Movements to the Bahamas

The remaining six turtles (CCL ranging from 101.3 to 107.8 cm; Table 1) migrated to the Bahamas (Fig. 3d). Apart from turtles FL05-3 and FL07-3 who first entered Charlotte Harbour and moved westward, respectively, all the turtles steadily moved southward to the Florida Keys. Then they entered the Florida Current and migrated north-eastward before moving south-eastward until they reached their foraging ground, 25 days (turtles FL06-3 and FL06-4) to 46 days (turtle FL07-3) after departure from Casey Key. Surface currents estimated along the turtles’ tracks were very strong in the Florida Current (>80 cm s−1; Fig. 4b). It is worth noting that except turtle FL05-3, all the turtles reached very close final residence areas on the Great Bahama Bank, south of Andros Island, ~720 km away from Casey Key (Fig. 3d). Moreover, four turtles (FL05-2, FL06-3, FL06-4 and FL07-3) showed very similar migration patterns from the south-east of Florida, with a clear change of direction as they exited the Florida Current sweeping past the northwestern edge of the Great Bahama Bank (see the 200-m isobath in Fig. 3d), and a final leg along the western side of the bank. The last turtle (FL05-3) took a slightly different route and ended her migration on another foraging ground. When she was in the Florida Current, she moved north-eastward until Grand Bahama Island before shuttling within the archipelago and ending her migration on the Eastern side of the Bahamas, between Wells Cay and Great Guana Cay, at a beeline distance from Casey Key of 647 km. It is worth noting that when she reached the Florida Current, this turtle experienced stronger currents than the other individuals (surface velocities between and 102 and 128 cm s−1; Fig. 4b). All the turtles remained on their foraging grounds until the end of tracking, 71 days (turtle FL05-2) to 275 days (turtle FL07-3) after arrival.

Orientation efficiencies

Except for one notable outlier, all the loggerheads showed rather good orientation efficiencies when moving towards their final foraging grounds (n = 27 turtles, mean Ot = 0.75, SD = 0.14; range: 0.41–0.94; Table 2). The track orientation efficiency of turtle FL06-1, however, was particularly low (Ot = 0.15; Table 2). When looking at the migration sections in oceanic waters, it appears that the motor path orientation efficiency Oom was higher than the track path orientation efficiency Oot in nine out of 11 cases (turtles FL05-1, FL05-2, FL05-4, FL06-1, FL06-2, FL06-3, FL06-4, FL07-3 and FL07-4; Table 2). The low orientation efficiency observed for turtle FL06-1 for the whole post-nesting migration was confirmed for the oceanic journey, both at the track (Oot = 0.07) and the motor path levels (Oom = 0.33). The general tendency of turtles to move and head towards their foraging ground can be highlighted in a chart representing the track and motor oceanic sections with respect to the foraging destination (see Girard et al. 2006 for details). Figure 5 illustrates the cases of two turtles (FL07-4 and FL06-3) which may be taken as examples of the general patterns obtained for the individuals crossing the Loop Current and the Florida Current. It clearly shows the mechanical impact of the Loop Current and the Florida Current: in most cases, the turtles swam towards their final destination (Fig. 5b) but the currents made them take more indirect routes (Fig. 5a).

Table 2 Orientation efficiencies of the entire migration tracks (Ot) and the oceanic track (Oot) and motor path (Oom) sections (see “Materials and methods” for details)
Full size table
Fig. 5
figure 5

Full migration tracks and oceanic motor paths of two loggerhead turtles, expressed with respect to the foraging ground directions (FG). a Whole migration tracks. Dashed sections indicate the oceanic legs, either through the Loop Current (turtle FL07-4) or the Florida Current (turtle FL06-3). b Oceanic motor paths. Axes indicates distance values (km; see Girard et al. 2006 for details)

Full size image

Discussion and conclusions

Movements and foraging destinations

This study reports for the first time migratory patterns and foraging destinations of loggerhead turtles nesting in the Eastern Gulf of Mexico as a component of South Florida Nesting Subpopulation. Our results reveal a large range of post-nesting movements and foraging destinations. For the 28 individuals tracked, 24% remained relatively close to their breeding site, while 52% moved to distant neritic foraging grounds inside the GOM in Southwest Florida, Northeast and South GOM, and 21% to the Bahamas. These results are in accordance with preliminary information obtained from flipper tag recovery (Schroeder et al. 2003). It is noteworthy that no turtle migrated towards the western part of the Gulf (91°W–98°W longitude) or towards the Eastern Florida coast, where adult loggerheads are reported in foraging grounds (e.g. Renaud and Carpenter 1994; McDaniel et al. 2000). Little is known about adult loggerheads’ feeding habitats in the Gulf of Mexico. It would be informative to conduct a mixed stock survey at foraging grounds to determine through genetic studies the nesting subpopulation to which they belong (Bowen et al. 2004; Shamblin 2007).

The spatial scale of the migrations ranged from a few tens of kilometres for the turtles showing a residential behaviour, to more than 1,200 km for turtles moving to Campeche Bay, Mexico (Table 1). Such short- and long-distance post-nesting movements have been observed through tag, satellite tracking and genetic studies in other loggerhead nesting populations, in the Atlantic (Plotkin and Spotila 2002; Hawkes et al. 2007), the Pacific (Limpus and Limpus 2001; Hatase et al. 2002) and Indian Oceans (Papi et al. 1997), and in the Mediterranean Sea (Godley et al. 2003; Broderick et al. 2007; Zbinden et al. 2007).

Navigation abilities

In previous studies, Girard et al. (2006) and Luschi et al. (2007) investigated the homing abilities and navigation behaviour of green turtles (Chelonia mydas). They showed that green turtles experimentally displaced from their nesting site corrected for the passive displacements due to the action of currents, but did not compensate for it by heading (i.e. swimming) in a biased direction such that the resultant movement would point to their goal. Indeed, they appeared to swim more or less accurately towards their goal despite currents taking them along more indirect paths. Contrary to what occurs in homing experiments where a well-defined reference for the turtles’ intended direction is provided, the goal of post-nesting movements is usually not known in advance. In the present study we made the strong assumption that all the turtles tracked reached their initial goal, and calculated the orientation efficiencies accordingly. Except for turtle FL06-1 (Ot = 0.15), and to a less extent for turtle FL05-3 (Ot = 0.41), orientation efficiencies of the entire migration tracks were relatively high (>0.50; mean Ot = 0.76, SD = 0.13), which indicates that turtles moved along rather direct routes. Moreover, in nine of 11 oceanic sections, the orientation efficiency of the motor path was higher than that of the track path (i.e., heading vectors were better oriented towards the final foraging ground than track vectors; Table 2). If all the turtles actually reached their initial foraging goal, then most of our results suggest that like green turtles, loggerheads do not compensate for the current drift.

Our migration data, however, may be analysed in a different way. Indeed, based on the results of several studies (e.g. Lohmann et al. 2001, 2004), we may hypothesize that sea turtle species share similar navigation systems and that since green turtles do not compensate for the current drift (Girard et al. 2006; Luschi et al. 2007), neither do loggerheads. Although such a hypothesis should be first confirmed through well-defined homing experiments, we would like to highlight the potential interest of calculating orientation efficiencies at the track and motor path levels in post-nesting migration studies. Indeed, if we assume that loggerheads do not compensate for the current drift, then we may use orientation efficiencies to determine whether or not migrating turtles reached their expected foraging ground: a migrating turtle ending on the initial goal should show a better orientation efficiency at the motor level than at the track path one. An opposite result would suggest that the currents made the turtle reach a foraging area which was not her initial goal. Under that assumption, we would consider that at least eight of the nine individuals which showed better orientation efficiencies in their oceanic motor path (Oom) than in their corresponding track path (Oot) actually reached their initial goal (Table 2). The very low values obtained for turtle FL06-1 both at the motor and the track paths levels suggest a more complex migration story which is discussed below. At the contrary, the remaining two turtles (FL05-3 and FL07-5) may have reached unexpected foraging grounds. Other considerations suggest that such a hypothesis is plausible (see below).

It is worth noting that in four out of six migrations journeys towards Bahamas waters, critical changes in tracks’ direction occured in the south-east of Florida, close the edge of the Great Bahama Bank. A close look at the track and heading vectors shows that these changes occurred not only at the track path level but also at the motor path one. It means that they were not only due to changes in ocean currents conditions, but also to modifications in the turtles’ swimming directions. Our study had no independent diving behavioural data to corroborate this observation, but the quality and quantity of location fixes substantially improved whenever turtles ceased to move within or across strong current zones. To test the hypothesis of elevated energetic costs while swimming in current would require time depth recorders or satellite relayed archival tags.

Migration strategies

Adult loggerhead turtles are mostly benthic feeders. However, phenotypically linked dichotomy in foraging strategies has been reported for adult loggerheads from Japan (Hatase et al. 2002) and Cape Verde Islands (Hawkes et al. 2006): after completing their nesting cycle, small adult females foraged in oceanic waters, whereas larger turtles moved to coastal areas. In the present study, all the turtles travelled to neritic foraging habitats (defined as<200 m; Lalli and Parsons 1993), although one large individual (turtle FL06-1) first made an excursion in oceanic waters before going to a foraging ground in the Florida Keys. Once she left the continental shelf, she performed one clockwise loop, and then returned back to neritic waters and finally reached the Florida Keys (Fig. 3b). Overlays of the track and ocean currents clearly show that the turtle moved around a large and energetic anti-cyclonic eddy originating from the Loop Current (ESM S1). One may hypothesize that turtle FL06-1 first foraged in deep waters and then in coastal waters.

Herein, loggerheads’ body sizes ranged from 82.2 (turtle FL07-14) to 112.0 cm CCL (turtle FL05-5; Table 1). Among the large individuals (CCL > 90 cm, n = 21), 57% moved to long distance (BD > 600 km) foraging sites in USA (Alabama, Louisiana), Mexico (Campeche Bay, Yucatán Shelf) and Bahamas. The remaining turtles stayed in the Eastern GOM, either close to Casey Key or in South Florida (BD < 300 km). Among the smaller individuals tracked (CCL < 90 cm, n = 7 turtles), 86% foraged along the Eastern GOM, either close to their breeding site (n = 1 turtle) or in the Florida Keys (n = 5 turtles). Only one small turtle (FL07-5; CCL = 86.3 cm) left the Florida Shelf, ending her migration in Cuba (Fig. 3c; see below). Therefore, while large individuals showed various migration strategies, most of the small individuals performed relatively short distance migrations and remained on the continental shelf of the Eastern GOM. One may wonder if this difference in the migration strategies of large and small adult loggerheads is somehow related to the presence of highly dynamic circulation features in the GOM. Current speeds estimated in the Loop and the Florida Currents (typically > 80 cm s−1; Fig. 4) highlight the physical barrier to be crossed by turtles to reach foraging grounds in Mexico and Bahamas. Two interesting cases also tend to show that strong ocean currents constitute major constraints during turtles’ post-nesting migrations. The first case is the migration of turtle FL05-3. Like five other individuals, this turtle moved towards Bahamas but she took a slightly different route and ended her migration on another foraging area of the Great Bahamas Bank (Fig. 3d). Orientation efficiencies of the oceanic track and motor paths may indicate that the turtle did not end on the expected foraging ground (Oot > Oom). Interestingly, that turtle also experienced stronger currents than the other individuals: while the maximum current speed estimated along her track was 128 cm s−1, highest values obtained along the others ranged from 82 (turtle FL07-3) to 108 cm s−1 (turtle FL06-3). Although large (CCL = 106.6 cm), she may have not managed to rapidly cross the particularly strong Florida Current and reach the more quiet waters of the Great Bahama Bank. Nevertheless, this interpretation, should be taken with caution and validated by other tracking data in similar ocean current conditions. The second case concerns the migration of turtle FL07-5. After leaving her nesting site, this small turtle first moved in a similar direction as the other South GOM migrants (Fig. 3c). Whereas the latter were drifted southward and northward by the Loop Current and finally reached the Yucatán Shelf, turtle FL07-5 moved south-eastward and reached Cuba. The higher orientation efficiency obtained for the oceanic motor path than for the corresponding track path (Oom = 0.52, Oot = 0.78) may indicate that due to the currents, the turtle did not reach her initial foraging goal, possibly located in Mexico. This hypothesis is supported by similar satellite-tracking data of female loggerheads migrating from Casey Key to Mexico by passing very close to Cuba (tracks available on http://www.seaturtle.org/tracking; Tucker, unpublished data). Since for turtle FL07-5 all tracking data in Cuba were located on land, they are very unlikely to represent the actual behaviour of that turtle on a foraging ground. It is unknown whether the transmitter detached from the turtle’s carapace and stranded on a beach or whether the animal was harvested.

To summarize, our results suggest that in some cases, the major circulation features of the GOM modify the final foraging destinations of loggerheads nesting in Sarasota County. To succeed in reaching their distant foraging grounds, loggerhead turtles may need to encounter favourable ocean conditions and reach a minimum body size. It would be informative to repeat tracking of the same individuals to test whether they return to the same foraging areas (e.g. Broderick et al. 2007), and compare consecutive migration routes with respect to the ocean current conditions.

Conservation issues

This study shows that loggerheads nesting on Florida’s Gulf coast not only migrate and forage through habitats under the jurisdiction of USA, but also travel to foraging grounds in Mexico, Bahamas and Cuba. These migration patterns and foraging destinations highlight the importance of international management effort to promote the conservation of the South Florida Nesting Subpopulation. Nevertheless, the coast and continental shelf of the Eastern Gulf of Mexico appear as priority areas to reduce anthropogenic threats on adult loggerheads of the GOM. Indeed, the Florida’s Gulf coast concentrates the most important loggerhead rookeries of the Gulf and the Southwest Florida shelf not only encompasses the inter-nesting movements of breeding females, but also a large proportion of their post-nesting migrations and foraging habitats. It is also reported as a shrimping activity area, although most of the shrimp are caught in Louisiana and Texas waters (Magnuson et al. 1990). Since bycatch in shrimp trawl fisheries is a major source of mortality for large juvenile and adult sea turtles (Turtle Expert Working Group 2000), the continental shelf of the Eastern Gulf of Mexico appears as a key area to mitigate the impact of shrimp trawl fisheries on adult female loggerheads of the South Florida Nesting Subpopulation. A separate issue in bycatch of sea turtles now occurs in GOM longlines [http://sero.nmfs.noaa.gov/pr/TurtlesReefFish.htm]. The spatial data discussed herein can also serve as a first step to quantifying the effects of sea turtle interactions in the bottom longline Gulf of Mexico Reef Fish Fishery.