Abstract

Kobayashi, D. R., Cheng, I-J., Parker, D. M., Polovina, J. J., Kamezaki, N., and Balazs, G. H. 2011. Loggerhead turtle (Caretta caretta) movement off the coast of Taiwan: characterization of a hotspot in the East China Sea and investigation of mesoscale eddies. – ICES Journal of Marine Science, 68: 707–718.

Satellite tags were attached to 34 non-reproductive loggerhead turtles (Caretta caretta) caught as bycatch in the Taiwanese coastal poundnet fishery from 2002 to 2008. Transmission durations ranged from 6 to 503 d (median 172 d), with 5860 d tracked in total. Horizontal track data were processed using the Bayesian state-space modelling to extract the most likely daily positions, taking into account ARGOS data quality and other forms of statistical error. A region of high occupancy in the East China Sea, covering 433 549 km2 of coastal and pelagic area next to Taiwan, China, Japan, and South Korea, was characterized from the tracking data. Various attributes of this hotspot are described using satellite tracks and remotely sensed data. The tracks were merged with oceanographic data, emphasizing a new global dataset characterizing mesoscale eddies from satellite altimetry data. A proximity-probability approach coupled with odds ratio testing was used to infer orientation to eddy features. Comparisons against random points, simulated particle tracks, and drifter buoys were used to demonstrate turtle differential responses to eddies inside and outside the hotspot, depending on eddy features (i.e. cyclonic vs. anticyclonic, edges vs. centres). Turtles inside the hotspot utilize fewer strong cyclonic eddy edges than those outside.

Introduction

The open ocean contains a large portion of global habitat, but it remains difficult to survey and sample effectively. Remotely sensed satellite data products offer the unique opportunity to observe large expanses of ocean efficiently over short periods. Remotely sensed data have been used for many applications, including studies related to the management of pelagic marine species. These applications have assisted in improving the efficiency of catching harvested species (Zainuddin et al., 2006), as well as improving the efficiency of protection for species of concern taken incidentally (Howell et al., 2008). There is much potential for remotely sensed data products to assist in the management and recovery of marine species that are protected, endangered, or threatened, because such indirect approaches are non-extractive and non-invasive and can address processes at the requisite spatio-temporal domains of long-lived, wide-ranging species (Druon, 2009). Increased knowledge of their pelagic habitat requirements can be used to minimize harmful anthropogenic interactions that could jeopardize populations.

Loggerhead turtles (Caretta caretta) are one such species of concern whose management could benefit from remotely sensed data. Satellite tags have yielded many insights into their pelagic behaviour and habitat use worldwide. In the Mediterranean Sea, loggerheads in certain areas exhibit a combination of passive drift and site fidelity (Cardona et al., 2005; Revelles et al., 2007), which may be a function of age and/or size (Eckert et al., 2008). In the North Pacific, satellite tags have revealed that many pelagic juvenile loggerheads, after hatching in Japan, have a protracted residence in the Kuroshio Extension Bifurcation Region (KEBR) and eastward adjacent areas of the central North Pacific (Polovina et al., 2000, 2006; Kobayashi et al., 2008), where they forage and interact with pelagic fisheries (Polovina et al., 2004; Kobayashi and Polovina, 2005; Parker et al., 2005). A prominent number of pelagic juveniles is similarly quasi-resident in the productive waters off Baja California, where they feed extensively on pelagic red crabs (Pleuroncodes planipes) and interact with coastal fisheries (Nichols et al., 1999; Koch et al., 2006; Peckham et al., 2007). Both the KEBR and the Baja region are thought to be physical and biological hotspots, where marine predators accumulate (Palacios et al., 2006). Eastward and westward transpacific movements of loggerhead turtles have been documented using conventional and satellite tags (Uchida and Teruya, 1988; Nichols et al., 2000), but the prevalence of this type of directed migration remains unknown. Despite these widespread movements and complex patterns of spatial occupancy, the North Pacific population of loggerhead turtles appears to be a single stock originating from nesting sites in Japan (Bowen et al., 1995; Kamezaki et al., 2003). Loggerhead turtles in the North Pacific are often found in areas not directly in the pathway between these nesting sites and the above-mentioned foraging sites. The reasons for this remain unclear.

In the western Pacific, loggerhead turtles are commonly caught as bycatch in poundnets off the coast of Taiwan (Cheng and Chen, 1997), far southwest of nesting sites in Japan. The Taiwanese poundnet fishery utilizes >100 individual nets moored in shallow water (<20 m deep) along the coastline (Gilman et al., 2010). This type of fishery can exert a significant impact on loggerhead turtle stocks via incidental catch, yet mortality remains low in the Taiwanese fishery (Cheng and Chen, 1997). This appears to result from frequent tending (2–3 times per day) of the gear by fishers, and the shallow depth, coupled with air-accessible catch chambers that allow turtles to breathe (Ishihara, 2007). Taiwan is not known to have any loggerhead nesting sites, and genetic analysis indicates that these individuals are likely from the Japanese nesting stock (P. Dutton, Southwest Fisheries Science Center, pers. comm.). That stock is currently thought to be in a healthy condition (Kamezaki et al., in press), and it is an important component of the worldwide loggerhead population. Any additional sources of mortality require careful examination and mitigation. The purpose of this study was to analyse the movement patterns of satellite-tagged loggerhead turtles taken as bycatch in the Taiwanese coastal poundnet fishery. Remotely sensed satellite data were used to infer patterns of habitat use with respect to regional oceanographic features.

Material and methods

In all, 34 non-reproductive loggerhead turtles were tagged with satellite transmitters after being caught as bycatch in the coastal poundnet fishery off the Pacific coast of Taiwan (Table 1). Catches were made from 2002 to 2008, and animals ranged in size from 64 to 92 cm straight-line carapace length (SCL; 69–95 cm curved carapace length). Several different types of tag were used during the study: Telonics ST-18 (n = 10), ST-14 (n = 2), ST-20 (n = 14), ST-24 (n = 2), and Wildlife Computers SPLASH tags (n = 6). The duty cycle of the tags varied between 12:48 (12 h on, 48 h off) and 6:48 for the Telonics tags, and variable 6:42 for the Wildlife Computers tags. Inconsistencies in the datastreams attributable to this variable duty cycle, i.e. irregular temporal spacing/clumping of data, were resolved using a Bayesian model approach, as described below. Tags were affixed to turtle carapaces using polyester resin and fibreglass cloth (Balazs et al., 1996), and the animals were then released. Positional data were downloaded from ARGOS (http://www.argos-system.org/) and archived locally for processing. The study concluded when all 34 tags ceased transmitting. The study region and adjoining areas are illustrated in Figure 1.

Table 1.

Summary of tracking information for 34 loggerhead turtles (Caretta caretta) caught as poundnet bycatch in the Taiwanese fishery, equipped with electronic tags, and tracked by satellite.

Turtle number SCL (cm)  Date deployed Start latitude (°N) Start longitude (°E) Date terminated End latitude (°N) End longitude (°E) Days tracked Integrated distance (km) Endpoint distance (km) 
T01 71.8 25 January 2007 24.50 121.80 13 July 2007 33.31 123.57 170 5 603.27 992.65 
T02 64.0 2 April 2003 24.49 121.90 16 August 2004 28.32 123.58 503 7 997.52 456.62 
T03 74.5 16 January 2004 24.46 121.94 28 August 2004 32.28 126.98 226 5 809.04 997.47 
T04 67.7 19 February 2003 24.45 122.00 10 August 2003 22.46 117.98 173 4 130.11 465.84 
T05 74.0 22 March 2004 24.47 121.89 27 August 2004 35.11 161.06 158 8 642.19 3 966.55 
T06 72.0 30 November 2004 24.47 121.84 23 May 2005 34.66 145.11 175 5 980.64 2 522.40 
T07 68.7 8 March 2004 24.49 121.81 20 March 2005 22.52 117.86 377 3 905.98 459.12 
T08 72.1 19 April 2004 24.50 121.80 25 June 2004 35.79 151.68 68 4 666.79 3 138.90 
T09 76.5 5 May 2002 24.56 122.00 9 October 2002 32.97 128.22 158 3 453.35 1 112.86 
T10 71.3 21 December 2003 24.50 121.80 31 January 2005 29.32 149.72 408 14 285.62 2 824.89 
T11 74.5 23 December 2008 25.59 122.20 16 February 2009 21.42 115.46 52 1 225.34 829.64 
T12 66.5 4 December 2007 25.38 122.66 1 July 2008 32.46 132.56 210 2 483.99 1 244.03 
T13 74.5 17 October 2006 24.50 121.80 20 January 2007 28.17 130.27 96 3 814.74 938.55 
T14 92.0 19 October 2007 24.51 122.03 5 April 2008 14.47 111.16 169 3 707.65 1 593.81 
T15 82.5 14 January 2008 26.37 125.20 2 August 2008 26.45 127.64 90 2 151.86 243.85 
T16 70.0 18 May 2008 25.94 122.80 9 November 2008 27.32 126.04 174 2 254.59 357.92 
T17 78.0 10 June 2008 25.49 123.30 28 June 2009 29.21 126.53 382 4 929.34 521.97 
T18 69.5 9 February 2007 24.33 122.10 20 April 2007 21.52 116.90 70 2 009.11 617.56 
T19 72.0 14 January 2005 25.51 123.00 8 July 2005 23.09 118.10 173 3 258.40 565.38 
T20 70.0 19 May 2005 24.58 121.90 11 March 2006 30.08 153.28 295 10 960.24 3 164.93 
T21 69.7 2 June 2003 24.50 121.80 7 June 2003 25.80 123.13 205.43 195.97 
T22 74.5 26 June 2005 24.49 121.87 22 August 2006 30.23 125.61 387 5 017.65 736.52 
T23 79.0 21 November 2005 24.49 121.86 16 January 2006 28.59 125.73 57 1 115.18 595.71 
T24 83.0 14 April 2005 24.49 121.81 7 November 2005 27.77 123.66 208 4 674.89 407.11 
T25 67.3 15 March 2005 24.50 121.80 21 November 2005 28.28 122.03 251 5 957.75 419.13 
T26 74.5 21 December 2005 25.23 122.70 7 March 2006 22.46 117.88 77 1 755.23 579.00 
T27 76.0 4 April 2005 24.91 123.70 30 June 2005 28.34 121.83 87 2 214.29 423.35 
T28 80.5 1 February 2005 24.44 121.90 29 May 2005 5.61 114.89 118 5 199.59 2 216.07 
T29 72.5 28 April 2006 24.44 122.59 24 September 2006 37.02 141.75 82 4 139.80 2 304.64 
T30 73.0 10 March 2006 22.41 118.83 4 June 2006 26.38 120.71 73 661.39 478.52 
T31 78.5 17 January 2006 24.56 121.89 17 May 2006 29.68 124.83 62 1 666.05 637.86 
T32 76.0 10 February 2006 24.59 121.50 14 July 2006 2.36 104.41 154 4 968.15 3 079.59 
T33 73.0 13 May 2006 24.64 122.00 8 September 2006 25.13 121.76 60 1 547.36 58.75 
T34 77.0 4 January 2007 25.47 122.40 29 April 2007 27.00 124.50 111 1 459.65 270.37 
Turtle number SCL (cm)  Date deployed Start latitude (°N) Start longitude (°E) Date terminated End latitude (°N) End longitude (°E) Days tracked Integrated distance (km) Endpoint distance (km) 
T01 71.8 25 January 2007 24.50 121.80 13 July 2007 33.31 123.57 170 5 603.27 992.65 
T02 64.0 2 April 2003 24.49 121.90 16 August 2004 28.32 123.58 503 7 997.52 456.62 
T03 74.5 16 January 2004 24.46 121.94 28 August 2004 32.28 126.98 226 5 809.04 997.47 
T04 67.7 19 February 2003 24.45 122.00 10 August 2003 22.46 117.98 173 4 130.11 465.84 
T05 74.0 22 March 2004 24.47 121.89 27 August 2004 35.11 161.06 158 8 642.19 3 966.55 
T06 72.0 30 November 2004 24.47 121.84 23 May 2005 34.66 145.11 175 5 980.64 2 522.40 
T07 68.7 8 March 2004 24.49 121.81 20 March 2005 22.52 117.86 377 3 905.98 459.12 
T08 72.1 19 April 2004 24.50 121.80 25 June 2004 35.79 151.68 68 4 666.79 3 138.90 
T09 76.5 5 May 2002 24.56 122.00 9 October 2002 32.97 128.22 158 3 453.35 1 112.86 
T10 71.3 21 December 2003 24.50 121.80 31 January 2005 29.32 149.72 408 14 285.62 2 824.89 
T11 74.5 23 December 2008 25.59 122.20 16 February 2009 21.42 115.46 52 1 225.34 829.64 
T12 66.5 4 December 2007 25.38 122.66 1 July 2008 32.46 132.56 210 2 483.99 1 244.03 
T13 74.5 17 October 2006 24.50 121.80 20 January 2007 28.17 130.27 96 3 814.74 938.55 
T14 92.0 19 October 2007 24.51 122.03 5 April 2008 14.47 111.16 169 3 707.65 1 593.81 
T15 82.5 14 January 2008 26.37 125.20 2 August 2008 26.45 127.64 90 2 151.86 243.85 
T16 70.0 18 May 2008 25.94 122.80 9 November 2008 27.32 126.04 174 2 254.59 357.92 
T17 78.0 10 June 2008 25.49 123.30 28 June 2009 29.21 126.53 382 4 929.34 521.97 
T18 69.5 9 February 2007 24.33 122.10 20 April 2007 21.52 116.90 70 2 009.11 617.56 
T19 72.0 14 January 2005 25.51 123.00 8 July 2005 23.09 118.10 173 3 258.40 565.38 
T20 70.0 19 May 2005 24.58 121.90 11 March 2006 30.08 153.28 295 10 960.24 3 164.93 
T21 69.7 2 June 2003 24.50 121.80 7 June 2003 25.80 123.13 205.43 195.97 
T22 74.5 26 June 2005 24.49 121.87 22 August 2006 30.23 125.61 387 5 017.65 736.52 
T23 79.0 21 November 2005 24.49 121.86 16 January 2006 28.59 125.73 57 1 115.18 595.71 
T24 83.0 14 April 2005 24.49 121.81 7 November 2005 27.77 123.66 208 4 674.89 407.11 
T25 67.3 15 March 2005 24.50 121.80 21 November 2005 28.28 122.03 251 5 957.75 419.13 
T26 74.5 21 December 2005 25.23 122.70 7 March 2006 22.46 117.88 77 1 755.23 579.00 
T27 76.0 4 April 2005 24.91 123.70 30 June 2005 28.34 121.83 87 2 214.29 423.35 
T28 80.5 1 February 2005 24.44 121.90 29 May 2005 5.61 114.89 118 5 199.59 2 216.07 
T29 72.5 28 April 2006 24.44 122.59 24 September 2006 37.02 141.75 82 4 139.80 2 304.64 
T30 73.0 10 March 2006 22.41 118.83 4 June 2006 26.38 120.71 73 661.39 478.52 
T31 78.5 17 January 2006 24.56 121.89 17 May 2006 29.68 124.83 62 1 666.05 637.86 
T32 76.0 10 February 2006 24.59 121.50 14 July 2006 2.36 104.41 154 4 968.15 3 079.59 
T33 73.0 13 May 2006 24.64 122.00 8 September 2006 25.13 121.76 60 1 547.36 58.75 
T34 77.0 4 January 2007 25.47 122.40 29 April 2007 27.00 124.50 111 1 459.65 270.37 
Figure 1.

Reference map of the study area in the western North Pacific. EEZs are delineated by solid red lines. The white star at the northeast corner of Taiwan indicates the average release location for the 34 tagged loggerhead turtles.

Figure 1.

Reference map of the study area in the western North Pacific. EEZs are delineated by solid red lines. The white star at the northeast corner of Taiwan indicates the average release location for the 34 tagged loggerhead turtles.

Raw ARGOS positional data from the satellite tags were processed using a Bayesian state-space model (SSM; Jonsen et al., 2005). The SSM produced the most likely trajectory through the datapoints, taking into account ARGOS data-quality codes and temporally adjacent positions. The SSM also recast the tracks into daily streams of points, so removing the effects of variable duty cycles and/or intermittent tag-transmission behaviour. These new daily datastreams were archived and plotted using the mapping software Generic Mapping Tools (GMT; Wessel and Smith, 1991). Kernel density was also calculated with the R language subroutine kde2d using the SSM tracks and a smoothing bandwidth (h) of 2.5° longitude and latitude. As a kernel density estimate of habitat utilization can be influenced unduly by potentially arbitrary release locations and possible post-release effects on behaviour, an additional kernel density was performed for a subset of track data after 30 d at liberty (∼17% reduction in the track database, omitting 996 early track days across all 34 individuals). A polygon representing a high-occupancy area was extracted by contouring the output of kde2d with the GMT subroutine grdcontour. The surface area of this polygon (in a spherical context and for marine areas only) was calculated with the R language subroutine areaPolygon, coupled with a high-resolution bathymetric database (Smith and Sandwell, 1997). Daily track locations and all marine regions of the kernel density polygon were tabulated to exclusive economic zones (EEZs) using the EEZ boundary database available from the Flanders Marine Institute at http://www.vliz.be/vmdcdata/marbound/download.php. The track and kernel density locations were assigned to closed polygon regions using a ray-casting algorithm written as a QuickBASIC v4.5 subroutine.

Tracks were first merged to a suite of available oceanographic, bathymetric, and magnetic data products. These included NOAA Pathfinder sea surface temperature (SST), AVISO altimetry products (sea surface height, SSH; geostrophic u-component; geostrophic v-component), SeaWiFS ocean colour, Smith and Sandwell (1997) bathymetry, and earth magnetic-field data from the IGRF-10 model (total force, declination, inclination). Data were examined daily and integrated over the entire track duration by averaging across the daily exposures per individual. These oceanographic data were extracted from weekly (SST, AVISO) or monthly (SeaWiFS, earth magnetic field) grids by interpolation of corresponding trackline values using the GMT subroutine of grdtrack (Wessel and Smith, 1991). Most grids were 0.1° resolution, except for the Smith and Sandwell (1997) bathymetry, which was 0.033° resolution. Trackline aggregated values were examined for evidence of multimodality in an attempt to distinguish significant groupings of turtle behaviour. Some variables of interest (e.g. SST, SeaWiFS) were tabulated over the trackline to compare average conditions inside and outside a region of high occupancy.

The SSM tracks were also merged to a new oceanographic data product that quantifies individual eddies from a time-series of remotely sensed altimetry fields (Chelton et al., 2007, in press). These energetic mesoscale features have scales of tens to hundreds of kilometres and tens to hundreds of days, and account for most of the variability in the ocean; they are one of the primary dynamic features in the ocean, along with large oceanic currents and gyres (Klein and Lapeyre, 2009). The dataset was created with a procedure that automatically processes SSH data from satellite altimeters and tracks eddy trajectories over time and space, using the SSH fields in the AVISO Reference Series, which are constructed from pairs of simultaneously operating altimeters, one in a 10-d repeat orbit (TOPEX/Poseidon followed by Jason-1 then by Jason-2) and the other in a 35-d repeat (ERS-1, followed by ERS-2, then by ENVISAT; Ducet et al., 2000). Eddies were identified with this procedure from closed SSH contours, as described in detail by Chelton et al. (in press) in their eddy documentation. During the period 1992–2008, 176 614 eddy tracks (89 704 cyclonic, 86 910 anticyclonic) with lifetimes of 4 weeks or longer were identified in the global database. Over the domain of the current study (2002–2008, 100–180°E 0–45°N), there were 61 273 eddy tracks (31 292 cyclonic, 29 981 anticyclonic).

In addition to their identification in time and space, eddies can be characterized by attributes such as effective radius, amplitude, edge SSH, and geostrophic rotational speed. The median effective radius of the eddies in the study domain was 95 km (14–378 km). Eddy shapes were approximated as circles using the effective radius, and the daily SSM turtle positions were compared with points along the circumference. Radii at 5° intervals of arc originating in the central locations of the eddies were used for circle construction, and these radii-endpoint locations were used for comparison with the SSM data. The central locations of all eddies were also compared with the SSM data. Eddies were classified as either cyclonic or anticyclonic by the nature of their SSH anomaly (negative SSH anomaly = cyclonic; positive SSH anomaly = anticyclonic) and further classified by eddy strength, as indicated by the ratio of vertical amplitude and the effective eddy radius, which is highly correlated with the maximum rotational geostrophic speed within the eddy. Amplitude values in the dataset were defined as the absolute difference between the SSH contour along the edge of the eddy and the most extreme value within the eddy interior. For the purposes of this study, ratios >0.065 were classified as strong eddies. This value represents the median eddy strength over the entire domain of eddies in this analysis (range 0.0054–1.2913; n = 61 273). Using this merging of datasets, 12 distance-measure metrics were calculated, as summarized in Table 2, reflecting a nested ordering based on eddy type (cyclonic or anticyclonic), strength (any strength or strong), and the feature of interest (eddy edge or centre), respectively.

Table 2.

Summary of distance measures evaluated in this study, with distance measures 1–12 reflecting a nested ordering based on the eddy type (any type, cyclonic, or anticyclonic), eddy strength (any strength or strong only), and the feature of interest (eddy edge or eddy centre), respectively.

Distance measure Eddy type Eddy strength Feature of interest 
Any type Any strength Eddy edge 
Cyclonic Any strength Eddy edge 
Anticyclonic Any strength Eddy edge 
Any type Strong Eddy edge 
Cyclonic Strong Eddy edge 
Anticyclonic Strong Eddy edge 
Any type Any strength Eddy centre 
Cyclonic Any strength Eddy centre 
Anticyclonic Any strength Eddy centre 
10 Any type Strong Eddy centre 
11 Cyclonic Strong Eddy centre 
12 Anticyclonic Strong Eddy centre 
Distance measure Eddy type Eddy strength Feature of interest 
Any type Any strength Eddy edge 
Cyclonic Any strength Eddy edge 
Anticyclonic Any strength Eddy edge 
Any type Strong Eddy edge 
Cyclonic Strong Eddy edge 
Anticyclonic Strong Eddy edge 
Any type Any strength Eddy centre 
Cyclonic Any strength Eddy centre 
Anticyclonic Any strength Eddy centre 
10 Any type Strong Eddy centre 
11 Cyclonic Strong Eddy centre 
12 Anticyclonic Strong Eddy centre 

Some discussion of data resolution is appropriate here, because we are attempting to resolve differences of the order of tens of kilometres to determine whether turtles prefer edges or centres of eddies. Most satellite-altimetry data products have an approximately weekly sampling cycle and tracklines ∼100–300 km apart. Trackline data are then processed with sophisticated algorithms to resolve mesoscale processes with wavelength scales of ∼200 km, which corresponds to an eddy radius of ∼40 km (Chelton et al., in press). The SSM daily positions include the Bayesian credible intervals in both latitudinal and longitudinal directions. The average 95% interval was 0.64° (∼71 km) around latitude and 0.73° (∼70 km) around longitude. If an actual turtle position was in maximal error, it would still likely be closer to the edge or centre of the eddy as originally measured. Therefore, loggerhead turtle positions were known with reasonable certainty with respect to mesoscale eddy dimensions and feature locations.

Distance measurements were calculated for all daily positions along the SSM tracks. All daily positions within 1° latitude/longitude were tabulated as being in proximity to the eddy feature of interest. These tabulations were converted to proximity probabilities, and comparisons were made using odds ratio tests. Two subsets of the SSM track data were examined separately for those daily positions found within a high-occupancy region and those found outside it.

Proximity probabilities to eddy features were calculated for three sources of spatial data in addition to the SSM daily turtle positions. First, a set of random-walk tracks was examined. These were essentially a sequence of random points in space constrained to span the study domain. A sample of 100 tracks was examined. Second, a set of simulated passive-particle tracks was generated and analysed. These particles were driven by NOAA OSCAR (Ocean Surface Current Analyses-Real Time) current fields, estimated using data from satellite altimeters and scatterometers. The movement of particles was simulated using a Lagrangian advection-diffusion model (Polovina et al., 1999). Currents were matched spatially and temporally to that of the domain of this analysis, with releases from the same average release location of the bycatch turtles. Time at liberty was set at 1 year, with 50 replicates released at the beginning of each of the study years 2002–2008. Third, data from drifting buoys deployed 15 m deep (referred to as surface drifters) that were encountered in the spatial and temporal domain of this analysis (n = 1291 buoys with 32 963 individual locations) were examined. Surface drifters are routinely deployed by research vessels and vessels of opportunity as part of a programme studying global circulation patterns (Lumpkin and Pazos, 2006). The surface drifter dataset used in this analysis is administered by the Global Drifter Program at NOAA (http://www.aoml.noaa.gov/phod/dac/index.php).

As these three additional sources of spatial data were non-sentient, the a priori hypothesis was that none of the three tests would detect patterns resulting from active orientation and swimming behaviour. However, if there were passive accumulation or dispersal processes within the eddy features (e.g. convergence or divergence), such processes may manifest themselves as significant test results in the examinations of proximity probabilities. Comparison of the SSM test results and these additional applications might also yield insights into the underlying mechanism of eddy utilization patterns by loggerhead turtles and the role of passive vs. active orientation. A matrix of odds ratio tests was applied, using the reduced-degrees-of-freedom approach (Kobayashi et al., 2008) to alleviate concerns of pseudoreplication (Hurlbert, 1984). Pseudoreplication is a potential issue, because of the lack of independence among the many days of track data.

The array of proximity probabilities for the two subsets of turtle tracks, the random points, the simulated particles, and the surface drifters were used to characterize patterns of track behaviour using non-metric multidimensional scaling (NMDS) ordination. This attempted to distinguish significant groupings of turtles based on their utilization of the high-occupancy area and the other types of spatial track. The NMDS was done with the software package PC-ORD (McCune and Grace, 2002).

Results

The 34 tags transmitted for an average of 172 d (range 6–503 d), providing 5860 individual days tracked before transmissions ceased. The Bayesian SSM was successfully applied to the 34 tracks and the data are plotted in Figure 2. Some turtles moved northeast into the KEBR; others moved southwest. However, the dominant pattern was a quasi-resident behaviour in a region between Taiwan, China, Japan, and South Korea. Loggerhead turtles spent portions of their track time in 12 EEZs (Figure 3). The primary EEZ was China (28.9%), with Japan, Taiwan, and the Conflict Zone being the only other EEZs with at least 10% occupancy. High seas locations represented 12.5% of the total track time.

Figure 2.

Individual loggerhead turtle tracks using the same map coordinates as in Figure 1. Refer to Table 1 for summary information corresponding to individual tags T01–T34. The number below the tag designation is the number of days at liberty with successful tag transmission. Continents and other landmasses are outlined in grey. Black star, starting location; black circle, last transmission location.

Figure 2.

Individual loggerhead turtle tracks using the same map coordinates as in Figure 1. Refer to Table 1 for summary information corresponding to individual tags T01–T34. The number below the tag designation is the number of days at liberty with successful tag transmission. Continents and other landmasses are outlined in grey. Black star, starting location; black circle, last transmission location.

Figure 3.

Loggerhead turtle percentage occupancy of different EEZs, all loggerhead turtle tracks combined (2002–2008). The secondary chart on the right is a further breakdown of the “other” category on the primary chart, representing an aggregate 4.1% of the total days.

Figure 3.

Loggerhead turtle percentage occupancy of different EEZs, all loggerhead turtle tracks combined (2002–2008). The secondary chart on the right is a further breakdown of the “other” category on the primary chart, representing an aggregate 4.1% of the total days.

The kernel density distribution for all tracks combined is displayed in the lower panel of Figure 4. This distribution suggests a localized pattern of occupancy in the East China Sea around Taiwan, China, Japan, and South Korea. However, that distribution (i.e. Figure 4, lower panel) also assimilates release locations, which were all similar and located in proximity to the area of occupancy. Post-release behaviour, e.g. lethargy or flight, may have an undesirable effect on the analysis. The upper panel of Figure 4 displays the kernel density distribution using track data after only 30 d of liberty. This suggests that the observed pattern is not heavily influenced by the similarity and/or proximity of release locations to the area of occupancy. It is also likely, although not quantified, that turtle behaviour was not significantly altered because of the tagging and release process. A polygon enclosing the kernel density distribution at values >0.005 encompasses 433 549 km2 of ocean, after correcting for intersections with land features (Figure 4, upper panel). The area spans diagonally along the Taiwan Strait between Taiwan and China and extends northeast into the East China Sea, starting at 116°43'E 21°35'N and ending at 128°17'E 33°19'N. The depth there ranges from 0 to 3117 m, with an average of 195 m. Similar to the tracking breakdown, the bulk of the East China Sea kernel density distribution is situated in the Chinese EEZ (40.73%), with the Joint Japan Korea Area (15.73%), Taiwan (14.48%), Japan (13.29%), Conflict Zone (11.81%), and South Korea (3.97%) also represented (Figure 5).

Figure 4.

Kernel density distribution for all loggerhead turtle tracks combined (2002–2008), using a smoothing bandwidth (h) of 2.5° longitude and latitude. (Upper panel) Kernel density distribution using only track data after the initial 30 d of each of the tagged individuals; (lower panel) kernel density distribution for all days.

Figure 4.

Kernel density distribution for all loggerhead turtle tracks combined (2002–2008), using a smoothing bandwidth (h) of 2.5° longitude and latitude. (Upper panel) Kernel density distribution using only track data after the initial 30 d of each of the tagged individuals; (lower panel) kernel density distribution for all days.

Figure 5.

Kernel density distribution for different EEZs, excluding land and only considering the breakdown of the marine portion of the hotspot.

Figure 5.

Kernel density distribution for different EEZs, excluding land and only considering the breakdown of the marine portion of the hotspot.

Individual mesoscale eddy tracks identified over the domain of this study are displayed in Figure 6, including a snapshot of eddy activity at the time when the maximum number of tagged turtles were at liberty (n = 7; 25 May 2005). Eddy activity was evident over the entire spatial and temporal domain.

Figure 6.

Altimetric eddy tracks over the entire study domain (upper) and a specific example of eddy size (circle size) and location on 25 May 2005 (lower). For both panels, blue represents cyclonic eddies, and red shows anticyclonic eddies. In the lower panel, thicker circles distinguish strong (the ratio of amplitude and radius >0.065) from weaker eddies. Black stars indicate the location of seven loggerhead turtles at liberty as this manuscript was being prepared.

Figure 6.

Altimetric eddy tracks over the entire study domain (upper) and a specific example of eddy size (circle size) and location on 25 May 2005 (lower). For both panels, blue represents cyclonic eddies, and red shows anticyclonic eddies. In the lower panel, thicker circles distinguish strong (the ratio of amplitude and radius >0.065) from weaker eddies. Black stars indicate the location of seven loggerhead turtles at liberty as this manuscript was being prepared.

Proximity probabilities are presented in Figure 7, and the results of the individual odds ratio tests in Table 3. An odds ratio test value is an indication of the odds differential between the two categories being compared for a particular attribute; hence, the test indicated that the odds of being in proximity to a strong cyclonic eddy edge were approximately ninefold greater for turtle tracks outside, than for those inside, the hotspot. There were also significant results for strong eddy edges, centres, and cyclonic centres. Turtle tracks outside the hotspot did not have significantly higher odds of being in proximity to eddy features than the three non-sentient spatial data sources. Tracks inside the hotspot differed from the three non-sentient spatial data sources with respect to proximity to strong eddy edges and strong cyclonic edges, revealing much lower odds of proximity.

Table 3.

Summary of odds ratio tests of turtle track proximity probabilities comparing two subsets (outside and inside the hotspot) and with the three non-sentient spatial data sources (random points, simulated particles, and surface drifters), with the statistical test showing either not significant (n.s.) or significant at p = 0.05 (*).

 Odds ratio test result and significance
 
Proximity Outside: Inside Outside: Random Outside: Simulation Outside: Drifters Inside: Random Inside: Simulation Inside: Drifters 
Any eddy edge 2.23 n.s. 1.08 n.s. 0.94 n.s. 0.67 n.s. 0.48 n.s. 0.42 n.s. 0.30 n.s. 
Cyclonic eddy edge 2.61 n.s. 1.21 n.s. 1.09 n.s. 0.84 n.s. 0.46 n.s. 0.42 n.s. 0.32 n.s. 
Anticyclonic eddy edge 1.77 n.s. 0.88 n.s. 0.82 n.s. 0.64 n.s. 0.50 n.s. 0.46 n.s. 0.36 n.s. 
Any strong eddy edge 4.68* 1.90 n.s. 1.05 n.s. 1.21 n.s. 0.41 n.s. 0.22* 0.26* 
Strong cyclonic eddy edge 9.05* 2.24 n.s. 1.16 n.s. 1.53 n.s. 0.25* 0.13* 0.17* 
Strong anticyclonic eddy edge 2.54 n.s. 1.32 n.s. 0.90 n.s. 0.89 n.s. 0.52 n.s. 0.35 n.s. 0.35 n.s. 
Any eddy centre 1.86 n.s. 1.20 n.s. 1.23 n.s. 0.87 n.s. 0.65 n.s. 0.66 n.s. 0.47 n.s. 
Cyclonic eddy centre 1.89 n.s. 1.47 n.s. 1.62 n.s. 1.07 n.s. 0.78 n.s. 0.86 n.s. 0.56 n.s. 
Anticyclonic eddy centre 1.62 n.s. 0.91 n.s. 0.87 n.s. 0.69 n.s. 0.56 n.s. 0.54 n.s. 0.43 n.s. 
Any strong eddy centre 4.28* 1.93 n.s. 1.47 n.s. 1.31 n.s. 0.45 n.s. 0.34 n.s. 0.31 n.s. 
Strong cyclonic eddy centre 5.67* 2.62 n.s. 1.89 n.s. 1.75 n.s. 0.46 n.s. 0.33 n.s. 0.31 n.s. 
Strong anticyclonic eddy centre 2.50 n.s. 1.14 n.s. 0.96 n.s. 0.81 n.s. 0.45 n.s. 0.38 n.s. 0.33 n.s. 
 Odds ratio test result and significance
 
Proximity Outside: Inside Outside: Random Outside: Simulation Outside: Drifters Inside: Random Inside: Simulation Inside: Drifters 
Any eddy edge 2.23 n.s. 1.08 n.s. 0.94 n.s. 0.67 n.s. 0.48 n.s. 0.42 n.s. 0.30 n.s. 
Cyclonic eddy edge 2.61 n.s. 1.21 n.s. 1.09 n.s. 0.84 n.s. 0.46 n.s. 0.42 n.s. 0.32 n.s. 
Anticyclonic eddy edge 1.77 n.s. 0.88 n.s. 0.82 n.s. 0.64 n.s. 0.50 n.s. 0.46 n.s. 0.36 n.s. 
Any strong eddy edge 4.68* 1.90 n.s. 1.05 n.s. 1.21 n.s. 0.41 n.s. 0.22* 0.26* 
Strong cyclonic eddy edge 9.05* 2.24 n.s. 1.16 n.s. 1.53 n.s. 0.25* 0.13* 0.17* 
Strong anticyclonic eddy edge 2.54 n.s. 1.32 n.s. 0.90 n.s. 0.89 n.s. 0.52 n.s. 0.35 n.s. 0.35 n.s. 
Any eddy centre 1.86 n.s. 1.20 n.s. 1.23 n.s. 0.87 n.s. 0.65 n.s. 0.66 n.s. 0.47 n.s. 
Cyclonic eddy centre 1.89 n.s. 1.47 n.s. 1.62 n.s. 1.07 n.s. 0.78 n.s. 0.86 n.s. 0.56 n.s. 
Anticyclonic eddy centre 1.62 n.s. 0.91 n.s. 0.87 n.s. 0.69 n.s. 0.56 n.s. 0.54 n.s. 0.43 n.s. 
Any strong eddy centre 4.28* 1.93 n.s. 1.47 n.s. 1.31 n.s. 0.45 n.s. 0.34 n.s. 0.31 n.s. 
Strong cyclonic eddy centre 5.67* 2.62 n.s. 1.89 n.s. 1.75 n.s. 0.46 n.s. 0.33 n.s. 0.31 n.s. 
Strong anticyclonic eddy centre 2.50 n.s. 1.14 n.s. 0.96 n.s. 0.81 n.s. 0.45 n.s. 0.38 n.s. 0.33 n.s. 
Figure 7.

Proximity probabilities for two subsets (outside and inside the East China Sea hotspot) of loggerhead turtle tracks, random points, simulated particles, and surface drifters. Each bar represents the proportion of days where the position of the object was within 1° of the latitude/longitude of a particular eddy feature.

Figure 7.

Proximity probabilities for two subsets (outside and inside the East China Sea hotspot) of loggerhead turtle tracks, random points, simulated particles, and surface drifters. Each bar represents the proportion of days where the position of the object was within 1° of the latitude/longitude of a particular eddy feature.

The NMDS-ordination results are summarized in Figure 8. The loggerhead turtle tracks inside the hotspot were clearly responding differently from those of turtles outside the hotspot or the three other sources of spatial data. Interestingly, the turtles outside the hotspot were most closely associated with the simulated particle tracks, whereas the random points and the surface drifters formed another clustering in the NMDS.

Figure 8.

NMDS ordination of proximity probabilities for two subsets of loggerhead turtle tracks and three non-sentient spatial data sources.

Figure 8.

NMDS ordination of proximity probabilities for two subsets of loggerhead turtle tracks and three non-sentient spatial data sources.

Discussion

Differential utilization of certain portions of mesoscale eddies by loggerhead turtles is intuitively easy to understand from an ecological perspective. There may be mechanisms that result in apparent attraction or repulsion from such features passively, cues that lead the turtles to orientate to them actively, or a combination of these factors. Conventional eddy pumping theory (McGillicuddy et al., 1998) suggests that primary dynamics are in the centre of the eddy, where vertical velocities and biological responses may be greatest. Many studies have documented strong primary production patterns related to eddy dynamics (Vaillancourt et al., 2003). Some empirical research has indicated that cyclonic eddy centres and anticyclonic eddy edges have more biological productivity (Mizobata et al., 2002). Some of these findings are consistent with application of the standard omega equation (Hoskins et al., 1978) to estimate vertical velocities when the three-dimensional density field and horizontal velocities are known (Klein and Lapeyre, 2009). Doming of the pycnocline or thermocline would be most pronounced at the centre and could reflect a different thermo-saline habitat. The eddy perimeter, where the geostrophic rotational velocities peak, likely functions as an ecotone between distinctly different water masses/habitats and may have an ecological function not easily delineated by its measurable attributes.

There could be passive accumulation in regions of convergence or downwelling. A floating object entrained in such a flowfield would be retained over regions of downwelling, because it cannot be advected downwards, though further horizontal movement would be dampened by the converging flowfields. This mechanism is known to accumulate floating debris in Langmuir cells and surface slicks generated by internal waves and other subsurface phenomena (Shanks and Wright, 1987). A turtle in a passive floating mode of transport or basking (Bentivegna et al., 2007; Hochscheid et al., 2010) could become entrained in eddy regions of convergence, most likely located near the centre. This would then likely keep the turtle near the centre, although also transporting it horizontally in synchrony with the typically westward eddy propagation. Such rafting could be an energy-efficient means of moving in any direction across an ocean basin.

Active orientation to eddy features could be linked to many types of behaviour, such as targeting locations for foraging, social/reproductive opportunities, and thermal habitat. Alternatively, there may be no proximate cause beyond the innate curiosity common in animals (Pisula, 2009); a turtle may simply be exploring a perceived discontinuity in an otherwise homogenous pelagic environment. Mesoscale eddies are more physically and biologically dynamic than regions without such features. Their potential use as waypoints for navigation or “meeting places” for social facilitation should also not be discounted; similar to how fish aggregating devices influence pelagic fish species (Fréon and Dagorn, 2000; Castro et al., 2001).

The differential response by loggerhead turtles based on their time tracks inside the hotspot is perplexing, but suggestive of aversion behaviour. There might be preferred areas >1 km from the eddy edge or quasi-stationary eddy features correlated with other important ecological considerations. Such a differential response warrants further investigation. As indicated above, eddies were present throughout the spatial and temporal domain of the study area, but the spatial patterning of eddy fields may need further attention. Eddy density is obviously lower in the coastal and shallow regions of the domain (Figure 6), and there may be other eddy-size-related or eddy-type-related distributional patterns which relate to this finding. An alternative approach to examining orientation to eddy features is being developed and will involve temporal randomization (eddy field randomization). This should preserve spatial patterns of the eddy field distribution and abundance, but still allowing quantitative investigation of orientation behaviour to specific parts of eddies based on temporal match–mismatch. This and partitioning or constraining of the non-sentient spatial datasets to regions inside and outside the hotspot are also topics for future investigation.

Enhanced foraging opportunity appears to be a likely mechanism for the prolonged residence in waters north of Hawaii (Polovina et al., 2004; Parker et al., 2005). These productive waters of the Subtropical Frontal Zone (Roden, 1980) and the Transition Zone Chlorophyll Front (Polovina et al., 2001) contain many frontal features that may aggregate forage items. Mesoscale eddy activity in these areas is correlated with diving behaviour, suggesting that subsurface foraging may be important (Howell et al., 2010). Most pelagic diet studies have indicated that loggerhead turtles forage on gelatinous zooplankton or other slow-swimming or sessile pelagic invertebrates. Benthic foraging is also common in coastal areas or when the bottom depth is within diving range (Plotkin et al., 1993). Opportunistic foraging on fishery discards has also been documented (Seney and Musick, 2007; Wallace et al., 2009). Pelagic forage could accumulate in regions of oceanographic convergence; if these regions were heavily fished or quasi-resident over shallow continental shelf areas, there could be local enhancement of other types of forage (benthic organisms or fishery discards). Extended residence by loggerhead turtles off the coast of Baja California appears to be related to the patterns of abundance of pelagic red crab (P. planipes), a forage species (Nichols et al., 1999; Peckham et al., 2007). The distribution and abundance of appropriate forage items in and around these mesoscale eddies and eddy location in time and space are important topics for future work. Particular attention needs to be paid to quasi-stationary eddy features over shallow continental shelves. Gelatinous zooplankton and biota inhabiting floating objects are often overlooked in ecosystem surveys, because of sampling difficulties or a focus on detrimental effects of floating objects, such as marine debris.

The area of high residence in the East China Sea continental shelf appears to be a localized region of biological activity similar to hotspots seen in the KEBR (Polovina et al., 2000, 2006) and off the Baja California coastline of Mexico (Nichols et al., 1999; Koch et al., 2006; Peckham et al., 2007). This region is of great interest because of its large size and proximity to the shore. The hotspot also straddles several EEZs, with the principal area in Chinese waters (Figure 5). There are also jurisdictional issues to resolve in the area, considering that two prominent locations within this hotspot lie in contested waters of the Conflict Zone or the Joint Japan Korea Area under joint sovereignty. Notably, 15 of the 34 turtles spent at least 90% of their time in the hotspot, accounting for 51% of the total track days (Figure 9). This behaviour was most clearly displayed by loggerhead T02 (Figure 1), which spent more than 500 d and all its time there. Clearly, this underscores the need to examine behaviour on an individual level, because it is difficult to characterize an average behaviour with this dataset.

Figure 9.

Individual turtle hotspot occupancy patterns, each bar corresponding to a tabulation of the 34 tracks into bins of 10%.

Figure 9.

Individual turtle hotspot occupancy patterns, each bar corresponding to a tabulation of the 34 tracks into bins of 10%.

Some sections of the hotspot area are on the continental shelf next to the Yangtze River, which could be an important foraging area. The dynamics of the region where the Yangtze River plume meets the Kuroshio Current intrusion are complex (Du et al., 2000); these intersecting water masses are likely a primary factor for the region's high productivity (Liu et al., 2003; Tseng and Shen, 2003). The seabed there is shallow enough for benthic foraging, yet also contains much eddy activity (Figure 6). Comparison of chlorophyll a data with SSM tracks inside the hotspot indicated that productivity there was nearly twice as high as in waters encountered by turtles outside the hotspot (Figure 10). The area is also intensively fished, primarily by boats from China. The incidental or targeted take of loggerhead turtles by these and other fisheries over the continental shelf is largely unknown and requires additional investigation. Loggerhead turtle diet and community structure of both the benthic and pelagic habitat are not well understood in this hotspot and require more study. It seems likely that a wide spectrum of marine predators converge in the region where the Yangtze River plume meets the Kuroshio Current intrusion, so future studies should concentrate on a wide array of species in its ecosystem.

Figure 10.

Average chlorophyll a concentration encountered by loggerhead turtles inside and outside the East China Sea hotspot. Error bars represent 95% confidence intervals.

Figure 10.

Average chlorophyll a concentration encountered by loggerhead turtles inside and outside the East China Sea hotspot. Error bars represent 95% confidence intervals.

To conclude, a hotspot region of high occupancy by loggerhead turtles was characterized in the East China Sea between Taiwan, China, Japan, and South Korea. Oceanographic investigations provided new insights using a global ocean-eddy dataset. The statistical analysis of eddy association was extended to three other sources of spatial data: random tracks, simulated particles, and surface drifters. Turtles inhabiting the hotspot appear to use eddy features differently from turtles outside the region and when compared with other types of spatial data. Specifically, turtles within the hotspot appear to associate less with strong cyclonic eddy edges than those located elsewhere, or compared with the non-sentient spatial datasets. Further regional oceanographic work, studies of fine-scale horizontal and vertical distributions of loggerhead turtle forage, and quantification of fishery bycatch in the presumed foraging region within the hotspot are key areas for future research.

Acknowledgements

We thank Dudley Chelton, Evan Howell, Asuka Ishizaki, and Hiroyuki Suganuma for their insightful comments, and we particularly appreciate the sharing of the global eddy dataset by Dudley Chelton and his colleagues. Constructive comments were also received from the audience at the Sea Turtle Association of Japan annual meeting in Miyazaki, Japan (November 2009), and at the Fifth International Fishers Forum in Taipei, Taiwan (August 2010), where some parts of this analysis were presented. We thank Karin Forney for sharing her ray-casting code, and the officers and crew of NOAA RV “Oscar Elton Sette” for providing comfortable accommodation to the first author during critical phases of manuscript preparation on cruise OES 10-01.

References

Balazs
G. H.
Miya
R. K.
Beavers
S. C.
Keinath
J. A.
Barnard
D. E.
Musick
J. A.
Bell
B. A.
Procedures to attach a satellite transmitter to the carapace of an adult green turtle, Chelonia mydas
Proceedings of the 15th Annual Symposium on Sea Turtle Biology and Conservation
 , 
1996
NOAA Technical Memorandum, NMFS-SEFSC-387
(pg. 
21
-
26
355 pp
Bentivegna
F.
Valentino
F.
Falco
P.
Zambianchi
E.
Hochscheid
S.
The relationship between loggerhead turtle (Caretta caretta) movement patterns and Mediterranean currents
Marine Biology
 , 
2007
, vol. 
151
 (pg. 
1605
-
1614
)
Bowen
B. W.
Abreu-Grobois
F. A.
Balazs
G. H.
Kamezaki
N.
Limpus
C. J.
Ferl
R. J.
Trans-Pacific migrations of the loggerhead turtle (Caretta caretta) demonstrated with mitochondrial DNA markers
Proceedings of the National Academy of Sciences of the USA
 , 
1995
, vol. 
92
 (pg. 
3731
-
3734
)
Cardona
L.
Revelles
M.
Carreras
C.
San Félix
M.
Gazo
M.
Aguilar
A.
Western Mediterranean immature loggerhead turtles: habitat use in spring and summer assessed through satellite tracking and aerial surveys
Marine Biology
 , 
2005
, vol. 
147
 (pg. 
583
-
591
)
Castro
J. J.
Santiago
J. A.
Santana-Ortega
A. T.
A general theory on fish aggregation to floating objects: an alternative to the meeting point hypothesis
Reviews in Fish Biology and Fisheries
 , 
2001
, vol. 
11
 (pg. 
255
-
277
)
Chelton
D. B.
Schlax
M. G.
Samelson
R. M.
Global observations of nonlinear mesoscale eddies
Progress in Oceanography
  
in press
Chelton
D. B.
Schlax
M. G.
Samelson
R. M.
de Szoeke
R. A.
Global observations of large ocean eddies
Geophysical Research Letters
 , 
2007
, vol. 
34
 pg. 
L15606
 
Cheng
I-J.
Chen
T-H.
The incidental capture of five species of sea turtles by coastal setnet fisheries in the eastern waters of Taiwan
Biological Conservation
 , 
1997
, vol. 
82
 (pg. 
235
-
239
)
Druon
J-N.
Habitat mapping of the Atlantic bluefin tuna derived from satellite data: its potential as a tool for the sustainable management of pelagic fisheries
Marine Policy
 , 
2009
, vol. 
34
 (pg. 
293
-
297
)
Du
Y.
Zhou
C. H.
Shao
Q.
Su
F.
Sea surface temperature and purse net productivity in East China Sea
Geoscience and Remote Sensing Symposium, 2000
 , 
2000
Hawaii, USA
IEEE 2000 International, Honolulu
(pg. 
1872
-
1874
)
Ducet
N.
le Traon
P. Y.
Reverdin
G.
Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and -2
Journal of Geophysical Research
 , 
2000
, vol. 
105
 (pg. 
19477
-
19498
)
Eckert
S. A.
Moore
J. E.
Dunn
D. C.
Van Buiten
R. S.
Eckert
K. L.
Halpin
P. N.
Modeling loggerhead turtle movement in the Mediterranean: importance of body size and oceanography
Ecological Applications
 , 
2008
, vol. 
18
 (pg. 
290
-
308
)
Fréon
P.
Dagorn
L.
Review of fish associative behaviour: toward a generalisation of the meeting point hypothesis
Reviews in Fish Biology and Fisheries
 , 
2000
, vol. 
10
 (pg. 
183
-
207
)
Gilman
E.
Gearhart
J.
Price
B.
Eckert
S.
Milliken
H.
Wang
J.
Swimmer
Y.
, et al.  . 
Mitigating sea turtle by-catch in coastal passive net fisheries
Fish and Fisheries
 , 
2010
, vol. 
11
 (pg. 
57
-
88
)
Hochscheid
S.
Bentivegna
F.
Hamza
A.
Hays
G. C.
When surfacers do not dive: multiple significance of extended surface times in marine turtles
Journal of Experimental Biology
 , 
2010
, vol. 
213
 (pg. 
1328
-
1337
)
Hoskins
B. J.
Draghici
I.
Davies
H. C.
A new look at the omega-equation
Quarterly Journal of the Royal Meteorological Society
 , 
1978
, vol. 
104
 (pg. 
31
-
38
)
Howell
E.
Dutton
P.
Polovina
J.
Bailey
H.
Parker
D.
Balazs
G.
Oceanographic influences on the dive behaviour of juvenile loggerhead turtles (Caretta caretta) in the North Pacific Ocean
Marine Biology
 , 
2010
, vol. 
157
 (pg. 
1011
-
1026
)
Howell
E. A.
Kobayashi
D. R.
Parker
D. M.
Balazs
G. H.
Polovina
J. J.
TurtleWatch: a tool to aid in the bycatch reduction of loggerhead turtles Caretta caretta in the Hawaii-based pelagic longline fishery
Endangered Species Research
 , 
2008
, vol. 
5
 (pg. 
267
-
278
)
Hurlbert
S. H.
Pseudoreplication and the design of ecological field experiments
Ecological Monographs
 , 
1984
, vol. 
54
 (pg. 
187
-
211
)
Ishihara
T.
Japan coastal bycatch investigations
North Pacific Loggerhead Sea Turtle Expert Workshop
 , 
2007
Hawaii, USA
Western Pacific Regional Fishery Management Council, Honolulu
(pg. 
21
-
22
)
Jonsen
I. D.
Flemming
J. M.
Myers
R. A.
Robust state-space modeling of animal movement data
Ecology
 , 
2005
, vol. 
86
 (pg. 
2874
-
2880
)
Kamezaki
N.
Chaloupka
M.
Matsuzawa
Y.
Omuta
K.
Takeshita
H.
Goto
K.
Long-term temporal and geographic trends in nesting abundance of the endangered loggerhead sea turtle in the Japanese Archipelago
Endangered Species Research
  
in press
Kamezaki
N.
Matsuzawa
Y.
Abe
O.
Asakawa
H.
Fujii
T.
Goto
K.
Hagino
S.
, et al.  . 
Bolten
A. B.
Witherington
B. E.
Loggerhead turtles nesting in Japan
Loggerhead Sea Turtles
 , 
2003
Washington, DC
Smithsonian Books
(pg. 
210
-
217
352 pp
Klein
P.
Lapeyre
G.
The oceanic vertical pump induced by mesoscale and submesoscale turbulence
Annual Review of Marine Science
 , 
2009
, vol. 
1
 (pg. 
351
-
375
)
Kobayashi
D. R.
Polovina
J. J.
Evaluation of time area closures to reduce incidental sea turtle take in the Hawaii-based longline fishery: generalized additive model (GAM) development and retrospective examination
 , 
2005
NOAA Technical Memorandum, NMFS-PIFSC-4
pg. 
39 pp
 
Kobayashi
D. R.
Polovina
J. J.
Parker
D. M.
Kamezaki
N.
Cheng
I-J.
Uchida
I.
Dutton
P. H.
, et al.  . 
Pelagic habitat characterization of loggerhead sea turtles, Caretta caretta, in the North Pacific Ocean (1997–2006): insights from satellite tag tracking and remotely sensed data
Journal of Experimental Marine Biology and Ecology
 , 
2008
, vol. 
356
 (pg. 
96
-
114
)
Koch
V.
Nichols
W. J.
Peckham
H.
De La Toba
V.
Estimates of sea turtle mortality from poaching and bycatch in Bahía Magdalena, Baja California Sur, Mexico
Biological Conservation
 , 
2006
, vol. 
128
 (pg. 
327
-
334
)
Liu
K. K.
Peng
T. H.
Shaw
P. T.
Shiah
F. K.
Circulation and biogeochemical processes in the East China Sea and the vicinity of Taiwan: an overview and a brief synthesis
Deep Sea Research II: Topical Studies in Oceanography
 , 
2003
, vol. 
50
 (pg. 
1055
-
1064
)
Lumpkin
R.
Pazos
M.
Griffa
A.
Kirwan
A. D.
Mariano
A. J.
Ozgokmen
T.
Rossby
T.
Measuring surface currents with Surface Velocity Program drifters: the instrument, its data, and some recent results
Lagrangian Analysis and Prediction of Coastal and Ocean Dynamics
 , 
2006
Cambridge, UK
Cambridge University Press
(pg. 
39
-
67
500 pp
McCune
B.
Grace
J. B.
Analysis of Ecological Communities
 , 
2002
Gleneden Beach, OR
MJM Press
pg. 
300 pp
 
McGillicuddy
D. J.
Robinson
A. R.
Siegel
D. A.
Jannasch
H. W.
Johnson
R.
Dickey
T. D.
McNeil
J.
, et al.  . 
Influence of mesoscale eddies on new production in the Sargasso Sea
Nature
 , 
1998
, vol. 
394
 (pg. 
263
-
266
)
Mizobata
K.
Saitoh
S.
Shiomoto
S.
Miyamura
T.
Shiga
N.
Toratani
M.
Kajiwara
Y.
, et al.  . 
Bering Sea cyclonic and anticyclonic eddies observed during summer 2000 and 2001
Progress in Oceanography
 , 
2002
, vol. 
55
 (pg. 
65
-
75
)
Nichols
W. J.
Resendiz
A.
Mayoral-Russeau
C.
Kalb
H.
Wibbels
T.
Biology and conservation of loggerhead turtles (Caretta caretta) in Baja California
Proceedings of the Nineteenth Annual Symposium on Sea Turtle Biology and Conservation
 , 
1999
NOAA Technical Memorandum, NMFS-SEFSC-443
(pg. 
187
-
189
291 pp
Nichols
W. J.
Resendiz
A.
Seminoff
J. A.
Resendiz
B.
Transpacific migration of a loggerhead turtle monitored by satellite telemetry
Bulletin of Marine Science
 , 
2000
, vol. 
67
 (pg. 
937
-
947
)
Palacios
D. M.
Bograd
S. J.
Foley
D. G.
Schwing
F. B.
Oceanographic characteristics of biological hotspots in the North Pacific: a remote sensing perspective
Deep Sea Research II: Topical Studies in Oceanography
 , 
2006
, vol. 
53
 (pg. 
250
-
269
)
Parker
D. M.
Cooke
W. J.
Balazs
G. H.
Diet of oceanic loggerhead sea turtles (Caretta caretta) in the central North Pacific
Fishery Bulletin US
 , 
2005
, vol. 
103
 (pg. 
142
-
152
)
Peckham
S. H.
Diaz
D. M.
Walli
A.
Ruiz
G.
Crowder
L. B.
Nichols
W. J.
Small-scale fisheries bycatch jeopardizes endangered Pacific loggerhead turtles
PLoS ONE
 , 
2007
, vol. 
10
 pg. 
e1040
 
Pisula
W.
Curiosity and Information Seeking in Animal and Human Behaviour
 , 
2009
Boca Raton, FL
Brown Walker Press
pg. 
148 pp
 
Plotkin
P. T.
Wicksten
M. K.
Amos
A. F.
Feeding ecology of the loggerhead sea turtle Caretta caretta in the northwestern Gulf of Mexico
Marine Biology
 , 
1993
, vol. 
115
 (pg. 
1
-
5
)
Polovina
J.
Uchida
I.
Balazs
G.
Howell
E. A.
Parker
D.
Dutton
P.
The Kuroshio Extension Bifurcation Region: a pelagic hotspot for juvenile loggerhead sea turtles
Deep Sea Research II: Topical Studies in Oceanography
 , 
2006
, vol. 
53
 (pg. 
326
-
339
)
Polovina
J. J.
Balazs
G. H.
Howell
E. A.
Parker
D. M.
Seki
M. P.
Dutton
P. H.
Forage and migration habitat of loggerhead (Caretta caretta) and olive ridley (Lepidochelys olivacea) sea turtles in the central North Pacific Ocean
Fisheries Oceanography
 , 
2004
, vol. 
13
 (pg. 
36
-
51
)
Polovina
J. J.
Howell
E. A.
Kobayashi
D. R.
Seki
M. P.
The transition zone chlorophyll front, a dynamic global feature defining migration and forage habitat for marine resources
Progress in Oceanography
 , 
2001
, vol. 
49
 (pg. 
469
-
483
)
Polovina
J. J.
Kleiber
P.
Kobayashi
D. R.
Application of TOPEX/POSEIDON satellite altimetry to simulate transport dynamics of larvae of spiny lobster, Panulirus marginatus, in the northwestern Hawaiian Islands, 1993–1996
Fishery Bulletin US
 , 
1999
, vol. 
97
 (pg. 
132
-
143
)
Polovina
J. J.
Kobayashi
D. R.
Ellis
D. M.
Seki
M. P.
Balazs
G. H.
Turtles on the edge: movement of loggerhead turtles (Caretta caretta) along oceanic fronts in the central North Pacific, 1997–1998
Fisheries Oceanography
 , 
2000
, vol. 
9
 (pg. 
71
-
82
)
Revelles
M.
Cardona
L.
Aguilar
A.
San Félix
M.
Fernández
G.
Habitat use by immature loggerhead sea turtles in the Algerian Basin (western Mediterranean): swimming behaviour, seasonality and dispersal pattern
Marine Biology
 , 
2007
, vol. 
151
 (pg. 
1501
-
1515
)
Roden
G. I.
On the subtropical frontal zone north of Hawaii during winter
Journal of Physical Oceanography
 , 
1980
, vol. 
10
 (pg. 
342
-
362
)
Seney
E. E.
Musick
J. A.
Historical diet analysis of loggerhead sea turtles (Caretta caretta) in Virginia
Copeia
 , 
2007
, vol. 
2007
 (pg. 
478
-
489
)
Shanks
A. L.
Wright
W. G.
Internal-wave-mediated shoreward transport of cyprids, megalopae, and gammarids and correlated longshore differences in the settling rate of intertidal barnacles
Journal of Experimental Marine Biology and Ecology
 , 
1987
, vol. 
114
 (pg. 
1
-
13
)
Smith
W. H. F.
Sandwell
D. T.
Global seafloor topography from satellite altimetry and ship depth soundings
Science
 , 
1997
, vol. 
277
 (pg. 
1957
-
1962
)
Tseng
R-S.
Shen
Y-T.
Lagrangian observations of surface flow patterns in the vicinity of Taiwan
Deep Sea Research II: Topical Studies in Oceanography
 , 
2003
, vol. 
50
 (pg. 
1107
-
1115
)
Uchida
S.
Teruya
H.
Uchida
I.
Transpacific migration of a tagged loggerhead, Caretta caretta. B) Tag-return result of loggerhead released from Okinawa Island, Japan
International Symposium on Sea Turtles 1988 in Japan
 , 
1988
Japan
Himeji City Aquarium and Hiwasa Chelonian Museum
(pg. 
169
-
182
)
Vaillancourt
R. D.
Marra
J.
Seki
M. P.
Parsons
M. L.
Bidigare
R. R.
Impact of a cyclonic eddy on phytoplankton community structure and photosynthetic competency in the subtropical North Pacific Ocean
Deep Sea Research I: Oceanographic Research Papers
 , 
2003
, vol. 
50
 (pg. 
829
-
847
)
Wallace
B. P.
Avens
L.
Braun-McNneill
J.
McClellan
C. M.
The diet composition of immature loggerheads: insights on trophic niche, growth rates, and fisheries interactions
Journal of Experimental Marine Biology and Ecology
 , 
2009
, vol. 
373
 (pg. 
50
-
57
)
Wessel
P.
Smith
W. H. F.
Free software helps map and display data
EOS: Transactions of the American Geophysical Union
 , 
1991
, vol. 
72
 (pg. 
441, 445
-
446
)
Zainuddin
M.
Kiyofuji
H.
Saitoh
K.
Saitoh
S-I.
Using multi-sensor satellite remote sensing and catch data to detect ocean hotspots for albacore (Thunnus alalunga) in the northwestern North Pacific
Deep Sea Research II: Topical Studies in Oceanography
 , 
2006
, vol. 
53
 (pg. 
419
-
431
)