James P. Casey. Department of Biological Sciences. University of North Carolina Wilmington. Approved by. Advisory Committee

Transcription

1 BEHAVIOR AND HABITAT OF LEATHERBACK TURTLES (DERMOCHELYS CORIACEA) FROM THE ST. CROIX, U.S. VIRGIN ISLANDS NESTING POPULATION: EVIDENCE OF FEEDING DURING THE NESTING SEASON James P. Casey A Thesis to be Submitted to the University of North Carolina Wilmington in Partial Fulfillment of the Requirements for the Degree of Master of Science in Biology Department of Biological Sciences University of North Carolina Wilmington 2010 Approved by Advisory Committee Steve D. Emslie Andrew J. Westgate Amanda L. Southwood Williard Chair Accepted by Dean, Graduate School

2 TABLE OF CONTENTS ABSTRACT... iii ACKNOWLEDGEMENTS... iv DEDICATION... vi LIST OF TABLES... vii LIST OF FIGURES...x INTRODUCTION...1 METHODS...12 RESULTS...39 DISCUSSION...70 CONCLUSION...93 REFERENCES ii

3 ABSTRACT Leatherback turtles (Dermochelys coriacea) are long-distance ocean migrants that travel from foraging habitats in temperate latitudes, to tropical nesting beaches to lay eggs every 2-5 years. It is generally assumed that leatherback turtles, like other species of sea turtle, do not feed while offshore from nesting beaches, and rely instead on fat reserves to fuel reproductive activities. Recent studies, however, provide evidence that leatherbacks may forage during the internesting interval while offshore nesting beaches in the Western Atlantic Ocean and Caribbean Sea. Bio-logging technology was used to investigate the foraging behavior and habitat of adult female leatherback turtles from the St. Croix, U.S. Virgin Islands (USVI) nesting population. Leatherback gastrointestinal temperatures (T GT ) were analyzed for sudden fluctuations that were indicative of ingestion events and laboratory ingestion simulations were used to characterize temperature fluctuations associated with ingestion events of prey vs. seawater. Seven leatherbacks were documented to make a combined total of over one-hundred ingestion events of gelatinous prey during the internesting interval. The number of prey ingestion events ranged from 6 to 48 for individual turtles, and the majority (87.4%) of these events occurred during the daytime (05:00-19:00 h). Prey ingestions were most frequently associated with V-shaped dives, and the mean (±1S.D.) depth of turtles at the detection of prey ingestion events ranged from 113 ± 72 to 170 ± 90 m. Although not statistically different, daytime foraging dives were typically deeper than nighttime foraging dives, and the depth at the start of foraging events tended to be deeper during the day compared with night. The satellite-derived tracks of leatherback movements, T GT data, and satellitederived SSH data indicate that leatherbacks do not specifically target meso-scale eddies iii

4 to forage during the nesting season, which may be a result of leatherbacks having limited time during the internesting interval to thoroughly track and explore oceanographic features. It may also reflect a conservative, opportunistic foraging strategy, as energy stores are most likely reserved for reproductive activities rather than foraging effort during the breeding season. Although leatherbacks were found to opportunistically feed during the internesting interval, prey ingestion rates indicate that energy reserves acquired prior to the breeding season are critical for successful reproduction by leatherbacks from the St. Croix, USVI nesting population. iv

5 ACKNOWLEDGEMENTS I am foremost thankful to my graduate advisor, Dr. Amanda Southwood Williard, for her untiring guidance and encouragement that made this project possible. I am forever indebted to Dr. Williard for providing me the opportunity to work on such a fascinating project and for teaching me countless lessons and insights on animal physiology and ecology and academic research in general. I wish to express warm and sincere thanks to my lab mates Lisa Goshe, Leigh Anne Harden, and Jessica Snoddy for their support and friendship. A heartfelt thanks to Dr. Jeanne Garner and Steve Garner, Directors of the West Indies Marine Animal Research and Conservation Service (WIMARCS) Inc., and many other staff and volunteers members of the WIMARCS, Inc. for their hard work, knowledge, and assistance at Sandy Point National Wildlife Refuge. Sincerest gratitude to Dr. Matthew Witt, University of Exeter, for his technical advice and assistance on GIS analysis. My thanks also goes to Dr. Michael Coyne, Director of seaturtle.org, for providing resources used by this project to manage satellite data collected from study turtles. My sincere thanks to John Mauser and Mark Neill at North Carolina Aquarium at Pine Knolls and the North Carolina Aquarium at Fort Fisher, respectively, for providing the jellyfish that I used for my laboratory simulations. I wish to thank scallop fisherman Joe Correia, for returning a satellite transmitter to our research team, and tuna spotter pilot George Purmont, for his generosity and help in locating a satellite transmitter during this study. Special thanks to Dr. Molly Lutcavage and the Large Pelagics Research Center, University of New Hampshire, for providing funding for this research project. Finally, I would like to thank the members of my graduate committee, Drs. Steven D. Emslie and Andrew J. Westgate for their constructive and detailed comments on my research, as well as the faculty and staff of the Department of Biology and Marine Biology at University of North Carolina Wilmington for their support and guidance during my graduate studies. v

6 DEDICATION To my family, the Casey s, for many years of love and support. vi

7 LIST OF TABLES Table Page 1. Summary of the instruments deployed on leatherback turtles nesting at St. Croix, USVI Summary of satellite data relayed from leatherbacks from the St. Croix, USVI. nesting population during 2007 and Summary of internesting diving behavior by leatherback turtles from the St. Croix, USVI nesting population Summary of internesting submergence time and dives with wiggle events by leatherback turtles from the St. Croix, USVI nesting population Diel comparison of hourly number of dives, dive duration, maximum dive depth, and post-dive surface time by leatherback turtles (n = 10) from the St. Croix, USVI nesting population Summary of the internesting gastrointestinal tract temperature (T GT ) and ambient temperature (T A ) for leatherback turtles (n = 8) from the St. Croix, USVI nesting population Summary of total number of dives, dive frequency, and maximum depth of dives by leatherback turtles (n = 8) from the St. Croix, USVI nesting population, while their gastrointestinal temperatures were being monitored using stomach temperature telemetry Summary of dive duration, post-dive surface time and dive shape by leatherback turtles (n = 8) from the St. Croix, USVI nesting population, while their gastrointestinal temperatures were being monitored using stomach temperature telemetry Summary of ingestion events documented in gastrointestinal temperature records of leatherback turtles (n = 7) from the St. Croix, USVI nesting population, using stomach temperature telemetry Summary of laboratory ingestion simulations conducted using artificial seawater and moon jellyfish (Aurelia aurita) Summary of integral index values and duration ingestions events characterized as prey ingestions leatherback turtles (n = 7) from the St. Croix, USVI nesting population vii

8 12. Summary of the dive depth associated with ingestion events characterized as prey ingestions in leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval Summary of maximum depth, duration and the post-dive surface time of dives associated with prey ingestion events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval Summary of ambient temperature (T A ) at maximum depth and post-dive surface times of dives associated with prey ingestion events by leatherbacks turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval Summary of dive depth and duration for V-shaped and U-shaped dives associated with prey ingestion events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval Summary V-shaped and U-shaped dive phases associated with prey ingestion events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval Summary of dives associated with prey ingestion events containing wiggle events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval Summary of bathymetry and sea surface chlorophyll-a concentrations for the locations of prey ingestion events by seven leatherback turtles from the St. Croix, USVI nesting population during the internesting interval Summary of unidentified/non-characterized ingestion events by seven leatherback turtles from the St. Croix, USVI nesting population during the...internesting interval Summary of diving behavior and thermal conditions associated with unidentified/non-characterized ingestion events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population Summary of the dive phase associated with unidentified ingestion events by seven leatherback turtles from the St. Croix, USVI nesting population during the internesting interval Summary of wiggle events associated with unidentified/non-classified ingestion events for leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval viii

9 23. Prey mass estimates for ingestion events characterized as prey for leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval ix

10 LIST OF FIGURES Figure Page 1. Photos of adult leatherback sea turtles (Dermochelys coriacea) at sea and onboard a research vessel Map displays the location of Sandy Point National Wildlife Refuge (SPNWR), St. Croix, USVI Photos of leatherback sea turtles nesting at Sandy Point National Wildlife Refuge (SPNWR), St. Croix USVI after platform transmitter terminals (PTTs) were attached to their carapace during this study Photos of the platform transmitter terminal (PTT) attachment site for turtles during the 2007 field season Photos of the attachment site for platform transmitter terminals (PTTs) during the 2008 field season A photo of the stomach temperature pill (STP) insertion method used in this study Photos of stomach temperature pills (STPs) used in this study Dive depth profile and gastrointestinal tract temperature (T GT ) recorded for turtle #AAQ943 during her first 24 h at sea Stomach temperature fluctuations previously observed in cormorants and schematic diagrams of temperature fluctuations following ingestion events Photo of the platform transmitter terminal (PTT) attachment site for turtle #AAR264 immediately following the PTT removal process during the 2007 field season Photos of the platform transmitter terminals (PTTs) attachment site for turtle #XXZ456 and turtle #AAQ943, after being at sea for 29.8 d and 8.0 d, respectively Maps of tracks for turtle #AAR863, #PPQ234, #AAR591, and #PPQ244, during 8-10 d internesting intervals in Maps of tracks for turtle #AAR264, #XXZ142, #AAQ943 and #AAC261, during 8-10 day internesting intervals in 2007 or x

11 14. Map of the kernel-estimated habitat utilization distributions (KHUD) of internesting locations recorded for leatherback turtles from the St. Croix, USVI nesting population during one of their internesting intervals (8-10 d) in 2007 or Map of tracks for turtle #XXZ465 and #KL56, recorded during one of their internesting intervals in Map of track recorded for turtle # during one of its internesting intervals in Map shows the location of the satellite transmission in received for turtle #AAC271 during its 0.1 d tracking period Maps of post-nesting tracks recorded for turtle #AAR530, #XXZ126, and #VI Map of post-nesting migrations tracks for turtle #0XXZ481, #AAV935, #AAG914, #XXZ123, recorded during 2007 and Maps of post-nesting tracks for turtle #XXZ481and #AAV935, recorded during Maps of post-nesting tracks for turtle #AAG914 and #XXZ123, recorded during Satellite-derived relative frequency distributions of maximum dive depth by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2007 when the turtles were within the Caribbean Sea Satellite-derived relative frequency distributions of maximum dive depth by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2008 when the turtles were within the Caribbean Sea Satellite-derived relative frequency distributions of dive durations by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2007 when the turtles were within the Caribbean Sea Satellite-derived relative frequency distributions of dive durations by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2008 when the turtles were within the Caribbean Sea Satellite-derived time-at-temperature distributions by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2007 when the turtles were within the Caribbean Sea xi

12 27. Satellite-derived time-at-temperature distributions by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2008 when the turtles were within the Caribbean Sea Diel dive data for leatherback turtles from the St. Croix, USVI nesting population, recorded during one of their internesting intervals Dive profile recorded for turtle #AAC261 on day 2 of her internesting interval Archival time-series gastrointestinal tract temperature and dive profile recorded for turtle #PPQ234 during its gastrointestinal tract temperature monitoring period Archival time-series gastrointestinal tract temperature and dive profile record for turtle #AAR863 during its gastrointestinal tract temperature monitoring period Archival time-series gastrointestinal tract temperature and dive profile recorded for turtle #AAV935 during its gastrointestinal tract temperature monitoring Period Archival time-series gastrointestinal tract temperature and dive profile recorded for turtle #AAR264 during its gastrointestinal tract temperature monitoring period Archival time-series gastrointestinal tract temperature and dive profile recorded for turtle #XXZ465 during its gastrointestinal tract temperature monitoring period Archival time-series gastrointestinal tract temperature and dive profile recorded for turtle #AAC261 during its gastrointestinal tract temperature monitoring period Archival time-series gastrointestinal tract temperature and dive profile recorded for turtle # during its gastrointestinal tract temperature monitoring period Archival time-series gastrointestinal tract temperature and dive profile recorded for turtle #AAQ943 during its gastrointestinal tract temperature monitoring period Gastrointestinal temperature (T GT ) and ambient temperature (T A ) and depth recorded for turtle #XXZ465 during its tracking period from 21 May 2008 at 18:00 to 29 May 2008 at 18: xii

13 39. Diel dive frequency and maximum depth by turtles, while their gastrointestinal temperatures were monitored Diel post-dive surface time and dive duration by turtles, while their gastrointestinal temperatures were monitored Frequency of ingestion events identified in the gastrointestinal tract temperature data for turtle #PPQ235, #AAR863, #AAR264, #XXZ465, #AAC261, # and #AAQ943, recorded during one of their internesting intervals Integral index values of various combinations of jellyfish and artificial seawater used for laboratory ingestion simulations Integral index values for sudden fluctuations in gastrointestinal temperatures of leatherback turtles that were indicative of ingestion events Frequency of sudden fluctuations in gastrointestinal temperature characterized as prey ingestion events by turtle #PPQ235, #AAR863, #AAR264, #XXZ465, #AAC261, # and #AAQ943, recorded during one of their internesting intervals The difference in time between the start time of prey ingestion events and time at which turtle reached maximum depth (prey ingestion time time at maximum depth of ingestion dives) by seven leatherback turtles from the St. Croix, USVI nesting population during one of their internesting intervals Frequency of prey ingestion events recorded during each phase of the dive (descent, bottom, ascent and surface) for V-shaped and U-shaped dive types and the frequency of wiggle events associated with V-shaped and U-shaped dive types by seven leatherback turtles in the Caribbean Sea during their internesting interval Map of cubic interpolated tracks and locations of prey ingestion events for turtle #PPQ234 and #AAR863, recorded during each turtles gastrointestinal monitoring period, overlayed onto a sea surface height deviation image Maps display cubic interpolated tracks and locations of prey ingestion events for turtle #AAR264 from 12-May-2008 to 21-May-2008 and for #XXZ465 from 15- May-2008 to 20-May-2008, recorded during each turtles gastrointestinal monitoring period, overlayed onto sea surface height deviation image Maps display cubic interpolated tracks and locations of prey ingestion events for turtle #XXZ456 from 21-May-2008 to 27-May-2008 and from 28-May-2008 to 03-June-2008, recorded during the turtles gastrointestinal monitoring period, overlayed onto sea surface height deviation image xiii

14 50. Maps display cubic interpolated tracks and locations of prey ingestion events for turtle #XXZ456 from 04-June to 11-June-2008 and turtle #AAC261 from 19-May-2008 to 28-May-2008, recorded during each turtles gastrointestinal monitoring period, overlayed onto a satellite-derived sea surface height deviation image Maps display cubic interpolated tracks and locations of prey ingestion events for turtle # from 20-May-2008 to 27-May-2008 and for turtle #PPQ234 from 20-May-2008 to 28-May-2008, recorded during each turtles gastrointestinal monitoring period, overlayed onto a satellite-derived sea surface height deviation image Maps shows satellite derived sea surface height (SSH) anomaly data on May 14, 2008, May 28, 2008, and June 11, Diffuse attenuation coefficient at 490 nm in the Eastern Caribbean Sea between 5/20/2008 and 5/27/2008, recorded by Moderate Resolution Imaging Spectrodiometer (MODIS) sensors aboard NASA s Aqua and Terra satellite platforms Maps of interpolated locations of tracked turtles in 2007 and 2008 when they displayed sudden fluctuations in their gastrointestinal track temperature that were characterized as prey ingestion events, overlayed onto satellite-derived sea surface concentrations of chlorophyll-a image (SeaWiFI) Internesting foraging hot-spots for gravid leatherback turtles from the St. Croix, USVI nesting population, based on records of sudden fluctuations in the turtles gastrointestinal tract temperature that were characterized as prey ingestion events in 2007 and Frequency of sudden fluctuation in gastrointestinal tract temperature in seven leatherback turtles between nesting events at St. Croix, USVI recorded during each phase of the dive (descent, bottom, ascent and surface) for V-shaped and U- shaped dive type that were unidentified/non-characterized ingestion events The relationship between INT and energy for laboratory ingestion simulations with jellyfish (closed circles) and seawater (open circles) Map showing the current critical for leatherbacks in waters adjacent to SPNWR, St. Croix, U.S. Virgin Islands, which was redesignated by the National Marine Fisheries Service in xiv

15 INTRODUCTION The leatherback turtle (Dermochelys coriacea) is the sole remaining member of the family Dermochelyidae, a lineage thought to have diverged from other chelonians during the late Cretaceous or Jurassic Period, approximately million years ago (Pritchard, 1997). Leatherbacks have the most wide-ranging distribution of any reptile, spanning from tropical to boreal waters of the Atlantic, Pacific and Indian Oceans (Spotila, 2003). They are extremely large animals, with nesting females typically weighing between kg (Georges and Fossette, 2006). The largest leatherback on record is a male turtle that stranded dead off the coast of Wales in 1988 and weighed >900 kg (Morgan, 1989). Besides the leatherback s large size, the texture and general appearance of the leatherback s shell (dorsal carapace and ventral plastron) are among its most distinct morphological characteristics. Their shell lacks the boney scutes and scales present in other extant sea turtle species (Family Cheloniidae), and is composed of thousands of small, polygonal bones embedded in a fibrous matrix covered by a layer of smooth, dark, leathery skin that is typically white-spotted (Pritchard, 1997). The leatherback s shell is flexible due to its unique composition and has been reported to compress and become concave at the start of their dives (Spotila, 2004). Dives > 1000 m in depth have been recorded for leatherback turtles, although routine dives fall within the range of m (Eckert et al. 1989b; Hays et al., 2006; Sale et al., 2006). The carapace of mature adults has seven prominent and streamlined dorsal longitudinal ridges that taper to a supracaudal point and typically measures an astonishing cm (curved length) (Georges and Fossette, 2006; Fig. 1).

16 Leatherbacks also have several internal anatomical features that further demonstrate their uniqueness not only as a species of sea turtle, but also as a reptile. Beneath the leatherback s carapace is a thick layer of white adipose tissue (Greer et al., 1973; Davenport et al., 1998) that lies above a layer of brown adipose tissue (Goff and Stenson, 1988). The functional significance of brown adipose tissue has not been described for leatherbacks, but this tissue is found in many species of mammals and is well known for releasing metabolic heat through non-shivering thermogenesis (Cannon and Nedergaard, 2004). In addition to the insulative and potentially heat-generating properties of these adipose deposits, leatherbacks also possess an arrangement of blood vessels that may serve as counter-current heat exchangers at the base of their flippers (Greer et al., 1973). These morphological and physiological features are thought to function in combination with the leatherback s large body size to enhance the species ability to retain metabolic heat and maintain a thermal gradient between internal body temperatures (T B ) and ambient water temperatures (T W ) (Frair et al., 1972; James and Mrosovsky, 2004; Southwood et al., 2005; Wallace and Jones, 2008). In one study, a leatherback s T B was recorded to be approximately 26ºC after the animal was housed for more than 24 h in a holding tank with T W of 7.5ºC a temperature difference of approximately 18.5ºC (Friar et al. 1972). Alterations in activity patterns and behavior (i.e. traveling rate and shuttling between cold and warm waters) may also play an important role in thermoregulation (Southwood et al., 2005; Wallace et al., 2005; Bostrom and Jones, 2007). Satellite telemetry studies show that leatherbacks undertake long-distance oceanic migrations, sometimes across entire ocean basins, and travel seasonally between high and 2

17 low latitudes (e.g. Keinath and Musick, 1993; Hays et al., 2003, 2004; James et al., 2005a; Eckert, 2006; Shillinger et al., 2008). During the summer and fall, leatherbacks in the Atlantic Ocean are found in temperate waters off the coast of Europe and North America, including Ireland, the United Kingdom, Canada and the U.S. (Hays et al., 2003, 2004; James et al., 2005ab; Eckert, 2006; Houghton et al., 2006; James et al., 2006ab, Witt et al., 2007). These sites are known to be valuable foraging grounds for leatherback turtles where they exploit seasonally high concentrations of their prey, such as the jellyfish species Aurelia aurita, Rhizostoma octopus, Rhizostoma pulmo and Cyanea capillata (James and Herman, 2001; James and Mrosovsky, 2004; Hays et al., 2004; Houghton et al., 2006; Jonsen et al., 2007; Witt et al., 2007). The energy consequences of relying primarily on a diet of jellyfish, an organism that has a high water content and low caloric value, are an especially confounding aspect of leatherback biology. For example, the jellyfish Cyanea capillata has an average energy density of only 0.18 ± 0.05 kj g wet mass -1 (mean ± 1SD) (Doyle et al., 2007). Davenport (1998) suggested that for leatherbacks to meet resting metabolic demands, they need to consume a total mass of prey each day that is 50% of their body mass; a feat which was documented off the coast of France during leatherback surface feeding events (Duron, 1978, as cited in Davenport and Balazs, 1991). Leatherbacks leave high latitude foraging grounds in the North Atlantic during the fall and winter to begin southerly migrations, with some turtles heading to breeding and nesting areas in the tropics and subtropics (James et al., 2005ab; Eckert, 2006), and others utilizing low latitude open ocean foraging areas (James et al., 2005a; Eckert, 2006). The time between successive nesting years, known as the remigration interval 3

18 (RI), is typically 2 4 years for leatherbacks (Miller, 1997). There are data that support the idea that leatherbacks must obtain critical energy stores before they return to nesting and breeding areas and that leatherback RIs vary with environmental conditions, which affect prey distribution and abundance at putative foraging grounds (Saba et al., 2007; Saba et al., 2008). Researchers at Parque Nacional Marino Las Baulas (PNMB), Costa Rica, demonstrated that the numbers of nesting remigrants (i.e. turtles that have nested at the same beach in prior years) peaked following seasons of cool, highly productive sea surface temperature phases at the equatorial region of eastern Pacific Ocean, which were caused by La Niña phases of the El Niño-Southern Oscillation (ENSO) (Saba et al., 2007, 2008). In contrast, PNMB had low nesting numbers of remigrants following seasons of warm, low production sea surface temperatures at the equatorial region in the eastern Pacific Ocean, which were caused by the El Niño phase of the ENSO (Saba et al., 2007, 2008). Saba et al. (2007; 2008) suggested that leatherback RIs are strongly influenced by prey availability at putative foraging grounds in the south equatorial region of the Pacific Ocean. Environmental conditions associated with ENSO have also been reported to influence the RIs of green sea turtles nesting at Tortuguero, Costa Rica and Raine Island, Australia (Limpus and Nicholls, 1988; Broderick et al., 2001). Sea turtles are generally assumed to be capital breeders, relying on energy stores accrued at foraging grounds to fully support all activities associated with reproduction (Miller, 1997; Hays, 2003; James et al., 2005b; Reina et al., 2005). The capital breeding strategy is widespread among vertebrates, and numerous physiological and behavioral studies have been conducted to investigate how efficient use of energy stores by capital 4

19 breeders ensures successful reproduction (e.g. Parker and Holm, 1990; Doughty and Shine, 1997; Hendry and Berg, 1999; Madisen and Shine, 2006). Within the marine realm, animals as diverse as the crab eater seal (Lobodon carcinophagus) (Boyd, 2000), the northern elephant seal (Crocker et al., 2001), the common eider (Somateria mollissima) (Meijer and Drent, 2008) and the emperor penguin (Aptenodytes forsteri) (Robin et al., 1998) employ a capital breeding strategy. In contrast to the general assumption that sea turtles are capital breeders, green sea turtles display feeding behaviors during their internesting intervals, or the time between successive nesting events, at sites where potential food items are available (Hochscheid et al., 1999). In addition, gravid green sea turtles nesting at Raine Island, Australia, were documented to have fresh food material in their gastro-intestinal tract (Tucker and Read, 2001). Until recent years, leatherbacks were thought to conform to a capital breeding strategy. Eckert et al. (1989b) was the first to suggest that leatherbacks may not be capital breeders based on multiple lines of evidence. Eckert et al. (1989b) weighed female leatherbacks during the nesting season at St. Croix, USVI and reported a total decrease in mass of 2.0 ± 4.3 kg between consecutive nesting events. This weight loss was much lower than expected based on leatherback egg clutches having a mean mass of 9.1 ± 1.5 kg and the assumption that internesting weight loss should account for approximately 50% of the weight of egg clutch production (Eckert et al., 1989b). These data led Eckert et al. (1989b) to suggest that leatherbacks might feed during the internesting interval and that the internesting diel diving patterns displayed by female leatherbacks nesting at St. Croix, USVI. may be a result of the turtles foraging on the deep scattering layer (DSL) as it migrates vertically in the water column. 5

20 The DSL refers to an extended horizontal aggregation of marine animals that can be detected by the backscattering of sound waves from an echo-sounder or sonar on an echo-gram (Iida et al., 1996). Jellyfish, the leatherback s main prey item, and many other organisms are present in the DSL (Brahm, 1966; Balino and Aksnes, 1993; Iida et al., 1996; Brierly et al., 2001; Bamstedt, 2003; Kaartvedt et al., 2007). The DSL migrates vertically in the water column such that organisms normally reposition themselves from deep waters during the day to shallow habitats at night, a processed referred to as diel vertical migration (DVM) (reviewed in Hays et al., 2003). In other studies, animals such as the bigeye tuna (Thunnus obesus) (Dagorn and Josse, 2000), dusky dolphin (Lagenorhynchus obscurus) (Benoit-Bird et al., 2004), and northern elephant seals (Mirounga angustriostris) (Le Boeuf et al., 2000) have been found to forage on DSLs as they vertically migrate throughout the water column. Leatherback turtles in the Caribbean Sea conduct longer, deeper dives during the daytime and shorter, more frequent dives during the nighttime (Eckert et al. 1989b; Myers and Hays 2006), a diel dive pattern that may reflect a strategy to improve foraging success on gelatinous prey in the DSL as it moves vertically in the water column. Subsequent studies have confirmed these dive patterns for leatherback turtles and provided supplemental information on beak-mouth opening movements that could be indicative of prey ingestion during internesting dives (Hays et al., 2004; Myers and Hays 2006; Fossette et al., 2008b). Myers and Hays (2004) and Fossette et al. (2008b) used Inter-Mandibular Angle Sensors (IMASENs) to record beak-mouth opening movements by free-swimming leatherbacks during the internesting period offshore from Grenada and French Guiana, respectively. Beak-mouth opening movements occurred during descent, 6

21 ascent, and bottom phases of dives, as well as during wiggle events (i.e. sudden changes in vertical direction), leading the researchers to suggest that the leatherbacks were foraging during their internesting intervals (Myers and Hays, 2006; Fossette et al., 2008b). Under a capital breeding strategy, leatherbacks must acquire sufficient energy stores at foraging grounds to fuel long-distance migrations (3,000-5,000 km) from high latitude foraging areas to low-latitude nesting areas (James et al., 2005ab). They must also have enough energy stores to support all further activities associated with successful reproduction once they arrive at their breeding and nesting areas. The energetic cost of reproduction is high for sea turtles and encompasses vitellogenesis (egg production), as well strenuous terrestrial nesting activity. Based on the energy content of reptilian eggs, Wallace et al. (2006) estimated that the energy allocated by female leatherbacks to vitellogenesis, which appears to be complete by the time they arrive at breeding and nesting areas (Rostal et al., 1996; James et al., 2005ab), ranges from 390 to 473 x 10 3 kj season -1. During a single nesting season, leatherbacks lay 4-8 egg clutches and spend 8-10 days at sea between nesting events (Miller, 1997). A typical leatherback clutch contains 80 eggs, and their eggs are larger than eggs of other sea turtle species, with a diameter of 53.4 ± 0.5 mm (mean ± 1S.D.) and mean mass of 75.9 ± 4.2 g (Miller, 1997). To deposit their eggs, leatherbacks must haul out of the surf onto the sand, ascend the nesting beach, dig a body pit and egg chamber, oviposit, cover the egg chamber, disguise the nest site, and then crawl back into the surf (Miller 1997). 7

22 Metabolic rates of nesting female leatherbacks have been recorded to investigate the costs associated with the leatherback s terrestrial nesting behaviors. Lutcavage et al. (1990) used an open-flow system of respirometry to measure the oxygen consumption of ovipositing leatherbacks and reported a mass-specific metabolic rate of 0.25 ± 0.04 ml O 2 min -1 kg -1 (0.08 W kg -1 ) for leatherbacks during this activity. In comparison, the mean metabolic rates of leatherbacks during walking and nest covering on the beach, determined by an open-flow system of respirometry, were up to times higher on average than those recorded during oviposition (Paladino et al., 1990; Lutcavage et al., 1992; Paladino et al., 1996). The terrestrial nesting activities by leatherbacks, required for successful reproduction, account for a considerable portion of energy stores and are estimated to range from 17 x 10 3 to 22 x 10 3 kj season -1 (Wallace et al., 2006). Assessing energetic expenditure for leatherbacks during their onshore activities has been fairly easy, but determining energetic costs for leatherbacks at sea has proven to be difficult. One study successfully recorded leatherback field metabolic rates during internesting intervals using a technique called the doubly labeled water method (Wallace et al., 2005). Wallace et al. (2005) reported mass specific field metabolic rates of 0.20 to 0.74 W kg -1 for Pacific leatherbacks during the internesting interval offshore from Costa Rica. The field metabolic rates reported by Wallace et al. (2005) are the only available estimates of energetic expenditure by free-swimming leatherbacks and were recorded from turtles in warm tropical waters during their internesting periods a time when leatherbacks are likely to partition the majority of their energy resources toward reproductive activities, such as egg production (Reina et al. 2005; Wallace et al., 2005). 8

23 Post-reproductive movements and behaviors by leatherbacks are expected to have a higher energetic cost compared with the internesting interval, due to increased diving and swimming activity, thermoregulatory costs of foraging in colder waters, and upregulated salt excretion due to increased prey consumption (Davenport, 1998; Eckert et al., 2006; Wallace et al., 2006). Wallace et al. (2006) estimated that round trip migrations between foraging and nesting sites account for the majority of energetic costs associated with successful reproduction and ranged from 4.0 x 10 6 to 5.1 x 10 6 kj season -1. Given the high energetic cost of reproduction for leatherbacks, which is estimated to range from 4.9 x 10 6 kj season -1 to 6.3 x 10 6 kj season -1 and includes long-distance migrations, vitellogenesis, and strenuous terrestrial nesting activity (Wallace et al., 2006), it seems likely that leatherback turtles would forage opportunistically during the nesting season to take advantage of locally available food sources and augment energy reserves. Although previous studies have provided intriguing evidence to support this idea, documentation of prey ingestion and long-term monitoring of leatherback feeding behavior are lacking in the literature. Knowledge of energetic requirements and resource acquisition is critical to understanding and predicting reproductive success for leatherback turtles. Based on worldwide nesting records, leatherback populations have experienced drastic declines over the past several decades (Spotila et al., 1996). Leatherbacks are a currently listed as a Critically Endangered species by the World Conservation Union (IUNC) and are afforded protection under the Endangered Species Act of 1973 in the United States. 9

24 The primary objectives of my research were to (1) obtain detailed information on leatherback foraging during the nesting season, (2) characterize habitats in which successful foraging occurs, and (3) characterize the dive behaviors associated with foraging. The results of my research provide invaluable information on leatherback foraging ecology and energetics that may be applied to enhancing the prospects for the leatherback s survival as a species. Project Justification I investigated the internesting foraging behavior of female leatherback turtles nesting at Sandy Point National Wildlife Refuge (SPNWR) on St. Croix, USVI (Fig. 2). This site was chosen because previous studies had provided behavioral and circumstantial evidence supporting the idea that turtles from this nesting population feed during their internesting intervals (Eckert et al. 1989b), and this provided me with the opportunity to test specific hypotheses regarding foraging patterns. I used the method of stomach temperature telemetry to monitor feeding events in leatherback turtles during their internesting intervals. Stomach temperature telemetry has been widely used to investigate the foraging activity of marine animals, such as sea turtles (i.e. loggerhead turtles (Caretta caretta), Sato et al., 1994), marine birds (e.g. the adelie penguin (Pygoscelis adeliae), African penguin (Spehniscus demersus), gentoo penguin (Pygoscelis papua), King penguin (Aptenodytes patagonicus), wandering albatross (Diomedea exulans), and great cormorants (Phalacrocorax carbo sinensis) (e.g Wilson et al., 1992; Grémillet and Plӧs, 1994; Pütz and Bost, 1994; Wilson et al., 1995; Ancel et al., 1997)) and marine mammals (e.g. northern elephant seals (Mirounga angustiroustris), southern elephant 10

25 seals (Mirounga leonina), California sea lion (Zalophus californianus), grey seals (Halichoerus grypus), and harp seals (Phoca groenlandica) (e.g. Gales and Renouf, 1993; Austin et al., 2006; Kuhn and Costa, 2006; Horsburgh et al., 2008)). The technique of stomach temperature telemetry makes use of a temperature pill that must be ingested by the study animal and is capable of sensing and transmitting internal temperature data to an externally-mounted receiver. While inside the stomach, the pill provides data on the timing and frequency of prey consumed by the study animal. The use of stomach temperature telemetry to monitor feeding behavior of marine predators relies on temperature differences between the study animal and its prey. Accordingly, stomach temperature sensors are generally deployed on endothermic marine predators that prey on ectothermic marine animals. Ingestion of prey at ambient temperature (T A ) by animals that have warm core body temperatures results in a rapid decrease in stomach temperature. The animal s stomach temperature gradually rises back to previous levels as the animal s metabolic heat warms the prey contents inside the stomach (Wilson et al., 1995). Because leatherbacks maintain a significant thermal gradient between core body temperatures and T A and their prey are ectothermic, an investigation of leatherback feeding behavior using stomach temperature telemetry is feasible (Southwood et al., 2005). Temperature fluctuations recorded by stomach temperature telemetry may reflect ingestion of either prey or water. To strengthen the results of my research, I conducted laboratory simulations to characterize temperature fluctuations associated with ingestion of jellyfish and seawater, so that potential feeding and drinking by leatherbacks in our study could be identified in field records. 11

26 Satellite-linked transmitters, also known as platform transmitter terminals (PTTs), with archival capability were used in conjunction with stomach temperature telemetry so that the movements and dive behaviors associated with foraging events during the internesting interval could also be assessed. Opportunistic foraging by leatherback turtles during the internesting interval may be best served by adopting the most energetically efficient dive pattern that would result in contact with prey. If the leatherbacks are tracking vertically migrating prey in the DSL, this may be reflected by a diel dive pattern of short, shallow nighttime dives and long, deep daytime dives. By adopting this foraging strategy leatherbacks would minimize energy expenditure in their search for food and conserve energy stores for reproduction. Targeting vertically migrating prey while it is at shallow depth at nighttime would provide an energetically cheap method of foraging. Based on diel diving patterns previously documented for leatherback turtles during interesting intervals in the Caribbean Sea (Eckert et al., 1989b; Hays et al., 2004; Myers and Hays 2006), I hypothesized that leatherbacks would concentrate foraging efforts during nighttime and would therefore display a higher frequency of feeding events during the nighttime compared with daytime. I also expected that foraging events would occur at shallower depths at nighttime compared to the daytime because of their internesting diel diving pattern previously documented in the Caribbean Sea, which may be reflective of foraging on the deep scattering layer. METHODS Field Procedures Research was conducted at Sandy Point National Wildlife Refuge (SPNWR), St. Croix, USVI, a site managed by the United States Fish and Wildlife Service (USFWS) 12

27 since 1984 (Fig. 2). SPNWR is located in the southwest corner of St. Croix at 64 50'00'' W, 17 40'12'' N. The refuge has approximately 3 km of continuous coastline, and it currently hosts the largest nesting population of leatherbacks in the United States and its territories (Dutton et al., 2005). For more than 35 years, researchers at SPNWR have been collecting information on female leatherbacks that come ashore to lay eggs between February and August (Dutton et al., 2005). Long-term tagging records show that leatherbacks usually return to SPNWR to nest every 2-5 years, lay an average of 4-8 clutches in a nesting season, and spend 8-10 days at sea between each nesting event (Boulon et al., 1996). The nesting behavior of leatherbacks at SPNWR provided the opportunity to deploy instruments on nesting turtles and to retrieve the instruments when turtles returned to nest again. The West Indies Marine Animal Research and Conservation Service (WIMARCS), Inc., has monitored leatherback nesting on SPNWR since The WIMARCS, Inc. provided assistance in selecting individual turtles for my study, locating nesting turtles for instrument deployments, and deploying field instruments. Study turtles were selected based on their nesting history at SPNWR, with preference given to turtles that had at least a 3-year nesting history at SPNWR and had already laid 2-4 nests at the time of instrument deployment. Careful selection of study animals was critical, as it increased the likelihood that I would re-encounter the turtle on a subsequent nesting attempt and have the opportunity to retrieve instruments to download archived data. The primary leatherback nesting area on SPNWR was surveyed from 20:00 to 05:00 h to intercept nesting females for this study. 13

28 A combination of instruments were deployed on a total of 19 nesting leatherback turtles (standard curved carapace length (SCCL): ± 5.6 cm), during the 2007 (n = 9) and 2008 (n = 10) nesting seasons, to investigate their internesting foraging behavior and habitat. A PTT capable of transmitting data to the Argos satellite system, as well as archiving data (model Mk10-AL, Wildlife Computers, Inc., Redmond, WA,) (see Instrumentation: Platform Transmitter Terminals) was attached directly to the carapace of turtles while they were ovipositing. Prior to attaching the PTTs to the turtle s, the PTT attachment sites were cleaned with isopropyl alcohol (70%) followed by application of betadine antiseptic. A cordless drill (DC volt, DeWalt) with a 4 mm surgical drill bit (model QC 4.0 mm 195/40 mm, Apiary Medical, Inc., West Milford, NJ) was used to drill holes in the carapace for each PTT attachment (see below). The PTTs deployed in 2007 were attached to the leading edge of the carapace between the medial and 1 st lateral ridge (Fig. 3A). Four holes were drilled into the turtle s carapace to a depth of < 20 mm and at an angle perpendicular to the surface of the carapace. The spacing of the drill holes matched the spacing of holes in the PTT s armplates. Orthopedic-mini-anchors (OMAs) were inserted into the pre-drilled holes to a depth of < 15 mm, leaving ~15 mm of the OMA shaft protruding from the turtle s carapace (Fig. 4A). The holes were treated with an anti-bacterial ointment (Fura-zone, Neogen) prior to and after the insertion of the OMAs. The PTT was placed onto the turtle s carapace with the external shafts of the OMAs passing through holes on the PTT s arm-plates. Stainless steel washers were threaded onto the OMA shaft so that they lay flat across the PTT arm-plates, and a stainless steel hairpin was inserted into a hole at the top end of each OMA shaft to secure the PTT to the OMA (Fig. 4B). 14

29 The PTTs deployed in 2008 were attached directly to the medial ridge of the carapace, posterior to the turtle s scapulae (Fig. 3B). Two small holes (4 mm diameter) were drilled through the medial ridge and the holes were spaced evenly to match the spacing of the PTTs two arm-plates. Tygon-coated flexible stainless steel wire was passed through the drill holes in the medial ridge and the PTT s arm-plates (Fig. 5A). A biocompatible two-part cold-curing putty (Equinox TM Silicone Putty Products) was molded to the turtle s medial ridge prior to securing the PTT to provide a stable and flat surface for the PTT to rest on (Fig. 5B). The PTT was positioned on top of the putty as the putty was setting, and the ends of the stainless steel wire were then twisted together to secure the PTT to the turtle s carapace. A stomach temperature pill (STP) (model STP3, Wildlife Computers) (see Instruments: Stomach Temperature Pills) was inserted into the turtle s esophagus once ovipositing was complete and the turtle was covering her nest. Two pieces of flat nylon webbing (width 1.9 cm) were placed into the turtle s mouth and used to open the turtle s mouth and to keep the jaws agape (~15-20 cm) during STP insertion (Fig. 6). The STP was placed inside the lumen of a lubricated, flexible, braided PVC tube (inner diameter 3.9 cm; outer diameter 4.7 cm), which was inserted into the turtle s esophagus to a depth of approximately 40 cm. The STP was then pushed out of the PVC tube and into the turtle s esophagus using a plastic, rigid PVC trocar, threaded through the flexible PVC tube. The handle of the rigid trocar/ had a rubber stopper to limit its extension to 2 cm beyond the insertion end of the flexible PVC tube. PTTs were retrieved when turtles returned to nest again on SPNWR. We relocated 5 of the 9 turtles on which we deployed instruments in 2007, and 6 of the 10 turtles on 15

30 which we deployed instruments in The OMAs used to attach PTTs to the carapace during the 2007 field season were not removed, but the tygon-coated stainless steel wires used to attach PTTs to the medial ridge during the 2008 field season were clipped and removed with the PTTs. Following removal of the PTTs, each turtle s PTT attachment site was carefully examined, photographed, and treated with betadine antiseptic and antibacterial ointment. The archived data stored on the PTT s flash memory drive were downloaded to a computer using Wildlife Computers instrument software (MK-Host v ). The STPs were not retrieved, as they pass through the gastrointestinal tract of the turtle and were ultimately excreted into the ocean. All field procedures were approved by Institutional Animal Care and Use Committee (IACUC) of the UNCW (Protocol # ) and the USFWS (Permit #SPNWR for deployments in 2007 and Permit #SPNWR for deployments in 2008). Instruments: Stomach Temperature Pills (STPs) The STPs used in this study (model STP3, Wildlife Computers) detected temperature using a series of 4 thermistors and relayed temperature data acoustically to an archival PTT (see Instruments: Platform Transmitter Terminal). The components of the STPs were cast in a smooth cylinder of epoxy (63 mm length; 24 mm diameter) and arranged such that the instrument s four thermistors were evenly spaced and attached to a titanium ring that encircled the middle portion of the pill (Fig. 7). This arrangement minimized the time of heat transfer between the pill s external medium and thermistors, so that the pill could rapidly register temperature changes of the external medium. 16

31 The STPs were programmed to emit pulse-coded acoustic signals at 10-second intervals that corresponded to the coldest temperature detected by its thermistors. Battery life for STPs transmitting at this rate was ~22 d. The STPs were capable of measuring temperature from 0 to 50ºC with a resolution of 0.1ºC. Prior to field deployments, each STP was calibrated at temperatures of 15ºC, 22ºC and 30ºC, using a temperature controlled water bath (model Haake DC10-V26/B, Thermoscientific, Inc., Waltham, MA) and a NIST-traceable mercury precision thermometer, with an a resolution of 0.1ºC (model # , VWR International LLC, Batavia, IL). The STPs were coated with dissolvable biocompatible materials to temporarily increase the diameter of the instrument and thereby increase pill retention time in the leatherback s stomach. In 2007, alternating layers of gelatin (Knox Gelatin, Kraft Foods Global Inc.) and ethylcellulose (Ethocel Standard Premium 100, The Dow Chemical Company, Midland, MI) were added to the pill on either side of the thermistor ring (Fig. 7B). The thermistor/titanium ring was left uncovered so as not to interfere with temperature detection. Ethylcellulose forms a dissolvable cellulose matrix and is commonly used in pharmaceutical industry formulations for time-release medicine capsules. The combined layers of gelatin and ethylcellulose increased the diameter of the STPs from 24 mm to ~40 mm. As the gelatin and ethylcellulose dissolved, the overall diameter of the coated STP decreased until the pill was small enough to pass through the pyloric sphincter and into the small intestine. Modifications to the procedure for enlarging the pill s diameter were made after documenting short maximum STP retention times (< 60 h) by leatherbacks in In 2008, only ethylcellulose was used to increase the pill diameter to 40 mm 17

32 (Fig. 7C). The STPs coated with only ethylcellulose were not recovered and assumed to be excreted by the turtle. Efforts were made during the removal or retrieval of all PTTs in this study to listen for the chirping sound made by the STP to assess whether or not the STP had been excreted by the time the turtle returned to nest. Instruments: Platform Transmitter Terminals (PTTs) The archival PTTs used in this study (model Mk10-AL, Wildlife Computers) were configured with multiple sensors for monitoring environmental conditions as well as physiological parameters. Data were stored on the PTT s non-volatile FLASH memory (64 Mbytes for archival data). In addition to archiving data, the PTT s onboard software randomly selected data from the previous 12 d of archival data collection and transmitted the data to the Argos satellite system via the Wildlife Computers Cricket system. All of the PTTs deployed in this study were programmed to operate on a 24-hour duty cycle and satellite transmissions were controlled by activation of a salt water switch, which was triggered when turtle was at the surface and the PTT was exposed to the air. The PTTs operated on 512-KBytes of low-power static random-access memory (RAM) providing optimization for battery life, which was estimated as ~100 d under the user defined settings. The PTT s memory software code resided on its FLASH memory drive and was updated by Wildlife Computers prior to field deployments. Parameters for data collection, storage, and timing of satellite transmissions by the PTTs were programmed prior to deployments, using Wildlife Computers instrument software (MK-Host v ). The PTT s depth sensors and ambient temperature (T A ) sensors were programmed for a 10-second sampling interval. Depth sensors provided 18

33 measurements from 0 to 1000 m, with a resolution of 0.5 m, and accuracy of 1 m for measurements < 1000 m and ±1% of the reading for measurements >1000 m. The T A sensors provided an actual measured range of -40º to +60ºC, with a resolution of 0.05ºC and an accuracy of 0.1ºC. Temperature data transmitted by STPs were received and archived by the PTTs as time-series data at 10-second intervals that coordinated with depth and T A readings. The acoustic signals emitted by the STP were intercepted, interpreted, and archived by the PTT as temperature data with a resolution of ±0.1ºC. The STP-PTT communication range was 2 m. When the STP was activated and was either out of range of its associated PTT or its signal was not properly communicated/interpreted by its associated PTT, the PTT archived the STP data as STP Not Found or STP Repeated Reading. If communication between the STP and the PTT was permanently disrupted (i.e. no temperature readings) within 22 d of deployment, which was the estimated STP battery life, I assumed that the STP had moved out of reception range and had most likely been excreted. Permanent disruption of communication between the STP and PTT after a period 22 d was attributed to either STP battery failure or movement of the STP out of its reception range. Temperature data transmitted by the STP was broadly interpreted as representing gastrointestinal tract temperature (T GT ) because the exact location of the STP inside the turtle was not known while it was within range of the PTT. Fluctuations in T GT due to ingestion of prey or water are typically discernable when STP is in the esophagus or stomach of the study animal. In my study, I assumed the likelihood of detecting ingestion events was very low once the pill moved into the small intestine. 19

34 The PTTs deployed in 2007 were capable of archiving T GT data, but were not capable of relaying T GT data to satellite. The PTTs deployed in 2008 had an updated prototype software which permitted archiving of T GT data and transmission of T GT data to the Argos satellite system. I was essentially testing prototype software for the remote collection of T GT data for the manufacturer using our study turtles in The T GT data relayed by satellite in 2008 were not used for detailed investigation of foraging behavior, but were used to assess the STP retention times by turtles that did not return to nest again at SPNWR in PTTs deployed in both 2007 and 2008 were capable of relaying compressed data for depth and T A to the Argos satellite system, in addition to archiving detailed timeseries data for these variables. The PTT s onboard software analyzed its archival depth data, identified dives based on user-defined criteria ( 3 m depth, 60 s duration), sorted dives into user-defined bins for maximum dive depth and dive duration, and relayed these data to the Argos satellite system grouped into four 6-h blocks starting at 00:00 (GMT) of each day. User-defined bins (n=14) for maximum dive depth were as follows: 3-10 m; m; m; m; m; m; m; m; m; m; m; m; m; and > 500 m. User-defined bins (n = 14) for dive duration were as follows: min; 2-4 min; 4-6 min; 6-8 min, 8-10 min; min; min; min; min; min; min; min; min; and > 40 min. The PTT s software also analyzed T A data and sorted it into bins to describe the percent of time spent at a given temperature. User-defined bins (n = 14) for time-at-temperature were as follows: < 5 ºC; 5º-10ºC; 10º-12ºC; 12º-14ºC; 14º-16ºC; 16º- 18ºC; 18º-20ºC; 20º-22ºC; 22º-24ºC; 24º-26ºC; 26º-28ºC; 28º-30ºC; 30-32ºC; and > 32ºC. 20

35 Argos satellite systems provided Doppler-derived satellite locations for the PTTs, along with information on the location quality or class (LC) for each location fix, which is primarily dependent upon the number of uplinks (i.e. satellite messages) received from the PTT. The LC data information is presented as LC 3 (accuracy < 150 m), 2 (accuracy < 350 m), 1 (accuracy < 1000m), 0 (accuracy > 1000 m), A (accuracy undefined), B (accuracy undefined), and Z (location processing failed) (CLS America, Inc. 2007). Information collected by the Argos satellite system was available for download at the Argos website ( and data CDs were provided by Argos on a monthly basis. In addition, Argos locations and information on the quality of the locations were downloaded daily by the Satellite Tracking and Analysis Tool (STAT) online service ( (Coyne and Godley, 2005). Laboratory Procedures: Laboratory Ingestion Simulations To minimize errors in interpreting T GT fluctuations as evidence of foraging, it was necessary to conduct laboratory studies to characterize T GT fluctuations due to both food and seawater ingestion. Captive feeding studies with leatherbacks were not feasible, as adult leatherbacks fare poorly in captivity and the size of STP instruments precludes use with small animals. I conducted a series of laboratory ingestion simulations following methods used by Wilson et al. (1992) and Wilson et al. (1995) to determine if fluctuations in leatherback T GT documented in the field during this study were a result of successful foraging. For the laboratory ingestion simulations, an empty plastic bag, representing a leatherback s stomach, was suspended in a 26 L water bath (model Haake DC10-V26/B 21

36 Thermoscientific Inc., Waltham, MA) set to 28.1ºC to simulate core body temperature in a leatherback turtle. This temperature value was chosen because this was the mean T GT recorded from leatherback turtles during the 2007 season. The plastic bag was suspended in the water from a flat piece of wood (0.2 x 3.0 x 30 cm) anchored on both sides of the water bath. This arrangement provided some stability and easy access to the bag s opening for ingestion simulations. An ethylcellulose-coated STP (see Instruments: Stomach Temperature Pill), with a maximum diameter of 40 mm, was placed into the bottom portion of the bag such that it would remain submerged in the water bath throughout the ingestion simulation. An archival PTT (see Instruments: Platform Transmitter Terminals) was placed within 1 m of the water bath so that transmissions emitted by the STP during the simulation trial would be detected, interpreted, and archived by the PTT. An equilibration period of 5 min was followed by the introduction of a known mass of cold (24.4º-26.4ºC) moon jellyfish (Aurelia aurita) and/or seawater to simulate an ingestion event by a leatherback with an empty stomach (please see below for more details). The seal at the top of the plastic bag was zipped closed and binder clips were placed on the bag s sides at the water surface immediately following introduction of contents into the bag. These steps helped to minimize heat and air transfer between the bag s contents, which were submerged in the water bath, and the area of the bag that remained above water. Ingestion simulation trials were ended when the temperature of contents inside the bag were equal to the temperature of the water bath. A mercury precision thermometer was used to check the temperature of the bag contents at 15-min intervals beginning at 45 min into the simulation. Simulation trials lasted from min. At the end of each 22

37 simulation, the STP temperature data was downloaded from the PTT using Wildlife Computers instrument software (MK-Host v ). These data were viewed in Wildlife Computers data analysis software (Instrument Helper v 1.0) and then imported into OriginPro (v 8.0) graphing and analysis software (Origin Lab Corporation, Northampton, MA) for further analysis (see Statistics and Analysis: Laboratory Ingestion Simulations). A series of laboratory ingestion simulation trials were performed with different masses and combinations of jellyfish and seawater: 100 g of jellyfish with 400 g of seawater (n = 4); 200 g of jellyfish with 300 g of seawater (n = 6); 300 g of jellyfish with 200 g of seawater (n = 5); 500 g of jellyfish (n = 4); 500 g of seawater (n = 10); and 650 g of seawater (n = 6). Seawater used for laboratory simulations was made from Instant Ocean sea salt (United Pet Group, Inc., Cincinnati, OH). The salinity of the artificial seawater ranged from 33 to 36 ppt, and was determined using a handheld refractometer (model, REF201/211/201bp, Human New Era Technologies Co., Ltd,). Jellyfish were obtained from the North Carolina Aquarium at Pine Knoll Shores and The North Carolina Aquarium at Fort Fisher. Jellyfish were transported to a laboratory at the Department of Biology and Marine Biology at the UNCW and housed in a 50-gal tank with artificial seawater (33-36 ppt salinity) and two aerators. Only jellyfish that were alive, as indicated by muscular contractions, were used for laboratory feeding simulations. Jellyfish were removed from their holding tank, and weighed to the nearest 0.1 g using a calibrated digital balance (model 400 g 230 VAC, Ohaus Scout TM ). Prior to simulations, seawater and/or jellyfish were placed into a 500 ml beaker and kept in a circulating water bath (model 1160S, VWR International, LLC) 23

38 until contents inside the beaker reached a stable temperature between 24.4ºC and 26.4ºC for a duration > 10 min. The temperature of the jellyfish and/or seawater was measured with a mercury precision thermometer. Jellyfish were diced into small pieces (~2 x 2 x 2 cm) immediately prior to being used for simulations based on the assumption of leatherbacks shredding their prey into pieces as it travels through their esophagus to their stomach. Numerous factors, such as the volume of the animal s stomach, the fluidity of the contents inside the animal s stomach and the degree of churning or peristalsis by the stomach, must be accounted for to insure relevancy of the laboratory simulations to field data (Wilson et al., 1995). Each of these factors affects the speed at which an animal transfers heat from its stomach to the contents inside, and I attempted to address each of these potential sources of error in my laboratory simulations. Information on the stomach volume of leatherbacks is missing from the current literature. The volume of the bag representing the leatherback s stomach in the laboratory ingestion simulations was based on stomach volume measurements that I made from a single stranded leatherback turtle. The stranded turtle, which had a SCCL of cm, was found offshore of the Outer Banks, NC, and was mildly decomposed at the time of the initial necropsy. A continuous segment of the esophagus, stomach, and small intestines was removed from the stranded turtle and frozen for 8 weeks in a -21ºC freezer at UNCW. After thawing, the stomach capacity (i.e. volume) was immediately measured by closing off the pyloric sphincter with surgical clamps and then completely filling the stomach with water, which was poured into the stomach cavity through the esophageal 24

39 sphincter. I repeated stomach volume measurements 4 times, and found a mean capacity of 485 ± 36 ml. I assumed that stomach size of leatherback turtles scaled directly with SCCL and calculated an approximate stomach capacity of ml for turtles ranging between cm SCCL. This is the size range of turtles for which I documented ingestions during the internesting interval offshore of St. Croix, USVI. Given these data, a plastic bag with a maximum submerged capacity of 700 ml was used for all laboratory simulations. Adult leatherback turtles have a long (~2 m), muscular esophagus that is lined with keratinized spines, called papillae, which are used to squeeze water from their food and to trap the food inside the esophagus when the turtle expels water out of their mouth and nares (Reina et al., 2002). Jellyfish and other gelatinous organisms that leatherbacks prey upon have fragile body structures and could be easily damaged by the leatherback s papillae during prey manipulation and ingestion. Initial processing of prey in the esophagus would increase the speed in heat transfer from the leatherback to ingested food. I accounted for this potential source of error by dicing jellyfish into small pieces before using them in the simulations. Albeit no information exists on the degree to which muscular contraction in the stomach contributes to mechanical digestion of food in leatherbacks or other marine turtles, I assumed some churning or peristalsis took place as the stomach of the stranded turtle was thick and muscular. Working from this assumption, the contents introduced to the plastic bag were squeezed with tongs at 1-min intervals to simulate stomach churning. The portion of the tongs that was in contact with the plastic bag surrounding the seawater 25

40 and/or jellyfish remained submerged in the water bath throughout feeding simulations to prevent temperature disruptions during the trials. Statistics and Analysis: Internesting and Post-nesting movements Information on the horizontal movements of leatherbacks was derived from the Doppler-derived satellite locations provided by the Argos satellite system. Satellitelocation data were filtered with Satellite Tracking and Analysis Tool (STAT) (Coyne and Godley, 2005) following procedures used in other leatherback satellite telemetry studies (Hays et al., 2005; Eckert et al., 2006). Filtering parameters for locations were set as the following: speed 5 km/h, topography > 0 m, and turning angle < 25º. The turning angle criteria used for filtering in STAT has been found to provide more accurate results for estimating movements by leatherbacks based on comparisons between turtles being monitored simultaneous by the Argos Doppler-derived satellite location technology, which provides some locations with errors >1000 m, and FastLoc or Fast-Global Positioning Systems (FGPS) technology, which achieves highly accurate positions that are typically < 50 m ( (Michael Coyne, personal communication). Speed of travel by turtles was estimated based on the time and distances between the turtle s filtered locations. Daily travel rates (km d -1 ) were calculated by multiplying mean hourly travel rates (km h -1 ) for each day by 24. Filtered Argos locations and speeds were downloaded from STAT and plotted using geographic information system plotting and analysis software (ArgGIS Desktop v. 9.2, Environmental Systems Research Institute, Redlands, California). 26

41 To identify the important internesting habitat of the turtles from the St. Croix, USVI nesting population, I calculated a 25%, 50% and 75% fixed kernel habitat utilization distribution (UD), using Hawth s Analysis Tools for ArcGIS Desktop (v. 9.2) (Beyer, 2004). The smoothing parameter selected for the kernel analysis of internesting habitat UDs was determined using Home Range Tools (HRT) for ArcGIS Dekstop (v. 9.2) (Rodgers et al., 2005). One location nearest to mid-day for each turtle s internesting tracking period was incorporated into the kernel analysis to reduce temporal autocorrelation of the location data (De Solla et al., 1999). Turtles that returned to SPNWR after an internesting interval of 8-11 d and had an available satellite location within 6 h of mid-day on each day of their internesting interval were included in the habitat UD analysis. Turtles with exceptionally long internesting intervals in this study, ranging from 16 to 30.1 d (#XXZ465, #KL56 and # ), had 1.5 to 3.0 times the number of satellite tracking days and were excluded from the internesting habitat UD analysis to avoid weighted bias from their data. Hawth s Analysis Tools for ArcGIS Desktop (v. 9.2) was also used to calculate the surface area of the kernel estimated internesting habitat UDs. Statistics and Analysis: Diving Behavior (Satellite Dive Data) Satellite dive data were analyzed for the period of time when turtles were in the Caribbean Sea. The turtles were considered to be in the Atlantic Ocean and engaged in post-nesting migration when they were located west of 61º0'00'' W or north of 19º0'00' N. These locations are > 15 km north and east of islands in the Caribbean Sea and were arbitrarily selected as maritime boundaries between the Atlantic Ocean and Caribbean 27

42 Sea. Turtles that crossed this boundary (n = 4) did not return to the Caribbean Sea during their monitoring periods. Satellite data for dive depth, dive duration and time-at-temperature were received as histograms that displayed the frequency distribution of each variable in user-defined bins (see Instruments: Platform Transmitter Terminals (PTTs)) for 6-h GMT periods (00:00 06:00, 06:00 12:00, 12:00 18:00, 18:00 00:00). To allow for general comparisons among the individual turtles satellite dive depth and dive duration data, the frequency histograms for these data were converted into relative frequency histograms. General comparisons were also made among the turtles time-at-temperature satellite data, which was analyzed using the total percent time at each temperature bin for each turtle s entire recording period in the Caribbean Sea (see Statistics and Analysis: Internesting and Post-Nesting Movements). Statistics and Analysis: Internesting Diving Behavior (Archival Time-Series Data) The archival times-series data downloaded directly from the PTTs (.wch files) were imported into Wildlife Computers data analysis software for dive analysis (Instrument Helper v ). A zero-offset correction function was performed on the depth data to correct for possible drift during data recording. For dive analysis, a dive was defined as a submergence of 3.0 m with a starting and ending depth of 1.0 m. The Wildlife Computers dive analysis software (Instrument Helper v ) calculated the maximum depth, duration, surface time and the number of wiggle events (rapid change in depth > 1m), and the bottom time for each dive. The bottom time or phase of a dive was defined as the period during which depth was greater than 90% of the maximum 28

43 depth of a dive. A dive cycle was defined as an individual dive and the associated postdive surface interval. The dive analysis results were saved as.csv files, using Wildlife Computer data analysis software (Instrument Helper v ) and are hereafter referred to as archival dive summary data. The archival time series data containing depth, T A and T GT records were saved as.csv files, using Wildlife Computers data analysis software and are hereafter referred to as archival time-series data. Using the archival dive summary data, dives by leatherbacks were classified a V- shaped or U-shaped following procedures used in other leatherback studies (e.g. Fossette et al., 2007 and Fossette et al., 2008b). Dives with a bottom phase < 30% the dive duration were classified as V-shaped and dives with a bottom phase > 30% of the dive duration were classified as U-shaped. For each dive, the descent phase was defined as the period from the start of the dive (> 3 m) to the start of the bottom phase and the ascent phase was defined as the period after the bottom phase until the dive ended (< 1 m). The dive phases were used to assess the behavior of turtles during ingestion events (see Statistics and Analysis: Foraging Behavior). Internesting dives from the archival dive summary data were separated into daytime (05:00 18:59 h) and nighttime (19:00 04:59 h) categories to investigate diel behavior and test predictions based on my hypothesis that leatherbacks concentrate foraging efforts at night. Paired t-test was used to test for significant differences between daytime and nighttime mean number of dives (dives h -1 ), mean maximum dive depth, mean post-dive surface times, the proportion of dives containing a wiggle event and the proportion of V-shaped vs. U-shaped dives. I accepted differences to be statistically 29

44 significant at P < All statistical tests were performed using OriginPro (v.8.0) graphing and data analysis software (Origin Lab Corporation). Statistics and Analysis: Gastrointestinal tract temperatures and Ingestion Events (Archival time-series data) The internesting archival time-series data for T GT, T A, and depth were imported into OriginPro (v 8.0) graphing and data analysis software (Origin Lab Corporation). I used these detailed data to assess when ingestion events occurred and the behaviors and thermal conditions that were associated with the ingestion events. The archival timeseries T GT data were plotted and manually filtered for erroneous data, which occurred due to improper STP-PTT communication, prior to detailed analysis for documented ingestions. Erroneous archival times-series T GT data that were filtered had corresponding archival messages of STP Repeated Reading that were consecutively reported for >10 min or were irregular spikes (increases or decreases) in the data of ± 6º-25ºC. The irregular spikes in archival times-series T GT data typically lasted for < 2 min or were followed by corresponding archival time-series STP message of STP Not Found or STP Repeated Reading. A paired t-test was used to determine if there was a statistically significant difference between mean T A and mean T GT for the entire period of archival time-series T GT data for each turtle. This comparison allowed me to assess if leatherbacks maintained a significant thermal gradient between core temperatures and ambient water during the internesting interval at sea. 30

45 For the analysis of leatherback ingestion events, I visually analyzed each turtle s archival time-series T GT data and identified all fluctuations that could be interpreted as ingestion events (Grémillet and Plös, 1994; Southwood et al., 2005). It was common for the turtles T GT to steadily decrease during the first 6 h after departing from SPNWR with their instruments (29.2º-30.9ºC down to 27.7º-28.5ºC). Previous studies have shown that the high levels of activity while on the nesting beach result in elevated body temperatures (Sato et al., 1995; Southwood et al., 2005). The turtles in my study typically made frequent ingestions during their first 6 h at sea, as indicated by the rapid fluctuations in the turtle s archival time-series T GT data (Fig. 9), which may have been caused by the turtles drinking seawater to assist in lowering their body temperatures (i.e. behavioral thermoregulation) or to assist in moving the STP down their esophagus. To minimize error in documenting leatherback ingestion events that may be a result of foraging, T GT data collected during the first 6 h of recording were excluded from analysis. Leatherback ingestion events were identified as a decrease in T GT (ΔT D ) of 0.3ºC (3x the STP accuracy) at a minimum rate of 0.033ºC/min followed by temperature rise (T R ) event. The time at which the ΔT D began was designated as the initial temperature (T I ) of the ingestion event. To minimize error in identifying ingestion events and to account for long-term fluctuations in T GT of leatherbacks (Southwood et al., 2005), specific criteria were adopted for determining the final temperature (T F ) of the ingestion events (i.e. when the ingestion event ended): (1) an ingestion event ended and T F was attained when T GT returned to T I, (2) an ingestion event ended and T F was attained when T GT returned to within 0.1º-0.2ºC of the T I and remained stable for > 25 min or was followed by the start of another ingestion event, and (3) the T F of a sequence of multiple 31

46 ingestions was attained when T GT returned nearest to and within 0.1º-0.2ºC of the T I of the 1 st ingestion in the sequence and remained stable for > 25 min or was followed by the start of another ingestion event. When a stable T GT was obtained for > 25 min during a possible ingestion event prior to the ingestion event meeting at least one of the criterion listed above (1-3), the possible ingestion event was excluded from further analysis. Overall ingestion rates (ingestion h -1 ) for a turtle were based on the total number of documented ingestion events and the duration of the filtered archival times-series T GT data, excluding T GT data collected prior to the turtles being at sea for 6 h. Ingestion events were divided into daytime (05:00-18:59 h) and nighttime (19:00-04:59 h) categories, and the duration of each ingestion event was calculated as the duration between T I and T F. Statistics and Analysis: Laboratory Ingestion Simulations The size and shape of the area below the asymptote of a stomach temperature fluctuation provides important information on the speed at which the animal is able to warm the ingesta (mainly the duration of the T R event) in relation to the magnitude of the ΔT D of the fluctuation. For example, ingestion of a prey item at a cold temperature may result in a small ΔT D with a prolonged T R event compared to a large bolus of cold water that results in a large ΔT D with a rapid T R event (Fig. 9A) (Grémillet and Plös, 1994; Wilson et al., 1995; Wilson et al., 1995; Catry et al., 2004). One method used to analyze stomach temperature fluctuations is to measure the area or integral below the asymptote of the fluctuation from the moment the food is ingested (i.e. when temperature begins to drop) to the point when the temperature returns to the asymptote (Fig. 9B; Wilson et al., 1992, Wilson et al., 1995). Another method used for analysis of stomach temperature 32

47 fluctuations, referred to as the temperature rise integration method (TRIM), is to measure the area below the asymptote from the minimum temperature of the fluctuation (T MIN ) to the point when the temperature returns to the asymptote (Fig. 9C). The TRIM method was used by Grémillet and Plös (1994) to provide more accurate results for estimating the mass of prey ingested by great cormorants (Phalacrocorax carbo sinensis) in the field, based on captive feeding trials with great cormorants, because the birds stomach temperature fluctuations declined as a series of stages following single feeding events in captivity, rather than declining precipitously. In this study, the TRIM method was used to calculate the area of fluctuations in T GT identified as ingestion events because the T GT fluctuations typically declined as a series of stages. All TRIM values were calculated using OriginPro (v.8.0) graphing and data analysis software (Origin Lab Corporation). The TRIM method was also used for analysis of the temperature fluctuations recorded during laboratory ingestion simulations. The integral values measured for the ingestion simulations were used to characterize temperature fluctuations associated with ingestion of various combinations of jellyfish (i.e. leatherback prey) and seawater. This information was used to distinguish ingestions of prey from drinking of seawater in free-swimming leatherbacks (see below). The TRIM values were measured for ingestion events in leatherbacks that displayed a steady T R event from the ingestion T MIN to T F. The maximum change in temperature of the T R event, or the ΔT R, measured in laboratory simulations was 0.7ºC, which was the overall mean ΔT R (range 0.3º to 3.3ºC) for leatherback ingestions included in the integral analysis. The magnitude of the ΔT R may be affected both by the size (volume or mass) and consistency (water or prey) of the ingesta (Fig. 9A). I 33

48 standardized TRIM values for comparisons by dividing the TRIM by its associated ΔT R. I refer to the standardized TRIM values as integral index values. Each integral index value represents the recovery time of an ingestion event corrected for the magnitude of its associated ΔT R. In essence, this procedure allowed for comparisons of TRIM values between ingestions with widely varying magnitudes of ΔT R and to discern differences between prey and water ingestions. A Levene s test detected unequal variance among the integral index values of laboratory simulation groups (i.e. various combinations of jellyfish and seawater), so a non-parametric Kruskal-Wallis test was used to test for a difference between the integral values of the groups. The Kruskal-Wallis test was followed by post-hoc Mann-Whitney U-tests. I made comparisons between the seawater group with the largest mean integral index value (i.e. 650 g of seawater) and groups that contained 200 g of jellyfish. I accepted a difference to be significant for the Kruskal-Wallis test at P < To reduce the potential for Type I errors, a significance level of P < was set for each pair-wise comparison in the Mann-Whitney U-test based on a Bonferroni correction method (α/n). A minimum critical index value for distinguishing prey ingestions from seawater ingestions (i.e. drinking of seawater) was established based on the statistical differences that existed between ingestion simulation groups. All statistical tests were performed using OriginPro (v.8.0) graphing and data analysis software (Origin Lab Corporation). Statistics and Analysis: Foraging Behavior The following analyses were first conducted for all ingestions events identified as prey based on criteria developed from laboratory ingestion simulations (see Statistics and 34

49 Analysis: Laboratory Ingestion Simulations), as well as for ingestion events that did not meet the criteria for prey (unidentified ingestion events). First, ingestion events documented for each turtle were divided into daytime (05:00-18:59 h) and nighttime (19:00-04:59 h) time periods to investigate diel ingestion patterns. A paired t-test was used to test for a significant difference between the daytime (05:00-18:59 h) and nighttime (19:00-04:59 h) mean hourly prey ingestion rates to test the hypothesis that leatherbacks concentrate their foraging efforts to target the DSL when it migrates to shallow waters during the nighttime. The archival time-series data and dive summary data were examined to determine the turtle s behavior, depth, and T A when each prey ingestion event started. Behavioral information assessed for each prey ingestion event included the dive type that the event was associated with (i.e. V-shaped or U- shaped), the dive phase when the event started (i.e. descent, bottom, ascent or surface), and if a wiggle event occurred during the dive that was associated with the event. For each dive associated with a prey ingestion event, I calculated the time lapse between when the turtle reached the maximum depth of the dive and when the prey ingestion event started, and the difference in depth between the maximum dive depth and the depth where the prey ingestion event occurred. A paired t-test was used to test for significant diel differences in the following characteristics of prey ingestion events (diel tests not performed for unidentified ingestion events): mean depth at start of prey ingestion events, mean T A at the start of prey ingestion events, mean ΔT D of prey ingestion events, the mean post-dive surface intervals of dives with prey ingestion events, the mean maximum depth of dives with prey ingestion events, and the mean duration of 35

50 dives with prey ingestion events. Only turtles with at least three daytime and nighttime prey ingestion events were included in the statistical analysis. Comparisons between the proportion of prey ingestion events associated with dive phases of V-shaped dives were made using a one-way analysis of variance (ANOVA) test (significance level P = 0.05), followed by a post-hoc Tukey Test. To reduce the potential for Type I errors, a significance level of P < was set for each pair-wise comparison in the Tukey test based on a Bonferroni correction method (α/n). A Levene s test detected unequal variance among the proportion of U-shaped dives phases with prey ingestion events, so a non-parametric Kruskal-Wallis test was used to compare the proportion of prey ingestion events associated with dive phases of U- shaped dives. The Kruskal-Wallis test was followed by a post-hoc Mann-Whitney U-test. The significance level for the Kruskal-Wallis test was set at P < To reduce the potential for Type I errors, a significance level of P < was set for each pair-wise comparison in the Mann-Whitney U-test based on a Bonferroni correction method (α/n). A paired t-test was used to determine if there was a significant difference between the mean proportions of the dive types that contained both a wiggle event and a prey ingestion event and differences were accepted as significant at P < For dives that were associated with a prey ingestion event and contained a wiggle event, I also determined the time lapse between the start of the wiggle event and the start of the prey ingestion event. A paired t-test was used to compare the mean dive depth, dive duration, and postdive surface time between dives associated with prey ingestion events and dives associated with unidentified ingestion events. A paired t-test was also used to compare 36

51 the T A at the maximum depth between prey and unidentified ingestion events and the T A at the start of ingestion events between prey ingestion events and unidentified ingestion events. All statistical tests were performed using OriginPro (v.8.0) graphing and data analysis software (Origin Lab Corporation). Statistics and Analysis: Foraging Habitat To investigate the internesting foraging habitat of leatherback turtles, the filtered Argos locations of each turtle were first interpolated to provide a smoothed track at 15- min intervals, using MATLAB (v 7.8) (The Math Works, Inc., Natick, MA). The start time of foraging events were then paired to the nearest interpolated latitude and longitude location. The interpolated tracks and locations of foraging were plotted using geographic information system mapping and analysis software (ArgGIS Desktop v. 9.2, Environmental Systems Research Institute, Redlands, California). Oceanographic and biological data were incorporated into ArcGIS with the turtles tracks and locations of foraging for spatial and biological analysis. I used Hawth s Analysis Tools to determine the bathymetry of foraging locations using global one-arc minute grids (GEBCO One Minute Grid). Image composites for sea surface height deviation anomaly, sea surface chlorophyll a concentration, diffuse attenuation coefficient at 490 nm, and geostrophic currents were obtained from the National Oceanic and Atmospheric Administration (NOAA) CoastWatch THREDDS (Thematic Real-Time Environmental Distributed Data Services) catalog ( using Environmental Data Connector (EDC) (Applied Science Associated, Kingston, RI) for ArcGIS Desktop. 37

52 Data for sea surface height deviation (SSHD) (AVISO, Global, Science Quality) were derived by a combination of NOAA satellites (Jason-1, TOPEX/Poseidon, ERS 1/2, and GFO) and was calculated as the mean difference between the mean SSH and the expected mean SSH. SSHD data was downloaded as 4-km resolution one-day duration composite images for the following days: May 16, 2007; May 14, 2008; May 21, 2008; May 28, 2008; June 4, 2008; June 11, Data for chlorophyll-a concentration were collected by Sea-viewing Wide Field-of-view Sensor (SeaWiFS) (10 km resolution; accuracy: ± 30% of the reading) aboard National Aeronautics and Space Agency (NASA) Orbview-2 satellite and were downloaded as monthly duration composite images corresponding to the following periods: May 16, 2007 to June 16, 2007; May 16, 2008 to June 16, Smaller time-scales, such as 8-day composites, collected by SeaWiFS lacked spatial coverage for the area of interested and could not be used for analysis. Data for diffuse attenuation coefficient at 490 nm was collected by Moderate Resolution Imaging Spectroradiometer (MODIS) aboard NASA s Aqua and Terra satellite platforms. The diffuse attenuation coefficient data was downloaded at 4-km resolution as weekly composites from May 04, 2007 to June 21, 2007 and May 4, 2008 to June 21, Geostrophic currents (AVISO), interpreted from SSHD data, were downloaded as oneday image composites (for May 14, 2008; May 21, 2008; May 28, 2008; June 4, 2008; June 11, 2008) to overlay with the satellite-derived altimetry data. Hawth s Analysis Tools was used to identify chlorophyll-a concentration grid cells (10 km 2 ) encompassing the interpolated foraging locations, and for a fixed kernel density estimation to identify internesting foraging hot-spots. I experimented with kernelling options (i.e. grid size and smoothing parameter) and selected the most 38

53 parsimonious estimate, which was obtained with a smoothing parameter of 14 x 10 3 and a grid cell size of 100 m 2. RESULTS Field Observations and Data Collection The turtles in this study were tracked for a period ranging from 0.1 to 131 d (total tracking days logged: d, Table 1). Eleven turtles returned to SPNWR after a mean internesting interval of 13.5 ± 8.5 d (range: 8.1 to 30.8 d). All turtles that returned to nest at SPNWR displayed normal signs of nesting behavior, such as nest construction and egg deposition. Eight turtles did not return to SPNWR during their tagging year and were tracked for a period ranging from 0.1 to 133 d. A total of 11 PTTs were recovered from turtles in this study and all PTTs successfully archived the turtles depth and T A. Ten PTTs were recovered from turtles at SPNWR and 1 PTT (deployed on turtle #AAV935) was recovered in Fairhaven, MA from a scallop fisherman, who dredged up the instrument from the ocean s bottom on October 17, 2007 approximately 60 km offshore from Long Island, NY (Table 1). Turtles that returned to nest at SPNWR (n = 5) following instrument deployments in 2007 retained their OMAs and PTTs during their time at sea (range: 8.1 to 9.1 d, Table 1). The OMAs used for the attachment of the PTTs were present when the turtles returned to SPNWR, but they were no longer securely embedded in the carapace-insertion holes for two turtles. There were no observations of the turtles carapace chaffing from the OMA attachment technique and there were minimal to no signs of necrosis surrounding the turtles OMAs (Fig. 10). 39

54 Five of six turtles that returned to nest at SPNWR in the 2008 season had retained their PTT (Table 1). The PTTs remained securely attached to a turtle s medial dorsal ridge for 8-10 d, but the WIMARCS, Inc. observed two cases in which the instruments had shifted to the side and became loosely secured to the turtles medial dorsal ridge after > 15 d at sea (Fig. 11A). There were no observations of carapace chaffing from the medial dorsal ridge attachment technique, and there were minimal to no signs of necrosis immediately surrounding the drill-holes in the medial dorsal ridge (Figs. 11B-C). The modifications made in 2008 to the STP retention matrix, which was used to enlarge the diameter of the STPs, resulted in longer monitoring periods of turtle T GT. The mean monitoring period of turtle T GT was 1.5 ± 0.9 d in 2007 and 16.1 ± 10.3 d in 2008 (Table 1). In 2007, archival time-series T GT data was obtained from six turtles for a period that ranged from 0.8 to 2.9 d (Table 1). The short duration of T GT monitoring periods and the absence of the STP chirping sound during the PTT removal process for turtles in 2007 provided evidence that the turtles had excreted their STPs prior to returning to nest at SPNWR. I concluded that the T GT monitoring periods were reflective of the amount of time the STPs were retained in the gastrointestinal tract. PTTs were not recovered in 2007 from turtles #XXZ481, #XXZ126, and #AAR530 and, therefore, T GT data were not obtained for these turtles and their STP retention times could not be estimated (satellite T GT data were unavailable in 2007). In 2008, archival time-series T GT data were obtained from five turtles and their T GT were monitored over a period that ranged from 7.6 to 28.0 d (Table 1). The STP retention times could not be estimated for turtles that returned to SPNWR with their PTT 40

55 in 2008 as a result of one of the following: (1) the STP was present when a turtle returned to SPNWR, as indicated by the STP chirping sound during the PTT removal process, and the turtle returned to sea without a PTT to receive STP signals (e.g. turtle #AAR264, #AAC261, # and #AAQ943), or (2) archival time-series data indicated that T GT was monitored longer than the expected battery life of the STP and, therefore, STP battery failure may have occurred prior to the turtle excreting its STP into the ocean and out of range of its PTT (e.g. turtle #XXZ465; Table 1). In 2008, turtle #KL56 returned to nest at SPNWR after a 30.8 d period. The turtle shed/lost its PTT prior to coming ashore at SPNWR and was only monitored by the Argos satellite system for a 14.7 d period (Table 1). Satellite data indicate the turtle s PTT failed at sea while still receiving T GT data, so the turtle s STP retention time could not be estimated. The STP chirping sound was not heard from #KL56 when it returned to nest at SPNWR, but the STP battery may have failed prior this observation (Table 1). The STP retention times by turtles that did not return to SPNWR in 2008 could not be estimated either because satellite data indicated that a PTT failed while monitoring a turtle s T GT (e.g. #AAC271), or because it indicated that a turtle s T GT was monitored longer than the battery life expectancy of the STP and, therefore, STP battery failure may have occurred prior to the turtle excreting its STP (e.g. #VI1400, #AAG914 and #XXZ123; Table 1). I assumed that PTTs deployed on turtles #AAG914 and #XXZ123 ceased working after > 100 d because of PTT battery failure or shedding. Turtle #AAG914 logged a low PTT battery voltage (2.8 volts) on the last day of satellite transmissions (d 131), which supports battery failure as the cause for cessation of transmissions. 41

56 The following observations support premature PTT shedding for those instruments deployed in 2007 that failed in < 100 d: (1) the PTT deployed on turtle #AAV935 failed after 60 d, and was later dredged up from the ocean s floor by a scallop fisherman at approximately the same location as the last transmission (note: #AAV935 returned to SPNWR during the nesting season and laid a total of 5 nests (the WIMARCS, Inc., personal communication)), (2) the PTT deployed on turtle #AAR530 failed after 12.4 d (Table 1), and turtle #AAR530, with no transmitter attached, was observed laying five nests at SPNWR during the nesting season (the WIMARCS, Inc., personal communication), and (3) OMAs were not securely embedded within the carapace-insertion-holes for 2 turtles that returned to nest at SPNWR after 8.9 to 15.7 d at sea. Premature PTT shedding for those instruments deployed in 2008 that failed in < 100 d was suggested by the PTT deployed on turtle #KL56 that failed after 14.7 d and that this turtle was observed nesting at SPNWR without her transmitter ~15 d after its last satellite transmission was received. In addition, the PTTs were offset and not securely attached to medial dorsal ridge of two turtles (# and #XXZ465) after a period > 16 d at sea (Fig. 11A). Internesting Movements Internesting intervals at SPNWR are typically between 8-10 d in duration. Three of the 11 turtles returning to nest at SPNWR had exceptionally long ( d) interesting intervals, and their movements were analyzed separately from turtles with interesting intervals of 8-10 d. The turtles that returned to nest at SPNWR after an 42

57 internesting interval of 8-10 d (n = 8) had 12 to 80 Argos satellite locations after data filtering (Table 2). The turtles traveled a minimum straight-line distance that ranged from 313 to 495 km during their 8-10 d internesting intervals (Tables 1-2, Figs ), and had a maximum displacement from SPNWR that ranged from 70 to 110 km (rounded to nearest km). The turtles typically moved southward or westward for approximately 1-2 d after departing from SPNWR. Thereafter, the turtles generally travelled in a clockwise pattern and toward waters north of St. Croix. Interestingly, two individuals (turtle #AAC261 and #AAQ943) were located (Argos LC: 0 and A) < 10 km offshore from Vieques, Puerto Rico, an island where leatherbacks also nest within the Caribbean Sea during their 8-10 d internesting intervals. All turtles remained > 5 km offshore of SPNWR during the mid-period of their internesting interval, but began to move closer to shore 2 d prior to their next nesting event. The overall mean daily traveling rate during 8-10 d interesting intervals was 50.9 ± 6.9 km/d. Only turtles with a satellite location within 6 h of mid-day on each day of their internesting interval were included in the kernel-estimated internesting habitat UD analysis. Turtle #PPQ234 did not meet these criteria, so its data were not included in this analysis. The kernel-estimated 75% habitat UD, determined from 64 locations of seven turtles, had a total surface area of 5,975.7 m 2 and was centered approximately 30 km to the west of SPNWR (Fig. 14). The 75% UD extended north of SPNWR by 48 km, south of SPNWR by 32 km, west of SPNWR by 91 km, and east of SPNWR by 29 km. The total surface area of the 50% and 25% UDs for internesting locations was 2,515.2 m 2 and 722 m 2, respectively. The location of the 50% and 25% UDs show that the turtles spent the majority of their time within 50 km of SPNWR. 43

58 The number of filtered Argos locations for turtles (n = 3) that returned to SPNWR after 11 d ranged from 81 to 159 (Table 2). The turtles had a mean daily traveling rate that ranged from 57.6 ± 17.6 to 70.6 ± 17.9 km/d while at sea (Tables 1-2). The general movements by turtles during their d days at sea, prior to returning to nest at SPNWR, are described below. Turtle #XXZ465 Turtle #XXZ465 traveled a minimum straight-line distance of 2,165 km during her 28.9 d at sea, and had an overall mean traveling rate of 70.6 ± 17.9 km/d (Tables 1-2, Fig. 15A). Movements by turtle #XXZ465 during the first 2 d of her tracking period (15 16 May 2008) were mainly west of St. Croix. The turtle traveled northward for the next several days of its tracking period, passing through waters < 25 km offshore of St. Croix s west coast. Turtle #XXZ465 was located (Argos LC 0) 12.5 km northeast of St. Croix s east coast by day 5 of its internesting period (20 May 2008), and continued traveling eastward thereafter. The turtle conducted extensive traveling loops during days 5-27 (20 May 2008 to 11 June 2008) (mean traveling rate: 75.8 ± 15.8 km/d) at sea and generally restricted its movements to waters between St. Croix, British Virgin Islands (e.g. Tortola), Anguilla, and the island of St. Kitts and Nevis. The turtle s maximum displacement from SPNWR during her tracking period was km, which was recorded (Argos LC 0) ~17 km south of St. Kitts and Nevis at 62º 51' 50'' W, 17º 10' 52'' N (Table 2, Fig. 15A). The archival time-series dive data and satellite locations obtained for turtle #XXZ465 did not show evidence of a nesting event on islands other than St. Croix. 44

59 Interestingly, turtle #XXZ465 may have attempted to nest on St. Croix s southern coast, just prior to returning SPNWR. Archival time-series data indicate the turtle was at the surface for an extended period of time (> 3 h) on the night of 11 June 2008, and was located (Argos LC 1) < 1 km offshore of St. Croix s southern coast (67º44'35'' W, 17º41'42'' N) where leatherbacks occasionally nest (the WIMARCS, Inc., personal communication). Turtle #XXZ465 successfully nested at SPNWR on the night of 12 June 2008 and, so any attempts made by this turtle to lay eggs on the night of 11 June 2008 at one of St. Croix s southern beaches (e.g. Hey Penny Beach) were apparently unsuccessful. Turtle #KL56 Turtle #KL56 was tracked at sea for 14.7 d, but returned to nest at SPNWR after 28.9 d at sea (Table 1). Turtle #KL56 traveled a minimum straight-line distance of 849 km in the Caribbean Sea, traveled in a clockwise pattern, and had a mean traveling rate of 61.3 ± 27.6 km/d during the 14.7 d tracking period (Tables 1-2, Fig 15B). After departing from SPNWR, turtle #KL56 traveled west southwestward for 4 d (18-21 May 2008; traveling rate range: 52.2 to 74.5 km/d). The turtle s maximum displacement from SPNWR was km and was recorded on day 3 of its tracking period (Table 2). Between her days 5-8 (23 26 May 2008) at sea, turtle #KL56 traveled mainly north northwestward toward Puerto Rico (travelling rate range: 25.6 to 61.1 km/d). On the night of day 8 at sea, turtle #KL56 was located in waters < 12 km offshore of Puerto Rico s southern coastline and was tracked traveling east northeastward. Turtle #KL56 continued traveling generally east northeastward over its internesting tracking 45

60 days 9-12 (27 30 May 2008; traveling rate range: 47.4 to km/d). By mid-day on day 14 (1 June 2008) of its tracking period, turtle #KL56 began shifting from a southeastward to a westward traveling direction, and headed toward the east coast of St. Croix. The turtle s last available satellite location, was on her tracking day 14 at 20:26 h, and she was located (Argos LC 0) 4.0 km offshore of SPNWR. The turtle returned to nest at SPNWR without her PTT approximately 14 d after its last available satellite location. Turtle # Turtle # traveled a minimum straight-line distance of 911 km in the Caribbean Sea, and traveled in a clockwise pattern during its 16 d at sea (Tables 1-2, Fig. 16). Turtle # traveled eastward for five days after departing from SPNWR (mean traveling rate: 62.4 ± 22.0 km/d). On her tracking day 4, turtle # was located (Argos LC 2) km offshore of SPNWR at 64º41'35'' W, 17º35'17'' N, which was her maximum recorded displacement from SPNWR. Turtle # traveled northward toward Puerto Rico during her tracking days 4-5 (24-25 May 2008; mean traveling rate: 49.8 ± 13.7 km/d) and east northeastward during her tracking days 6-10 (26-30 May 2008; mean traveling rate: 45.3 ± 20.7 km/d). The turtle was located (Argos LC 1) < 4 km offshore of the southern coast of Vieques, Puerto Rico on her internesting day 10 (30 May 2008). There was no indication from the turtle s archival time-series data that it attempted to nest when it was located near the island of Vieques, Puerto (e.g. no extended surface times). After being located in close proximity to Vieques, Puerto Rico, turtle # continued travelling generally eastward and to the east of St. Croix, which is where it remained during its days (29-30 May 2008) at sea. During her final day at sea between nesting events at 46

61 SPNWR (day 16), turtle # traveled in close proximity (<1 km) (Argos LC 1) to St. Croix s northern coast and approached SPNWR from the northwest side of St. Croix. Post-nesting Movements in the Caribbean Sea and Atlantic Ocean The eight turtles that did not return to SPNWR traveled a minimum straight-line distance that ranged from 2 to 9,077 km during their tracking periods (Tables 1-2, Figs ). The total number of filtered Argos locations for the turtles that did not return to SPNWR ranged from 1 to 817 (Table 2). The overall mean traveling rate of individual turtles ranged from 47.6 ± 21 to 70.6 ± 16.2 km/d. Among the eight turtles that did not return again to SPNWR following instrument deployments during the tagging season, four turtles remained within the Caribbean Sea over their entire tracking periods and the other turtles embarked on post-nesting migrations in the Atlantic Ocean after spending d in the Caribbean Sea. The general movements by individual turtles during their post-nesting tracking periods are described below. Turtle #AAC271 Turtle #AAC271 was monitored for 0.1 d and logged only one offshore location (Fig. 17). Based on its single offshore location (Argos LC 1), turtle #AAC271 traveled a minimum straight-line distance of 2 km at a traveling rate of 1.8 km/h. The straight-line distance between the turtle s tagging and offshore location stretched over land areas occupied by SPNWR. The turtle s offshore location was accepted after determining the turtle could have reached the location by traveling at a rate of 2.7 km/h (i.e. < 5km/h), while remaining in waters < 1 km offshore of SPNWR. 47

62 Turtle #XXZ126 Turtle #XXZ126 traveled a minimum straight-line distance of 670 km in the Caribbean Sea, and had an overall mean traveling rate of 57.8 ± 26.5 km/d, during a 15.7 d post-nesting tracking period (Tables 1-2, Fig. 18A). The turtle traveled westward after departing from SPNWR and reached an offshore distance from SPNWR of km at 64º19'41'' W, 17º40'44'' N (Argos LC A) on day 4 of its tracking period. Satellite locations indicate that turtle #XXZ126 traveled mainly eastward toward St. Croix between its tracking days 4-9 (mean traveling rate: 39.8 ± 12.6 km/d). Around mid-day on day 10 of her tracking period (5/30/2008), turtle #XXZ126 was located (Argos LC A) 6.9 km offshore of SPNWR. Thereafter, the turtle s movements were away from SPNWR. On day 11 of her post-nesting tracking period, the turtle was tracked moving southwestward and reached waters > 50 km offshore of St. Croix s west coast. Following these movements, the turtle generally traveled toward the south or southeast in the Caribbean Sea. Turtle #XXZ126 logged a traveling rate of 86.7 km/d on her last tracking day, and its last available satellite location (Argos LC 0) was at 64º50'28'' W, 16º10'52'' N, which is > 170 km south of St. Croix. Turtle #AAR530 Turtle #AAR530 traveled a minimum straight-line distance of 593 km in the Caribbean Sea, and had an overall mean traveling rate of 51.0 ± 28.8 km/d all recorded during a 12.4 d post-nesting tracking period (Tables 1-2, Fig. 18A). The turtle conducted extensive traveling loops over the majority of its tracking period, and generally restricted its movements to areas west of SPNWR, east of 66º00'00'' W, south of Puerto Rico, and 48

63 north of 17º23'30'' N. Turtle #AAR530 remained in waters > 60 km offshore of SPNWR between days 8-12 of its tracking period, and had a traveling rate that ranged from 21.1 to 81.2 km/d. The turtle s last available satellite location (Argos LC 0) was at 65º44'02''W, 17º45'58'' N, which is 88.7 km west of SPNWR. Turtle VI1400 Turtle #VI1400 traveled a minimum straight-line distance of 1,660 km in the Caribbean Sea, and had an overall mean traveling rate of 69.3 ± 20.2 km/d, during a 26.0 d post-nesting tracking period (Tables 1-2, Fig. 18B). Turtle VI1400 was located km southwest of SPNWR by day 3 of its post-nesting tracking period (17 May 2008). Over her next 4 tracking days (18-21 May 2008), turtle #VI1400 traveled east northeastward between Puerto Rico and St. Croix (mean traveling rate: 73.8 ± 13.7 km/d). Turtle #VI1400 conducted extensive traveling loops during its post-nesting tracking days (26-30 May 2008), which were mainly in waters east of St. Croix, south of British Virgin Islands (e.g. Tortola), east of St. Kitts and Nevis, and north of 17º17'00'' N. The turtle s mean traveling rate over its tracking days was 84.1 ± 15.8 km/d. Turtle #VI1400 was located (Argos LC 1) 12.4 km north of SPNWR on day 17 (31 May 2008) of its post-nesting tracking period, and remained < 28 km offshore from SPNWR for the following seven days of her tracking period. The turtle s last available satellite location was received on her tracking day 24 (7 June :23), and was located (Argos LC 2) at 64º 54' 47'' W, 17º 41' 17'' N, which is 1.3 km offshore of 49

64 SPNWR. The turtle s mean traveling rate during her tracking days was 49.8 ± 12.4 km/d. Turtle #XXZ481 Turtle #XXZ481 traveled a minimum straight-line distance of 1,505 km during a 36-day post-nesting tracking period, which included a post-nesting migration in the Atlantic Ocean (Figs. 19, 20A). The turtle traveled west after departing from SPNWR and was located (Argos LC 0) at 65º26'13'' W, 17º54'00'' N on day 2 of its tracking period, which is 57.6 km offshore of SPNWR (Fig. 20A). Turtle #XXZ481 generally traveled eastward and toward St. Croix over the following several days of its tracking period. On its tracking days 8-9, turtle #XXZ481 was located (Argos LC A) < 10 km offshore of SPNWR. The turtle remained < 30 km offshore of SPNWR over days 8-12 of its tracking period, and its mean traveling rate recorded over this period was 34.4 ± 12.7 km/d. On day 12 of its tracking period (5/31/07), turtle #XXZ481 was located (Argos LC A) 0.5 km offshore of SPNWR. The turtle moved north of SPNWR and toward Vieques, Puerto Rico during its tracking day 13 (1 June 2007). On its following tracking day, the turtle was located (Argos LC A) < 3 km offshore of the southern coast of Vieques, Puerto Rico at 07:38 (local time). Argos satellite locations indicate the turtle s following movements were away from Vieques, Puerto Rico to the east or southeast, so nesting attempts by this turtle at Vieques, Puerto Rico may have occurred on the night of 1 June 2007 (tacking day 13). The turtle s satellite location data did not, however, indicate that it was hauled out when it was near Vieques. 50

65 The first indication of turtle #XXZ481 embarking on her post-nesting migration was observed on day 17 (5 June 2007) of its post-nesting tracking period, which was when it began traveling east northeastward and away from St. Croix. The turtle was located ~43 km north of Anguilla on day 20 of its post-nesting tracking period 8 June 2008). The turtle generally traveled east northeastward in the Atlantic Ocean until 6/23/07 (tracking day 36). Turtle XXZ481 s last location (Argos LC A) was at 51º 21' 50'' W, 20º 26' 17'' N. Turtle #XXZ481 had a mean traveling rate of 47.4 ± 24.4 km/d (max: km/d) while tracked on its post-nesting migration (5 June 2007 to 23 June 2007). Turtle #AAV935 Turtle #AAV935 traveled a minimum straight-line distance of 3,910 km during a 60 d post-nesting tracking period, which included a post-nesting migration in the Atlantic Ocean (Figs. 19, 20B). Turtle #AAV935 traveled in an anti-clockwise pattern over day 0-10 d of her post-nesting tracking period (Fig. 20B). After departing from SPNWR, turtle #AAV935 traveled southeastward and was located (Argos LC 1) 87.9 km offshore of SPNWR on her tracking day 2. Thereafter, turtle #AAV935 moved north northwestward and reached waters north of St. Croix by day 4 of its tracking period. Satellite locations indicate that turtle #AAV935 had a mean traveling rate of 52.2 ± 10 km/d and remained > 15 km offshore from SPNWR during its tracking days 4-9. Turtle #AAV935 was located (Argos LC 0) < 5.5 km offshore of the southern coast of Vieques, Puerto Rico on the night of day 10 of its tracking period (3 June 2007). The close proximity of turtle #AAV935 to Vieques, Puerto Rico, and the timing of this 51

66 event falling within the mean internesting interval for leatherback turtles (8-10 d), suggest that the turtle attempted to nest at Vieques, Puerto Rico. Analysis of the turtle s archival time-series data, however, did not show an extended surface time following a series of shallow dives (< 50 m) on 3 June 2007 to 5 June 2007, which was the typically diving pattern recorded for turtles in this study just prior to their return to SPNWR. Turtle #AAV935 first appeared to embark on her post-nesting migration 2 days after it was located near Vieques, Puerto, which was indicated by its movement eastward and away from Vieques, Puerto Rico. The turtle traveled northward and on the east side of the British Virgin Islands to reach the Atlantic Ocean. On her tracking day 15, turtle #AAV935 was located (Argos LC 1) north of 19º0'00''N. While turtle #AAV935 migrated northward in the Atlantic Ocean and was between 20 0'0'' N and 35 0'0'' N, she remained in pelagic waters > 700 km offshore of the United States. The turtle had a mean traveling rate of 68.9 ± 15.2 km/d (max: 98.6 km/d) between 20º0'0'' N and 40º0'0'' N, which was during its tracking days (12-18 July 2007). The turtle s last location (Argos LC 3) was at 72º 30' 00'' W, 40º15'00' N on day 60 of her tracking period and was ~ 60 km offshore of the southern coast of Long Island, NY. Turtle #AAV935 had a mean traveling rate of 44.1 ± 26.3 km/d (range: 8.8 to 71.8 km/d) when it was north of 40º0'0'' N. Turtle #AAG914 Turtle #AAG914 traveled a minimum straight-line distance of 9,077 km the farthest distance recorded in this study over a 131 d post-nesting tracking period, which included a post-nesting migration in the Atlantic Ocean (Figs, 19, 21A). After moving 52

67 offshore from SPNWR, turtle #AAG914 traveled west southwest (Fig. 21 A). Over the next several days, the turtle traveled in a clockwise pattern around St. Croix. On day 5 of its tracking period, turtle #AAG914 approached St. Croix from the east and was located (Argos LC 1) 3.5 km offshore of St. Croix s southern coastline. Thereafter, the turtle continued moving clockwise around St. Croix. On day 6 of its tracking period (22 May 2008), turtle #AAG914 was located 43.1 km northeast of St. Croix. Turtle #AAG914 s mean traveling rate over days 0-6 of its tracking period was 70.7 ± 10.3 km/d. During its tracking days 8-20 (24 May 2008 to 05 June 2008), turtle #AAG914 was generally located in waters east of St. Croix, north of 17 35'0''N, and between the British Virgin Islands, Anguilla and St. Kitts and Nevis (Fig. 21A). The turtle s mean traveling rate during its tracking days 8-20 ranged from 59.9 to km/d (overall mean traveling rate: 81.4 ± 14.6 km/d). Turtle #AAG914 was located (Argos LC 2 and 0) within < 3 km of Virgin Gorda (British Virgin Islands) on day 11 (27 May 2008) and 16 (1 June 2008) of its post-nesting tracking period. Between the days when turtle #AAG914 was located near the British Virgin Islands, the turtle spent time in close proximity to eastern Leeward Islands of the Lesser Antilles, and was located (Argos LC 3) onshore at Orient Beach, St. Martin around mid-day of its tracking day 13 (29 May 2008). The turtle s onshore location (Argos LC 3, accuracy <100 m) at St. Martin on day 13 of its tracking period suggests that it attempted to nest during this time. The turtle s satellite records, however, did not indicate that it was hauled out at a time that corresponded to when it was located near St. Martin, or the British Virgin Islands. 53

68 Around mid-day on its tracking day 21 (6 June 2008), turtle #AAG914 was located in waters < 13 km offshore of St. Croix s northern coastline and was traveling westward. The turtle was located 31.4 km offshore of SPNWR on its following tracking day. Over the next 4 days of its post-nesting tracking period, turtle #AAG914 traveled generally southward at a mean traveling rate of 72.0 ± 7.5 km/d. Turtle #AAG914 traveled as far south as 15º36'47'' N, prior to moving east northeastward and toward Guadeloupe of the Leeward Islands of the Lesser Antilles (Fig. 21A). Turtle #AAG914 was located between the islands of Montserrat and Guadeloupe during day 30 (16 June 2008) of its post-nesting tracking period and was traveling eastward (Fig. 21). Turtle #AAG914 was located east of 60º0'0'' W on day 32 of its tracking period, and was tracked on a pan-atlantic post-nesting migration for the following 101 d. Turtle #AAG914 generally traveled northeastward across the Atlantic Ocean and toward the Azores island. The turtle traveled north of Azores islands and on 26 September 2008 was last located (Argos LC 0) at 10º42'40''W, 44º40'52''N, which is ~220 km offshore of the northwestern coast of Spain. The turtle s mean speed during her pan- Atlantic post-nesting migration (17 June - 26 September 2008) ranged from 29.9 to km/d (overall mean: 69.0 ± 16.2 km/d). Turtle #XXZ123 The turtle traveled a minimum straight-line distance of 7,526 km over a 123 d post-nesting tracking period, which included a post-nesting migration in the Atlantic Ocean. After departing from SPNWR, turtle #XXZ123 traveled north of St. Croix (Fig. 21B). The turtle was located (Argos LC 0) 2.4 km offshore of the southern coast of Vieques, Puerto Rico on day 3 (16 May 2008) of its post-nesting tracking period. The 54

69 turtle traveled westward for the following 3 d and remained < 35 km offshore of Puerto Rico. On day 6 (27 May 2008) of its tracking period, turtle #XXZ123 traveled east northeastward and was < 25 km offshore of Puerto Rico. The turtle was located (Argos LC 0) < 5 km offshore of Puerto Rico on day 9 of its tracking period (27 May 2008). Thereafter, the turtle moved east northeastward and was located (Argos LC 0) < 5 km offshore of Vieques Puerto Rico on day 10 of its tracking period (28 May 2008). Although turtle #XXZ123 s satellite locations suggest that it may have attempted to nest on day 9 or 10 of its tracking period at mainland Puerto Rico or Vieques, Puerto Rico, the turtle s satellite records did not indicate that it was hauled out during these days. Turtle #XXZ123 was traveled east southeastward during its post-nesting tracking days 11-12, and was located > 90 km offshore of St. Croix s east coast on day 12 of its tracking period (30 May 2008). Thereafter, turtle #XXZ123 traveled northward and between the British Virgin Islands and Anguilla and St. Martin/St. Maarten. Turtle #XXZ123 was located north of 19º0'0'' N on day 18 (5 June 2008) of its post-nesting tracking period and was tracked for the following 106 d (5 June - 18 September 2008) on a post-nesting migration. The turtle s mean daily traveling rate that ranged from 37.9 to km/d during its post-nesting migration period (overall mean: 67.7 ± 13.9 km/d). The turtle traveled across the Atlantic Ocean basin, along a similar track as turtle #AAG914. The turtle s last available satellite location (Argos LC 0) was on 18 September 2008 at 13º58'19'' W, 48º05' 06'' N, which is ~660 km offshore of the northwestern coast of Spain. 55

70 Diving Behavior (Satellite Dive Data) Satellite dive data was received from the PTTs of all turtles in this study, with the exception of the PTT deployed on turtle #AAC271 which failed after 0.1 d (Table 2). The number of satellite relayed dive depth records (6-hour blocks) for individual turtles ranged from 14 to 354. The total number of dives logged in these records ranged from 179 to 35,504. The number of satellite relayed dive duration records (6-hour blocks) for individual turtles ranged from 16 to 368. The total number of dives logged in these records ranged from 228 to 36,814. Satellite data indicates the turtles generally conducted dives in the Caribbean Sea to maximum depths ranging from 50 to 250 m (Fig ) and for durations ranging from 16 to 30 min (Fig ). The overall mean relative frequency for maximum dive depths ranging from 50 to 250 m was 0.78 ± The overall mean relative frequency of dive durations ranging from 16 to 30 min was 0.57 ± Satellite data indicates that all turtles, with the exception of turtle #AAQ943, conducted at least one dive in the Caribbean Sea that was > 400 m. Dives by turtles > 400 m were, however, infrequent based on the satellite dive depth relative frequency distributions for each turtle (dive depth > 400 m maximum relative frequency: 0.02, Figs ). All turtles were recorded to have made at least one dive lasting > 40 min in the Caribbean Sea; however, dives > 40 min were infrequent based on the satellite dive duration relative frequency distributions for each turtle (dive duration > 40 min maximum relative frequency: 0.01, Figs ). Satellite time-at-temperature data show the turtles spent the majority (range: %) of their tracking periods in the Caribbean Sea at T A ranging from 24º to 30ºC 56

71 (Figs ). The percent of time spent by turtles at T A > 32ºC in the Caribbean Sea ranged from 0.3 to 9.0%. The lowest T A logged by turtles in the Caribbean Sea was 10º- 12ºC. Internesting Diving Behavior (Archival Time-series Data) Fine-scale analysis of internesting diving behavior was conducted with the archival time-series dive summary data of ten turtles that returned to nest again at SPNWR with their PTTs attached to their carapace (Table 3). The overall mean number of hourly dives made by turtles during their internesting intervals was 3.7 ± 0.9 dives h -1 (range: 2.9 ± 0.5 to 5.3 ± 0.9 dives h -1 ). The overall mean maximum dive depth by individual turtles was 92 ± 39 m (range: 35 ± 57 to 174 ± 79 m). Dives to depths > 250 m were infrequently conducted by turtles and 3 turtles conducted dives > 500 m (maximum dive depth: 524 m; duration 20.8 min). The overall mean dive duration by turtles was 12.6 ± 3.0 min (range: 7.8 ± 7.3 to 15.7 ± 6.0 min, Table 3). All turtles conducted dives lasting > 20 min, and 2 turtles conducted dives lasting > 40 min (maximum dive duration: 47.7 min; maximum dive depth 11 m). The overall mean post-dive surface time was 4.3 ± 1.4 min (range: 2.6 ± 2.2 to 6.5 ± 10.8 min). Eight of ten turtles had surface times > 30 min. The extended internesting surface events (>30 min) were typically recorded during the daytime between 08:00 and 14:00 h. The turtles spent the majority of their internesting interval (range: 63.6 to 83.8%) submerged below 3 m (Table 4), and the majority of dives (range: 75.0 to 94.1%) were V-shaped (Table 3). Mann-Whitney test showed no significant difference (P < 0.05) in mean dive depth, dive duration, post-dive surface time, dive 57

72 frequency (dives h -1 ), and the proportion of V-shaped and U-shaped dives between the field seasons of 2007 and Based on these results, data from 2007 and 2008 was pooled for diel analysis. A significant difference was detected between daytime (05:00-18:59 h) and nighttime (19:00-04:59 h) internesting diving behavior by turtles (Table 5, Fig. 28). Daytime dives were less frequent (3.3 ± 0.9 dive h -1 ) and deeper (104.5 ± 45.4 m) in comparison to nighttime dives (4.2 ± 1.1 dive h -1, t = 3.93, df = 9, P < 0.01; 79.3 ± 33.0 m, t = 5.23, df = 9, P < 0.001). Additionally, daytime dives were longer in duration (13.1 ± 3.3 min) in comparison to nighttime dives (12.0 ± 2.8 min, t = 3.45, df = 9, P < 0.001), and the post-dive surface times of daytime dives were longer in duration (5.8 ± 2.2 min) in comparison to post-dive surface times of nighttime dives (2.7 ± 0.7 min, t = 5.15, df = 9, P < 0.001). All turtles displayed wiggle events or sudden changes in vertical direction during their internesting dives (Table 4, Fig. 29). The proportion of internesting dives containing a wiggle event ranged from 18.1 to 33.1% for individual turtles (overall mean: 25.3 ± 5.9%). A significant difference was detected between the proportion of daytime and nighttime dives that contained a wiggle event, with a higher proportion wiggle events performed during daytime dives (28.0 ± 6.1%) in comparison to nighttime dives (20.5 ± 8.6%, t = 2.6, df = 9, P < 0.03). Gastrointestinal tract temperatures (Archival time-series data) Unfortunately, the majority of archival time-series T GT data (> 75%) obtained in 2007 from turtles #PPQ244, #AAR591 and #XXZ142 contained errors that rendered the 58

73 data unusable for a detailed analysis of T GT and foraging behavior. Usable time-series T GT data was obtained from two turtles in 2007 and six turtles in 2008, but these data were not contiguously recorded throughout the entire internesting interval (Figs ). The total amount of archival time-series T GT recorded for individual turtles at sea ranged from 0.8 to 28.6 d (Table 6). The paired t-test demonstrated that mean T GT of turtles was significantly higher than T A during the T GT monitoring periods (t = 7.77, df = 7, P < 0.001, Table 6, Fig. 38). The mean T GT of individual turtles at sea ranged from 28.1º ± 0.3ºC to 28.7º ± 0.6ºC, and the mean T A experienced by individual turtles ranged from 25.4º ± 3.3ºC to 27.4º ± 1.2ºC (Table 6). Dive Behavior During T GT Monitoring (Archival Time-Series Data) Dive summary files show that individual turtles conducted 59 to 2,006 dives during the T GT monitoring period (Table 7). The mean dive frequency was 3.6 ± 1.0 dives h -1 (range: 2.6 ± 0.7 to 5.1 ± 1.2 dives h -1 ), and the maximum depth of dives typically ranged from 50 to 150 m (overall mean: 90 ± 37). Dive durations ranged from 8.0 ± 7.4 to 15.8 ± 5.8 min (overall mean: 12.4 ± 3.1 min), while post-dive surface times ranged from 2.5 ± 4.0 to 6.8 ± 11.1 min (overall mean: 4.3 ± 6.1 min) (Table 8). V-shaped dives were more common than U-shaped dives (t = 7.2, df = 7, P < 0.001), and the overall mean percent frequency of V-shaped and U-shaped dives was 85.8 ± 5.9% and 14.1 ± 5.9%, respectively. The turtles displayed a diel diving pattern during the T GT monitoring period (Tables 7, 8, Figs ). Daytime dives were less frequent (3.2 ± 0.9 dive h -1 ) and deeper (100 ± 41 m) in comparison to nighttime dives (4.0 ± 1.3 dives h -1, t = 2.5, df 59

74 = 7, P = 0.04, 80 ± 33 m, t = 4.1, df = 7, P = 0.004). In addition, daytime post-dive surface times (5.6 ± 2.3 min) were longer in comparison to nighttime post-dive surface times (2.8 ± 0.8 min, t = 4.5, df = 7, P = 0.003). Interestingly, a paired t-test did not detect a difference between dive durations recorded for turtles during the daytime (12. 4 ± 2.9 min) and the nighttime (12.7 ± 3.4 min, t = 0.5, df = 7, P = 0.62). Ingestion Events in Gravid Leatherbacks Archival time-series T GT data recorded after the leatherbacks were at sea for > 6 h contained sudden fluctuations indicative of ingestion of either prey or seawater for 7 of 8 turtles (Table 9, Figs ). The total number of ingestion events documented for individual turtles ranged from 9 to 65. The overall mean ingestion rate was 0.2 ± 0.3 ingestions h -1 (range: 0.1 to 0.9 ingestions h -1 ), and the overall mean duration of ingestion events (T F -T I ) was 55.8 ± 17.8 min (range: 30.9 ± 14.2 to 77.9 ± 35.6 min). The majority of the documented ingestion events (70.0%) occurred while the turtles were at sea during internesting days 0-4, but ingestions events were also documented while the turtles were at sea between nesting events for 5 to 26 d (Fig. 41). In 2007, turtles #PPQ234 and #AAR864 made a total of 9 and 20 ingestion events, respectively (Table 9, Figs , 41). Ingestion events for these turtles were documented during internesting days 0-2 (Fig. 41). The last ingestion event documented for turtle #PPQ234 was approximately 5 h prior to when its STP was out of detection range and was assumed to have been excreted, which was on day 2 of its 8.9-day internesting interval. The last ingestion event documented for turtle #AAR864 occurred 60

75 approximately 5 h prior to when its STP was out of range of its PTT and was assumed to have been excreted, which was on day 0 (at 00:20) of its 9.1-day internesting period. In 2008, five turtles displayed ingestion events during their T GT monitoring periods (Table 9, Figs , 41). Turtle #AAR264 made 21 ingestion events over days 0-8 of its 10-day internesting interval. The turtle s last documented ingestion event was > 1.3 d prior to when it hauled out to nest again at SPNWR with its STP still transmitting (Fig. 41). Turtle #XXZ465 made 65 ingestion events the highest number of ingestion events recorded in this study while at sea for 28.9 d. The turtle s last documented ingestion event was on its day 26 at sea and 2 d prior to when its T GT monitoring period ended. Turtle #AAC261 made a total of 34 ingestion events, and its last documented ingestion was > 0.7 d prior to when it hauled out to nest again at SPNWR with its STP still transmitting. Turtle # made a total of 31 ingestion events over its 16 d at sea. The turtle s last documented ingestion event was > 11.5 d prior to when it hauled out to nest again at SPNWR with its STP still transmitting. Turtle #AAQ943 made 11 ingestion events during its 8-day internesting interval. The turtle s last documented ingestion event was approximately 0.4 d prior to when its T GT monitoring period ended at sea and was 0.7 d prior to when it hauled out to nest again at SPNWR with its STP still transmitting. Laboratory Ingestion Simulations Laboratory ingestion simulations were conducted to discriminate between fluctuations representative of prey and seawater ingestion in the field data. Significant differences were detected among the integral index values recorded for the laboratory 61

76 ingestion simulations groups (Kruskal-Wallis, H 2,35 = 21.2, P < 0.001, Table 10, Fig. 42). The ingestion simulation groups comprised of 300 g of jellyfish with 200 g of seawater and 500 g of jellyfish had significantly higher (P < 0.017) integral index values in comparison to the group comprised of 650 g of seawater. The group comprised of 200 g of jellyfish with 300 g of seawater had similar integral index values (P 0.017) in comparison to the ingestion simulation group comprised of 650 g of seawater. Based on these results, the mean integral index value of the group comprised of 300 g of jellyfish with 200 g of seawater (482 s -1 ) was used to identify leatherback prey ingestion events (i.e. successful foraging events) in the field data (see below). Evidence of Feeding During the Nesting Season The sudden fluctuations in T GT that were indicative of ingestion events and had an integral index value 482 s -1 were identified as successful foraging events (here after referred to as prey ingestion events), based on the criteria set forth by statistical analysis of integral index values derived from laboratory ingestion simulations. Sudden fluctuations in the T GT records that had an integral index value < 482 s -1 were classified as unidentified ingestion events. The criteria for distinguishing prey ingestions were conservative, so it is possible that unidentified ingestion events may have been comprised of a small amount of prey that was undetectable based on the statistical analysis of my laboratory ingestion simulations (Table 10, Fig. 42). Among the 191 ingestion events that were identified in this study, 43, or 22.5 %, contained temperature fluctuations during their warming periods and were excluded from comparisons with the laboratory ingestion simulations. At least six successful foraging 62

77 events were identified for each of the seven gravid leatherbacks that displayed ingestion events in their T GT record (maximum: 48 prey ingestion events, Table 11, Fig. 30, 31, 33-37, 43). A total of 111 prey ingestion events by leatherbacks were identified in this study, and the overall mean duration of the prey ingestion events was 60.1 ± 12.7 min (range: 36.6 ± 10.2 to 71.1 ± 32.8 min). The overall mean integral index value of prey ingestion events was 1067 ± 195 (range: 795 ± 388 to 1294 ± 650). A paired t-test detected no statistically significant difference between the mean integral index values of prey ingestion events documented during the daytime (1176 ± 108) and the nighttime (858 ± 236, t = 2.56, df = 3, P = 0.08). The mean rate of prey ingestion events by turtles ranged from 0.05 to 0.37 ingestions h -1 (overall mean: 0.11 ± 0.12 ingestions h -1, Table 11). The majority of the identified prey ingestion events occurred during the daytime (87.4 %) and prior to the turtles being at sea for 4 d (69%) (Table 11, Fig. 44). There was no significant difference, however, between the mean prey ingestion event rates (ingestions h -1 ) documented during the daytime (0.09 ± 0.11 ingestions hour -1 ) and the nighttime (0.04 ± 0.04 ingestions hour -1, t = 1.61, df = 6, P = 0.16). The overall mean maximum dive depth of dives associated with prey ingestion events (hereafter referred to as prey ingestion dives) was 187 ± 26 m (Table 12). The prey ingestion dives that occurred during the daytime were typically deeper (overall mean: 212 ± 47 m) than those that occurred during the nighttime (overall mean: 121 ± 64 m). The difference between the mean maximum dive depth of daytime prey ingestion dives was, however, not significantly different in comparison to nighttime prey ingestion dive depth (t = 2.49, df = 3, P = 0.08). 63

78 Prey ingestion events typically started between 1.0 min prior to and 5.0 min after the maximum dive depth was reached (overall mean: 1.5 ± 1.5 min; Fig. 45). The overall mean difference between the maximum depth of prey ingestion dives and the depth where the prey ingestion was detected was 46.7 ± 13.4 m (range: 35 ± 35 to 61 ± 77 m; Table 12). The overall mean depth of turtles at start of prey ingestion events was 137 ± 20 m. The prey ingestion events typically started at deeper depths during the daytime (overall mean: 150 ± 27 m) in comparison to the nighttime (overall mean: 66 ± 61 m; Table 12). A paired t-test, however, did not detect a significant difference between the mean starting depths of daytime and nighttime prey ingestion events (t = 3.02, df = 3, P = 0.06). The maximum depth at the start of prey ingestions during the daytime and nighttime by individual turtles ranged from 200 to 385 m and from 0 to 173 m, respectively. The overall mean duration of prey ingestion dives was 19.1 ± 2.4 min, and the overall mean post-dive surface time of prey ingestion dives was 7.6 ± 4.1 min (range: 3.5 ± 3.0 to 14.7 ± 18.6 min; Table 13). Mean daytime prey ingestion dive durations (20.0 ± 3.1 min) were not statistically different from mean nighttime prey ingestion dive durations (19.0 ± 4.2 min, t = 0.66, df = 3, P = 0.55). Post-dive surface times of prey ingestion dives were typically longer during the daytime (overall mean: 9.4 ± 5.9 min) than the nighttime (overall mean: 2.4 ± 0.4 min), but the difference was not statistically significant (t = 1.9, df = 3, P = 0.15). The overall mean T A at the maximum depth of prey ingestion dives was 21.2º ± 0.9ºC (range: 21.8º ± 3.6º to 23.8º ± 4.1ºC; Table 14). No significant difference was detected between the mean T A at the maximum depth of prey ingestion dives during the 64

79 daytime (20.8º ± 1.4ºC) and the nighttime (23.2º ± 2.4ºC, t =1.79, df = 3, P = 0.17). A significant difference was detected between the T A at the start of daytime and the nighttime prey ingestion events. The start of prey ingestion events during the daytime occurred at cooler T A (22.8º ± 1.5ºC) in comparison to the nighttime (25.6º ± 1.6ºC, t = 3.61, df = 3, P = 0.03; Table 13). The overall mean T A at the start of prey ingestion events was 23.3º ± 1.1ºC (range: 21.8º ± 3.6º to 23.8º ± 4.1ºC; Table 14). The mean ΔT D of prey ingestion events for individual turtles ranged from 0.5º ± 0.1º to 0.8º ± 0.3ºC (overall mean: 0.6º ± 0.2ºC; maximum: 3.3ºC; Table 14). Although prey ingestion events started at significantly cooler mean T A during the daytime in comparison to the nighttime, the mean ΔT D of daytime prey ingestion events (0.6º ± 0.2ºC) was not significantly different in comparison to those that occurred during the nighttime (0.4º ± 0.1ºC, t = 1.13, df = 3, P = 0.34). The mean ΔT D of prey ingestion events during the daytime and nighttime for individual turtles ranged from 0.5º ± 0.2º to 1.0º ± 0.1ºC and from 0.3º ± 0.0º to 0.5º ± 0.1º C, respectively. Prey ingestion events were more commonly associated with V-shaped dives (92.1 ± 11.5 %) than U-shaped dives (7.5 ± 11.5%, Z = 2.30, P < 0.05; Table 15). The overall mean maximum dive depth and duration of V-shaped dives was 189 ± 31 m and 19.2 ± 2.5 min, respectively, while the overall mean maximum dive depth and dive duration for U-shaped prey ingestion dives was 134 ± 68 m and 16.9 ± 4.1 min, respectively. There was a significant difference in the mean proportion of prey ingestion events occurring in the different phases of V-shaped dives (one-way ANOVA, F 3, 24 = 11.1, P < 0.001; Table 16, Fig 46A). The frequency of prey ingestions that started during the descent, bottom and surface phase of V-shaped dives were similar (P 0.008), but the 65

80 proportion of prey ingestion events that started during the ascent phase was higher in comparison to the descent (P < 0.008) and surface phase (P < 0.008). All prey ingestions associated with U-shaped dives occurred during the bottom-phase of dives, but no significant difference was detected between the frequency of prey ingestions among the phases of U-shaped dives (Kruskal-Wallis, H 2, 28 = 6.23, P = 0.10; Table 15, Fig. 46A). Among the prey ingestion dives, 28.8% contained a wiggle event (Table 17). The overall mean difference in time between the start of prey ingestion events and an associated wiggle event (start time of prey ingestion event start time of wiggle event) was 3.6 ± 3.5 min (mean range: 0.3 ± 8.1 to 8.6 ± 11.6 min). The majority of wiggle events associated with prey ingestion dives were documented during V-shaped dives (84.3%), but the proportion of the total number of V-shaped prey ingestion dives that contained a wiggle event was lower than the proportion of U-shaped prey ingestion dives that contained a wiggle event (t = 7.8, df = 3, P = 0.004, Table 17, Fig. 46B). Habitat Associated with Successful Foraging Events Prey ingestion events occurred mainly over deep waters (>1000 m) within the Caribbean Sea (Figs ). The overall mean sea depth where prey ingestion events were documented was 2,501 ± 1,232 m, and ranged for individual turtles from 2,118 ± 830 to 3,281 ± 799 m (Table 18). Prey ingestion events were documented up to 170 km offshore of St. Croix (Fig. 49B). Interpolated tracking analysis placed prey ingestion events for two turtles (#AAR264 and #AAC261) onshore at St. Croix. This error was caused by the start time of the ingestion events occurring between two consecutive Argos satellite locations that were spatially separated by St. Croix. The mean straight line 66

81 distance between the turtles interpolated locations of their prey ingestion events and SPNWR was 56.8 ± 38.4 km (Table 18). No SSH anomalies were detected within 200 km of St. Croix during the T GT monitoring periods in 2007 (18-24 May 2007). All prey ingestion events for turtle #PPQ234 and #AAR864 in 2007 occurred within 100 km of St. Croix, and therefore showed no associated with SSH anomalies (Fig. 47). A positive and a negative SSH anomaly was detected < 100 km south of St. Croix during the T GT monitoring periods in 2008 (12 May - 12 June 2008; Fig. 52). The anomalies were observed moving west northwest in the Caribbean Sea from islands in the arc of the Lesser Antilles. The edge of the negative sea surface height anomaly was observed west of St. Croix by 28 May 2008, while the edge of the positive sea surface height anomaly was observed west of St. Croix on 25 June A satellite-derived colored image of the diffuse attenuation coefficient at a wavelength of 490 nm shows an increase in light attenuation south of St. Croix between 66º42'32'' W, 63º47'51'' W, 17º22'12'' N, and 15º42'53'' N during May 2008; a sign of increased productivity (e.g. increased chlorophyll-a) that corresponded to the area covered by the pair of SSH anomalies during this period (Fig. 53). Prey ingestion events for turtle #ARR264 and #XXZ465 between May 2008 mainly occurred < 25 km north of St. Croix and did not appear to be associated with SSH anomalies, but turtle #AAR264 and #XXZ456 had five and two prey ingestion events, respectively, that occurred < 60 southwest of St. Croix and appeared to be associated with the edge of the negative SSH anomaly (Fig. 48). Between 21 May 2008 and 27 May 2008, turtle #XXZ465 had at least six prey ingestion events that appeared to 67

82 be associated with the edge of the positive SSH anomaly (Fig. #49A), while turtle #AAC261, # , and #AAQ943 had four, six, and 15 prey ingestion events, respectively, that appeared to be associated with the edge of the negative SSH anomaly (Fig. 50B, 51). During the same period, turtle #XXZ465, #AAC261, #AAQ943 and had three, seven, and one prey ingestion events, respectively, that did not appear to be associated with the pair of SSH anomalies (Fig. 49A, 50B, 51B). Between 28 May 10 June 2008, all 14 prey ingestion events identified for turtle #XXZ465 appeared to be associated with the edge of the positive SSH anomaly, which was < 250 km east and southeast of St. Croix (Fig. 49B, 50A). The locations of prey ingestion events were associated with satellite-derived sea surface chlorophyll-a concentrations that ranged from ± to ± mg m 3 (2007) and from ± 0.28 to ± 0.32 mg m 3 (2008; Table 18, Fig. 54). Three turtles had one or more prey ingestion event locations for which no chlorophyll-a concentration data were available: turtle #AAR264 (n = 3), turtle #AAR264 (n = 1) and #AAC261 (n = 1). The 75 % kernel density estimate of leatherback prey ingestion locations shows three main internesting foraging hot-spots (Fig. 55). One foraging hot-spot was centered approximately 20 km northwest of SPNWR and extended out from SPNWR 75 km to the west, 55 km to the north, 45 km to the south, and 61 km to the east (all values rounded to nearest km). The other main foraging hot-spots were centered approximately 140 km west southwest of SPNWR and approximately 95 km northeast of SPNWR. 68

83 Unidentified ingestion events Unidentified ingestion events occurred mainly during the daytime (83.2 %; Table 19), but there was no significant difference between daytime and nighttime rates of unidentified ingestion events (ingestions h -1 ; t = 1.28, df = 6, P = 0.25). Unidentified ingestion events occurred more frequently during V-shaped dives than U-shaped dives (t = 9.7, df = 7, P < 0.001; Table 20). No significant difference was detected between prey ingestion dives and unidentified ingestion dives for mean maximum dive depth (t = 1.0, df = 6, P = 0.34), dive duration (t = 2.0, df = 6, P = 0.10), post-dive surface time (t = 0.62, df = 6, P = 0.56), and T A at maximum dive depth (t = 0.58, df = 6, P = 0.60; Table 12-14, 20). No significant difference was detected between T A at the start of unidentified ingestion events and prey ingestion events (t = 0.74, df = 6, P = 0.49). A significant difference was detected between the proportion of dive phases associated with unidentified ingestion events for V-shaped dives (Kruskal-Wallis 2, 28, H = 17.4, P < 0.001), but not for U-shaped dives (Kruskal-Wallis 2, 28, H = 4.47, P 0.23) (Table 21, Fig. 56). Unidentified ingestion events occurred more commonly during the ascent phase compared with other dive phases (P < 0.008) of V-shaped dives, but the proportion of unidentified ingestion events was similar during decent, bottom and surface phase of V-shaped dives (P 0.008). Among the unidentified ingestion dives, 28.8% contained a wiggle event. The majority of the wiggle events (78.3%) occurred during V-shaped dives (Table 22), but there was no significant difference between the proportion of V-shaped and U-shaped unidentified ingestion dives that contained a wiggle event (t = 3.81, df = 2, P = 0.06). The 69

84 overall mean difference in time between the start of unidentified ingestion events and an associated wiggle event (start time of unidentified ingestion event start time of wiggle event) was 3.6 ± 3.5 min (mean range: 0.3 ± 8.1 to 8.6 ± 11.6 min), and no significant difference was detected between the start time of wiggle events and the start time of prey and unidentified ingestion events (t = 0.60, df = 3, P = 0.59). DISCUSSION Field Observations The deployment of PTTs and STPs did not appear to impact the welfare of turtles in this study. The majority of study turtles (11 of 19) returned to SPNWR and displayed normal signs of nesting behavior during the tagging season. Among the turtles that did not return to nest again during the tagging season, three turtles were tracked on postnesting migrations after remaining in the Caribbean Sea for a period ranging from 15 to 32 d and five turtles had PTTs fail after a period ranging from 0.1 to 26 d while they were still within the Caribbean Sea. Two turtles (#AAR530 and #AAV935) shed their PTTs in 2007 and one turtle (#KL56) shed its PTT in Other reasons for PTT failure after < 100 d include salt-water switch failure or damage to the satellite antennae on the PTTs (see Hayes et al., 2007). The nesting beach was patrolled hourly by WIMARCS, Inc. between 20:00 and 05:00 h from early April to 7-10 d after the last nest of season was documented, which is typically late July/early August, so it is unlikely that nesting attempts by tagged leatherbacks were missed at SPNWR. The turtles that did not return to SPNWR after instrument deployments during the tagging season and had PTTs that failed while they 70

85 were in the Caribbean Sea were therefore assumed to have embarked on post-nesting migrations shortly after their PTTs failed. Eight turtles returned to nest after 8.1 to 10 d, but three turtles (turtle #XXZ465, #KL56, and # , Table 1) had extended internesting intervals and returned to nest after a period of 16.0 to 30.8 d. It is possible that these turtles nested elsewhere in the Caribbean Sea prior to returning to SPNWR, since leatherbacks display interseasonal nesting shifts between SPNWR and Manchenil Bay beach, St. Croix, as well as between SPNWR and Playa/Resaca Brava, Isla Culebra, Puerto Rico (Eckert et al., 1989a; Boulan et al., 1996) and beaches on St. Kitts and Nevis (the WIMARCS, Inc., personal communication). Turtle #XXZ465, who had an extended internesting interval of 28.9 d, appeared to have attempted to nest at Manchenil Bay beach, St. Croix one day prior to returning to SPNWR, based on high quality Argos satellite locations (LC 1) and archival surface times (> 90 min). Archival surface times > 90 min also suggest that this turtle attempted to nest on tracking day 11 and 20, corresponding to interesting intervals of 11 and 9 d, respectively. The satellite records for turtle #XXZ465 and the other two turtles with extended internesting intervals, however, did not show a pattern of high quality locations indicative of nesting attempts prior to returning to SPNWR. Interestingly, each turtle with extended internesting intervals had prior records of 15 to 29.0 d separating consecutive nesting attempts at SPNWR. Eckert et al. (1989a) report that once a leatherback makes the switch to a different nesting beach the turtle continues to nest at that beach. This makes the extended internesting intervals recorded for turtles during this study and prior to this study an intriguing nesting pattern. High 71

86 quality Argos locations (LC 3 and LC 2) were not successfully obtained on a daily basis from any turtle in this study. Use of Argos locations without associated error estimates (e.g. LC 0 and A) may have caused us to miss nesting attempts at islands other than St. Croix by the turtles with extended internesting intervals. Five turtles tagged in 2007 displayed 2-year remigration intervals and laid 4 nests at SPNWR during the 2009 nesting season (the WIMARCs, Inc., personal communication). The instrument attachment sites had healed and were in good condition for all 2-year remigrants (the WIMARCs, Inc., personal communication). These observations demonstrate that the instruments deployed in this study did not have a longterm impact upon welfare of the turtles. Turtles #AAR530 and #AAV935, which did not return to nest in 2007 and had PTTs that failed after a period of d, were among the five remigrant turtles documented at SPNWR during the 2009 nesting season. Turtle #AAV935 s PTT was recovered ~60 km offshore from Long Island, NY by a scallop-fisherman. These observations confirm that PTT failure for these turtles was due to PTT shedding and not mortality. General Internesting Movements and Diving Behavior Turtles for which we have complete records of internesting movements, moved > 50 km offshore from St. Croix and traveled a minimum straight-line distance of several hundred kilometers during their internesting interval. These results are analogous to previous studies on leatherback internesting movements at SPNWR (Eckert, 2002) and other nesting beaches in the Atlantic Ocean (Eckert et al., 2006; Georges et al., 2007; 72

87 Witt et al., 2008; Bryne et al., 2009). The total surface area of the kernel estimated internesting habitat for turtles (25% UD: 6.0 km 2 ; 50% UD: 25.1 km 2 ; 70% UD: 60.0 km 2 ) highlights the extensive internesting movements. Study turtles were tagged while laying their 3 rd to 5 th documented clutch of the nesting season at SPNWR, and only two turtles returned to lay an additional nest at SPNWR following PTT removal. Since the turtles were tagged toward the end of their nesting season, internesting movements recorded in this study should not be considered a reflection of the turtles behavior over the entire nesting/breeding season. The internesting movements by turtles in this study may be more extensive compared to movements earlier in the nesting season as reported by Bryne et al. (2009) for leatherbacks nesting at Dominica, West Indies. The retrieval of archival PTTs after deployments allowed for fine-scale analysis of the turtles internesting diving patterns. The archival dive data obtained from turtles that returned to nest at SPNWR (n = 10) indicate they spent most of the internesting period (73.9 ± 5.7%) submerged below the top 3 m of the water. These records indicate that leatherbacks dive continuously during the internesting interval, a pattern previously documented for leatherbacks during the nesting season in the Atlantic (Eckert et al., 1989b) and Pacific (Southwood et al., 1999; Southwood et al., 2005; Wallace et al., 2005), as well as during post-nesting movements in the Atlantic and Indian Ocean (Sale et al., 2006). Internesting dives were typically V-shaped and ranged from m, but several turtles conducted dives > 500 m. This pattern is consistent with observations from previous studies of gravid leatherbacks at St. Croix (e.g. Eckert et al. 1986; Eckert et al., 1989b). 73

88 An internesting diel diving pattern was observed in this study, with turtles conducting deeper, longer and less frequent dives during the daytime compared to the nighttime. A similar diel diving pattern during the interesting interval was found for turtles nesting at St. Croix (Eckert et al., 1986; Eckert et al., 1989b) and at Grenada (Myers and Hayes, 2006) and is suggested to reflect foraging on the DSL as it vertically ascends to shallower waters at nighttime (Eckert et al., 1989b, Myers and Hayes, 2006). In my study, both daytime and nighttime dives contained wiggle events, which are commonly interpreted as foraging behavior (Bost et al., 2007). Additionally, seven turtles displayed sudden fluctuations in T GT indicative of feeding during both daytime and nighttime. The data support the idea that leatherbacks forage on the DSL during the internesting interval, but not that foraging is restricted to nighttime when the DSL is closer to the surface (see below: Evidence of Foraging During the Nesting Season). Based on archival data collected from turtles that returned to nest at SPNWR after 8-10 d (n = 8), dives were on average 31 m deeper and 2.6 min longer compared with dives recorded for gravid turtles (n = 6) in previous studies at SPNWR (Eckert et al., 1989b). Interestingly, the turtles in this study with internesting intervals ranging from 8.1 to 10.0 d had a mean nighttime dive depth (80 ± 36 m) that was nearly 60 meters deeper than the mean nighttime dive depth (21 ± 21 m) recorded for gravid leatherbacks (n = 6) with comparable internesting intervals at SPNWR (Eckert et al., 1989b). The exact reasons for the disparities of nighttime internesting diving behavior between turtles from the same nesting population could be caused by various factors, acting individually or jointly. Possibilities include differences in energy stores (see Bryne et al., 2009) and preferences for different thermal environments (Wallace et al., 2005; Fossette et al., 74

89 2009), variation in swimming ability of turtles caused by different instrument attachment techniques (i.e. harness attachment vs. direct-carapace attachment) (see Fossette et al., 2008a; Bryne et al., 2009), and variations in prey availability and distribution for turtles, both spatially and temporally, during the nesting season (see below: Evidence of Foraging During the Nesting Season). Characterizing Gelatinous Prey Ingestion Events Captive feeding trials with leatherbacks were not feasible for this investigation, since leatherbacks are extremely difficult to care for in captivity (see Jones, 2009) and the size of STP instruments precluded use with small animals. As an alternative to captive feeding trials, I used laboratory simulations to characterize ingestions of gelatinous prey. Laboratory ingestion simulations have been demonstrated to be a useful tool for assessing a study animal s stomach temperature fluctuations following ingestion events (Wilson et al., 1992; Wilson et al., 1995). Seawater and squid have been used in previous laboratory ingestion simulations (Wilson et al., 1992; Wilson et al. 1995), but my laboratory ingestion simulations were the first to include a species of jellyfish (Aurelia aurita), which was chosen to represented a typical prey item for a leatherback turtle in the Caribbean Sea. A salient finding of my laboratory ingestion simulations was that jellyfish and seawater could be distinguished from each other based on the integral index values of temperature fluctuations a standardized value of the TRIM integral (Grémillet and Plös, 1994). Integral index values of ingestion simulations that included at least 300 g of jellyfish were statistically different from simulations with only seawater. I was unable to 75

90 distinguish between jellyfish and seawater ingestion simulations when small amounts (i.e. < 300 g) of jellyfish were used. This important limitation translated into a very conservative analysis of field data, in which very small fluctuations in T GT were characterized as unidentified ingestion events. The laboratory ingestion simulations performed with seawater have been shown to slightly overestimate the integral values of seawater ingestions by free-living animals due to rapid passage of seawater from the animal s stomach to the small intestine (Wilson et al., 1992). It is reasonable to assume that seawater would pass through the stomach more rapidly than prey. Unfortunately, gastric motility and stomach retention times for prey and seawater are not known for leatherbacks. If seawater passes from stomach to small intestine prior to the stomach returning to pre-ingestion temperatures, the critical integral index value established by my laboratory simulations for indentifying prey ingestions is conservatively high and prey ingestion rates documented in this study are, albeit slightly, underestimated. Ingestion Events During the Internesting Interval Initial fluctuations in T GT occurred as turtles entered the surf following instrument deployments on the nesting beach. These fluctuations were likely caused by the turtle drinking seawater to either rehydrate (Sato et al. 1994), move the STP down their esophagus (Southwood et al., 1995), or to decrease elevated body temperatures due to the high activity associated with terrestrial nesting (see below). Temperature detected by the STP declined steadily during the first ~6 h at sea (Fig. 8) and an asymptote for TRIM 76

91 analysis, as defined by my criteria, could not be determined. Therefore, these initial fluctuations were excluded from analysis of foraging behavior. The criteria for identifying prey ingestion events in the T GT data of free-swimming leatherbacks were established using laboratory ingestion simulations. Ingestion events that did not meet criteria for prey ingestion were characterized as unidentified and may represent ingestion of either prey or seawater. One potential reason for leatherbacks ingesting seawater during the nesting season is for thermoregulation (i.e. to lower their body temperature). One argument against this is that leatherbacks are capable of altering their body temperatures (T B ) by changes in behavior and blood circulation in the absence of ingestion events (Paladino et al., Southwood et al., 2005; Wallace et al., 2005). For example, leatherback T B decreases with decreases in activity levels at depths > 20 m in the tropics (Southwood et al., 2005). In addition, Wallace et al. (2005) reported that leatherbacks increase the proportion of time spent in water temperatures (T W ) 24ºC during the first several days of the internesting interval, and inferred that T W 24ºC provide a heat sink for turtles following strenuous nesting activity and prevent them from overheating. Changes in blood circulation by leatherbacks have not been directly observed in the field due to difficulties in obtaining this data; however, Paladino et al. (1990) suggested that leatherbacks can alter their body temperature by regulating blood flow to peripheral tissues, based on a mathematical model. Southwood et al. (2005) reported that the thermoregulation model proposed by Paladino et al. (1990) closely predicted the gradient between leatherback T B and T W, as well as between leatherback T GT and subcarapace temperature (T S ), during the internesting interval in the tropics. These data provide evidence that leatherbacks increase blood flow to peripheral tissues to 77

92 prevent overheating while at low latitudes and decrease blood flow to peripheral tissues to minimize heat loss while at high latitudes (Southwood et al., 2005). It is also likely that their elongated, poorly insulated front flippers act as thermal windows for lowering T B, just as the dorsal fin and fluke of bottlenose dolphins (Tursiops truncatus) is used for convection of heat via blood during elevated T B (Noren et al., 1999; Meager et al., 2002; Meager et al., 2008). Another argument against leatherbacks drinking seawater to thermoregulate during the nesting season is that leatherbacks would incur an energetic cost to excrete additional salt absorbed from the intake of seawater (i.e. use of salt glands for maintenance of ion and water balance (reviewed by Lutz, 1997)). It seems unlikely that leatherbacks would choose to thermoregulate through a behavior that requires the use of additional energetic resources when they have demonstrated the ability to thermoregulate by energetically cheap methods, such as lowering activity levels and resting in relatively cool waters at depth (Southwood et al., 2005). Attempts to minimize additional salt intake through seawater ingestion have been demonstrated during feeding events by sea turtles (i.e. conservation of energy by avoiding seawater ingestion) (Lutz, 1997), supporting the idea that sea turtles normally avoid ingestion of seawater. Sea turtles have keratinized spines called papillae (see Fig. 6) lining their esophagus and they are used to trap food while they forcefully expel seawater out of their nares or mouth (Lutz, 1997). A second potential reason for leatherbacks to ingest seawater during the nesting season is to re-hydrate. Recently, Wallace et al. (2006) reported that water constituted 67% of the total wet mass of a leatherback egg. Based on the average wet mass of leatherback eggs (~76 g) (Wallace et al. 2006) and the average number of eggs in a clutch 78

93 (~80 eggs) (Miller, 1997), the amount of water invested by leatherbacks into a single clutch is 4.1 kg or 4.1 L. This amounts to 16.3 to 32.6 L of water invested by leatherbacks into egg formation during a nesting season, based on leatherbacks laying four to eight clutches. Studies of water sources for egg production by leatherbacks have not been attempted, but it seems reasonable that water requirements for egg formation could be provided by reserves, energy stores (i.e. metabolic water) or, if they are feeding, gelatinous prey (Doyle et al., 2007). I conclude that ingestion events characterized as either prey or unidentified in this study are reflective of opportunistic feeding during the internesting interval. In addition to the aforementioned arguments against leatherbacks drinking seawater, this conclusion is supported by Argos satellite locations that display extensive internesting movements (>100 km), reflective of using energy to search for prey. Furthermore, the depth of daytime and nighttime ingestion events characterized as prey and unidentified revealed a pattern, although not statistically significant, of foraging on vertically migrating animals. Lastly, the ingestion events in this study did not appear to be reflective of attempts to thermoregulate because they resulted in only small, evanescent changes to the T GT and were typically followed by extended periods in warm surface waters (< 3 m). Evidence of Foraging During the Internesting Interval The sudden fluctuations in T GT indicative of prey ingestion events occurred primarily during the first several days of the internesting interval (0-4 d). Behavioral data collected using IMASENs/magnetic hall sensors and time-depth recorders (TDRs) deployed on leatherbacks nesting at French Guiana and Grenada also provide evidence 79

94 of foraging early in the interesting interval, but these instruments remained attached to turtles for a maximum of only 2 d (Myers and Hayes, 2006; Fossette et al., 2008b). Short T GT monitoring periods (< 2.5 d) were a concern for three turtles in my study (e.g. turtle #PPQ234, #AAR863, and #AAV935), but T GT monitoring periods for the other five turtles lasted for 95 to 100% of their time at sea. The concentration of prey ingestion events during the first several days of the internesting interval and absence of prey ingestion events toward the last days of the internesting interval may reflect a shift from foraging efforts to reproductive activities in the final days at sea prior to a nesting attempt. There were no apparent shifts in internesting diving behavior in this study to support this idea; however, previous studies have reported changes in leatherback diving behavior toward the end of the internesting interval, characterized by a decrease in dive effort and increase in dive variability at French Guiana (Fossette et al., 2007), consistently shallower and shorter dives at Costa Rica (Southwood et al., 2005), and lower variability in swim speeds at St. Croix (Eckert et al., 2002). It is also possible that STPs may have moved from the stomach into the small intestine after a period of several days (as reported in Southwood et al., 2005). Under this scenario, the STP would be unable to detect ingestion events but would continue to transmit T GT data to the PTT (e.g. turtle # , Fig. 36, 41). A final explanation for the temporal pattern observed in prey ingestion is that patchy prey distribution resulted in a decrease in foraging success as turtles initiated return to nesting beach in the final days of the internesting interval. 80

95 Behavior Associated With Feeding Events Previous studies have suggested that the diel diving pattern displayed by leatherbacks during the internesting interval in the Caribbean Sea is reflective of foraging on the DSL as it vertically migrates from deeper water during the day to relatively shallow water at night (Eckert et al., 1989b; Myers and Hayes, 2006). Recent work has demonstrated the presence of vertically migrating DSLs near the internesting habitat occupied by turtles from the St. Croix population (Rovira-Peña, 2006). Data collected by 76.8 khz acoustic Doppler current profiler (ADCP) and Trucker trawl nets detected a persistent DSL between Puerto Rico and the Dominican Republic (~55 km offshore from the west coast of Puerto Rico) at the Central Mona Passage station, as well as at insular slope waters of Puerto Rico (Rovira-Peña, 2006). The DSLs detected west of Puerto Rico was comprised primarily of copepods and euphausiids and ascended from deep waters (400 m) around dusk and descended from shallow waters (0-150 m) around dawn. The nighttime prey ingestions dives in this study (mean maximum depth: 121 ± 64 m) correlate to the relatively deep depths occupied by the DSL at the Mona Passage at night (0-150 m) (Rovira-Peña, 2006). One potential reason for the discrepancy between the daytime depth of prey ingestion events (mean maximum dive depth: 212 ± 47) and the daytime location of the DSL at the MC (~400 m), is that leatherbacks foraged on siphonophores that vertically descended to daytime depths located m above the DSL detected at the Mona Passage (Rovira-Peña, 2006). Siphonophores are a known leatherback prey item (Hartog, 1980) and Rovira-Peña (2006) reported they were collected at the Mona Passage during trawls in October at three depths layers (0-60 m; m; m). Unfortunately, the diel behavior and vertical distribution of 81

96 siphonophores could not be thoroughly described because of constraints for conducting deep water trawls (maximum trawl depth 180 m; seafloor at station: > 450 m) (Rovira- Peña, 2006). It is possible that siphonophores were not detected as a second, vertically migrating DSL that maintain shallower daytime depths than the DSL reported by Rovira- Peña (2006) because the ADCP frequency was several times higher than the most useful frequency (24.5 khz) for detecting siphonophores (Warren et al., 2001; Rovira-Peña, 2006). In addition, siphonophores may have been present in low numbers at the Mona Passage, which would result in lack of detection of a second DSL. Observations offshore San Diego, California confirmed that siphonophores, in some cases, maintain daytime depths several hundred meters above aggregations of myctopids (Barham, 1963). A similar partitioning of habitat may exist at the Mona Passage between the DSL detected by Rovira-Peña (2006) and siphonophores because myctopids were constituents of the DSL. Another potential reason for vertical separation at the Mona Passage between siphonophores and the DSL is that copepods and euphausiids vertically descend to depths below siphonophore aggregations as a predator avoidance strategy (Purcell, 1981, 1997; Robison et al., 1998; Hays, 2003). Previous studies have reported DVM by species of siphonophores in the Western Pacific (Braham, 1966; Alvariño, 1967; Robinson et al., 1998; Silguero and Robinson, 2000), and the North Atlantic (Pugh, 1977; Mackie et al. 1987; Youngbluth et al., 1996; Pugh, 1999), as well as the Mediterranean (Laval et al. 1989) and Adriatic Sea (Lučić et al., 2000). Migration by siphonophores in the Caribbean Sea may be responsible for the 82

97 diel dive behavior by leatherbacks and prey ingestion events being typically deeper during the daytime than the nighttime. Other potential gelatinous prey for leatherbacks in the Eastern Caribbean Sea include the Australian or white-spotted jellyfish (Phylorhiza punctata) (Garcia and Durban, 1993), and pink meanie (Drymonemea dalmatinum) (Williams et al., 2001) and the moon jellyfish (Aurelia aurita) (Iverson and Skinner, 2006). Little is known about the behavior of these organisms in the Eastern Caribbean Sea because they are difficult to census by net tows due to their fragile body structures and they have patchy distributions. Strong correlations between leatherback foraging and specific prey will require additional studies on the distribution and behavior of gelatinous animals in the Caribbean Sea. Future studies to address such issues in the Caribbean Sea should include echosounders with frequencies that are most useful for the detection of gelatinous animals and/or ROVs (remotely operated vehicles) with video cameras to allow for in situ observations of animals. I expected that leatherbacks would adopt an energetically efficient strategy of foraging primarily at night, since their prey would be closer to the surface at this time based on the normal pattern of DVM by DSLs (Hays et al., 2003). The similar rates of prey ingestion events for daytime and nighttime did not, however, support the hypothesis that gravid leatherbacks forage primarily at nighttime (Table 11). Rather surprisingly, daytime prey ingestion rates (range: 0.04 to 0.35 prey ingestions h -1 ) were typically higher than during nighttime (range: 0.0 to 0.11 prey ingestions h -1 ), although the difference was not statistically significant. 83

98 The diel dive pattern observed for leatherback foraging dives in this study was not particularly strong. Daytime foraging dives were typically deeper than nighttime foraging dives, and the depth at the start of foraging events tended to be deeper during the day compared with night; however, neither of these differences was statistically significant. Caution is warranted for interpretation of these results because three of seven turtles were excluded from statistical analysis due to low sample size for nighttime foraging events (< 3 events). The lack of a strong diel signature for foraging behavior and prey ingestion may be due to limited availability of suitable prey and/or low prey encounter rates for leatherbacks at their internesting habitat. V-shaped dives are often interpreted as representing exploration or travel by sea turtles (Eckert et al., 1989b; Hochscheid et al., 1999; Houghton et al., 2008 ), penguins (Le Bouf et al., 2000; Ropert-Coudert et al., 2000; Shreer et al., 2001) and seals (Thompson and Fedak, 1993), whereas U-shaped dives are interpreted as a reflection of foraging by air-breathing predators at a maximum depth where prey are assumed to be located (Thompson et al. 1993; Le Bouf et al., 2000; Fossette et al. 2007; Fossette et al., 2008b). Pelagic foraging V-shaped dives were recently confirmed, however, in thickbilled murres (Uria lomvia) (Elliott et al., 2008), while foraging events during V-shaped dives have also been suggested for other species of seabird, such as the Royal penguin (Eudyptes schlegeli) and Gentoo penguin (Pygoscelis papua) (Schreer et al., 2001), as well as marine mammals, such as harbor seals (Phoca vitulina) (Lesage et al., 1998). Evidence of foraging by leatherbacks in my study was most commonly documented during dives that were deep (> 100 m), long (19.2 ± 2.5 min), and V- shaped. U-shaped dives associated with foraging were also deep (> 100 m) and long (

99 ± 4.1 min), but were less frequently (7.9 ± 11.5 %) associated with foraging compared to V-shaped dives. Previous investigations of the internesting diving behavior of leatherbacks at St. Croix have led other researchers to suggest that V-shaped dives by leatherbacks were foraging dives (Eckert et al., 1986; Eckert et al., 1989b), and results from my study support this conclusion. In contrast to the diving pattern associated with foraging by leatherbacks in this study, foraging dives by gravid leatherbacks at French Guiana are typically shallow (18.1 ± 6.3 m), short (7.5 ± 2.8 min), and either U-shaped or W-shaped (Fossette et al., 2007; Fossette et al., 2008b). The local bathymetry and differences in prey distribution and concentration can largely explain disparities between leatherback diving patterns associated with foraging at St. Croix (this study) and French Guiana (Fossette et al., 2007; Fossette et al., 2008b). At French Guiana, leatherback movements were restricted to the coastal shelf and U-shaped and W-shaped dives resulted in increased time spent at shallow depths (< 50 m) in highly productive areas, influenced by local rivers and the Amazon river, where leatherback prey, such as Stomolophus sp. and Aurelia sp. (Eisenberg and Frazer, 1883; Grant et al., 1996), have been documented in great numbers (Fossette et al., 2009). In contrast, leatherbacks from the St. Croix nesting population were monitored in easily accessible (< 6 km offshore from SPNWR, see Fig. 2), deep waters (> 200 m) where they targeted and captured vertically migrating prey associated with the DSL. Based on phytoplankton studies in the Caribbean Sea (Hargraves et al., 1970, Marhsall 1973) and trophic levels, gelatinous prey abundance in the oceanic waters of Eastern Caribbean Sea is likely to be sparse compared to coastal areas of north eastern 85

100 South America. Under these circumstances, a V-shaped dive pattern may provide greater opportunity for encountering prey than a U-shaped or W-shaped diving pattern. Although leatherbacks in this study were found to have wiggle events associated with feeding, they occurred in less than one-third of all documented prey ingestion dives and typically as single events. In contrast, leatherbacks at French Guiana display wiggle events during the majority of foraging dives (62.7%) and foraging dives typically contain multiple wiggle events (2.4 ± 1.0 wiggles per dive) (Fossette et al., 2008). The multiple wiggle events by leatherbacks during foraging dives at French Guiana may reflect the turtles locating prey slightly above and below their swimming depth as the move horizontally along the seafloor. There was a time lag of 3.6 ± 3.5 min between the wiggle events and the detection of the foraging events by the STPs in this study. This time lag seems reasonable due to leatherbacks having a long (> 2 m) esophagus. To my knowledge, however, no information exists on the passage time of food through the alimentary canal, specifically from the mouth to the stomach, of adult leatherback turtles. This makes it problematic to determine the specific depth of prey ingestion when using T GT to detect foraging events (see below Bioenergtics). Habitat Associated With Feeding Events Areas of high biological productivity serve as good foraging areas for all sorts of predators, ranging from fish to fisherman (Santos 2000; Bost et al., 2009). Biological productivity is not uniform throughout the world s oceans and is affected by abiotic factors, such as light and nutrients, as well complex and dynamic physical processes, such as eddies, fronts, and upwelling, that operate across a range of time and space. 86

101 Meso-scale eddies result in disturbances to productivity at horizontal scales ranging from tens to hundreds of kilometers. They are ubiquitous throughout the ocean and are frequently formed through meanders, or instabilities, in such ocean currents as the Gulf Stream (Lee et al., 1991), the Northern Brazil Current (Richardson, 2004; Morell et al., 2006) and the Kuroshio Current (Nakata et al., 2000). In the Northern Hemisphere the Coriolis affect results in eddies rotating counterclockwise (cyclonic) to bring deeper waters toward the surface (i.e. upwelling), and are identified in satellite-derived data as negative SSH anomalies due to their cooler, higher density core waters. Upwelling associated with cyclonic eddies result in regions of increased phytoplankton and cholorophyll-a concentrations and higher productivity (Williams et al., 1998; Mizobata, 2002). Anticyclonic eddies rotate clockwise, move surface waters to deeper depths (i.e. downwelling), and are identified as positive sea surface height anomalies due to their warmer, lower density core waters (Mizobata, 2002; Corredor et al., 2004). At the core of anti-cyclonic eddies, downwelling results in lower levels of nutrients, phytoplankton, and chlorophyll-a in the euphotic zone, while high nutrient levels, chlorophyll-a and phytoplankton are found at the edges of these eddies (Mizobata, 2002). Satellite-telemetry studies have led several researchers to suggest that leatherbacks forage at hot-spots associated with eddies and fronts in the Atlantic Ocean (Ferraroli et al., 2004; Hays et al., 2004; Eckert et al., 2006), Indian Ocean (Luschi et al., 2003) and Pacific Ocean (Markman and Schwartz, 2005). Polovina et al. (2005) suggested that pelagic stage loggerhead sea turtles also forage at the edges of productive 87

102 eddies in the Pacific Ocean, based on satellite tracks of turtle movements and remotely sensed oceanographic data. Studies have confirmed that satellite-derived SSH anomalies in the Caribbean Sea are a result of both cyclonic and anti-cyclonic eddies that originate from the Northern Brazil Current (NBC) in the Atlantic Ocean (Richardson, 2004; Morell et al., 2006). The increased light attenuation at the wavelength of 490 nm observed at the core of the negative SSH anomaly and at the edges of the positive SSH anomaly located near St. Croix during the 2008 field season is indicative of high levels of phytoplankton and confirms that these meso-scale features were cyclonic and anti-cyclonic eddies, respectively (Mizobata, 2002; Corredor et al., 2004). In my study, the satellite-derived tracks of leatherback movements, T GT data, and satellite-derived SSH data provide evidence that leatherbacks from the St. Croix nesting population occasionally forage along the edges of cyclonic and anticyclonic eddies (Figs 49B, 50B, 51B). Based on a previous report by Markman and Schwartz (2005), gelatinous prey abundance is predicted to be higher within the core of cyclonic eddies compared to the edges of either cyclonic or anticyclonic. Interestingly, the leatherbacks monitored during my study did not associate with the core of the cyclonic eddy observed in This may be a result of leatherbacks having a restricted amount of time to explore meso-scale eddies between nesting events (i.e. internesting interval 8-10 d.). Since leatherbacks have a limited amount of time between nesting events, as well as limited energy reserves during the nesting season, their interactions with eddies will most likely decrease as the distance between the nesting beach and the eddy increases. In contrast to the observations in this study, satellite-telemetry studies have revealed that 88

103 leatherbacks follow oceanographic features, such as eddies and currents, for prolonged periods of their post-nesting migrations (e.g. Luschi et al., 2003). The interaction of leatherbacks with eddies reported in this study should, therefore, only be interpreted as a reflection of their forging behavior associated with eddies during the nesting season. Foraging locations identified north of St. Croix in 2008 did not appear to be associated with SSH anomalies, and no SSH anomalies were associated with foraging events by turtles in As a result of these data, I did not find strong evidence that leatherbacks specifically target areas that are known to enhance primary productivity. It appears that SSH anomalies are not a good predictor of leatherback foraging grounds during the nesting season, and they play less important roles in leatherback foraging behavior during the nesting season than migration periods (e.g. Luschi et al. 2003). The impact of the SSH anomalies on the availability of prey for leatherbacks in this study was not directly investigated, but the observed leatherback feeding events in association with SHH anomalies during their internesting intervals suggests that these transient oceanographic features may promote an overall increase of prey in the oligotrophic waters of the Caribbean Sea. While eddies appear not be strong predictors of leatherback foraging behavior during the nesting season at St. Croix, a greater understanding of eddy disturbances on prey abundance is critical for understanding shifts in leatherback movements during the nesting seasons and migration periods. Bioenergetics The INT of a stomach temperature fluctuation (Fig. 9) was first described by Wilson et al. (1992) to be related to the mass of the food ingested (M F ), specific heat 89

104 capacity of the food (SHC f ), temperature of the food ingested (T f ), temperature to which the food must be heated (T a ), and how fast heat is passed from the animal to the food (m). This relationship can be written as the following: Eq. 1. INT = m * SHC f * (T a - T f ) * M F. (Wilson et al., 1992) The specific heat capacity of the food depends on the prey item ingested, while m in Eq. 1. is affected by several factors, including the animal s stomach volume, the insulation of the stomach and mixing of the food (Wilson et al., 1995). Wilson et al. (1992) also found that m is constant and describes the linear relationship between the INT and the energy invested by the animal into heating the food (i.e. SHC f * (T a - T f ) * M F ). Assuming one knows the relationship of the variables in Eq. 1 based on captive feeding studies (e.g. Kuhn and Costa, 2006; Wilson et al., 1992; Wilson et al., 1995) or laboratory ingestion simulations (e.g. Wilson et al., 1992), the INT values of stomach temperature fluctuations in free-swimming animals can be used to estimate the mass of the prey ingested (i.e. M F = INT/ m * SHC f * (T a - T f )). For my laboratory ingestion simulations, I assumed that jellyfish had a SHC value of 4.17 J ºC -1 s -1 because jellyfish are composed ~96% seawater (i.e jellyfish have a similar SHC to water) (Doyle et al., 2007). The relationship between INT and E for laboratory ingestion simulations with jellyfish gave an m value of 0.31± 0.02 ºC s -1 J -1 (INT = 0.31E , r 2 = 0.98, n = 6) (Fig. 57). Applying this m value to estimate the mass of prey ingested, based on measured INT from field T GT data, is problematic because the T f was unknown due to the dive behavior associated with foraging (i.e. ascending and descending during feeding). I estimated prey mass by setting T f equal to the T A at 1.0 min (T fa1.0 ) and at 3.5 min (T fa3.5 ) prior to the detection of prey ingestion 90

105 events. These time lags were selected for estimating T f because the mean time difference between the detection of prey ingestion and wiggle events was 3.6 min (Table #17). The mean difference in temperature between paired T fa1.0 and T fa3.5 (i.e. T fa1.0 - T fa3.5 ) ranged from -0.07º ± 1.2º to 0.89º ± 2.19ºC (maximum range: 2.15º to 6.15ºC). The T fa1.0 value was lower than its paired T fa3.5 value for 47% of the prey ingestions, and the opposite relationship between T fa1.0 and T fa3.5 occurred for 53% of the prey ingestion events. The mean prey mass estimates for T fa1.0 and T fa3.5 ranged from 60.2 to 1160 g and from -172 to 537 g (rounded to the nearest gram), respectively (Table #23). The differences between the mean and total prey mass estimates based on T fa1.0 and T fa3.5 were not statistically different (t = 1.75, df = 6, P = 0.16; t = 4.6, df = 6, P = 0.66). Prey mass estimates were all positive, with the exception of one ingestion event for turtle #AAR864. The negative estimate of prey mass (-2345 g) was a result of T fa3.5 being slightly warmer than the T I, or asymptotic temperature, of the ingestion (T fa3.5 = 28.35ºC; T I = 28.2ºC). In this case, if prey were equal to the ingestion s T fa3.5, the prey would not have led to a sudden drop in T GT (i.e. prey must be colder than the animal for detecting feeding), and therefore the time lag used for estimating T f was obviously incorrect. With the exception of the ingestion event with the negative prey mass estimate, comparisons of prey mass estimates between the paired T f values revealed differences up to 170 g. Applying the m value of laboratory ingestion simulations with jellyfish to the field data is also problematic because the composition or fluidity of the ingesta is unknown. The relationship between INT and E for my laboratory ingestion simulation with seawater gave a value for m of 0.02 ± ºC s -1 J -1 (INT = 0.02E , r 2 = 0.50, N = 91

106 16) (Fig. 57), and was significantly different than the laboratory simulations with only jellyfish (F test, P < 0.001). Therefore, if leatherbacks ingest prey with seawater, the fluidity of the ingesta increases and prey mass is underestimated by applying the m value of laboratory simulations with only jellyfish. A high m value is assumed to be responsible for prey mass estimates that were < 300 g because ingestion events characterized as prey were predicted to have consisted of at least 300 g of jellyfish, based on the statistical results of my laboratory ingestion simulations (Table 10, Fig. 42). Interestingly, #AAQ943 had an estimated prey mass of >4,000 g for its first prey ingestion event using T fa1.0 or T fa3.5 to estimate the T f. If it holds true that leatherbacks have a stomach volume of ~700 ml, this prey mass estimate is apparently overestimated by the estimates for T f. and/or the m value. To improve the use of stomach temperature telemetry for estimating prey mass ingested by leatherbacks, captive feeding studies with adult turtles need to be conducted with STP instruments. As mentioned before, these experiments have yet to be conducted because leatherbacks are difficult to care for in captivity (Jones, 2009) and the size of STP instruments precludes use with small animals. Additional studies on leatherback foraging behavior with stomach temperature telemetry should also consider simultaneous deployments of STPs with IMASENs and/or video cameras to determine the T f and/or the time lag between prey capture and the detection of prey by the STP. Wallace et al. (2005) estimated that leatherbacks from the St. Croix, USVI nesting population require 1.18 x 10 6 kj of energy during the nesting season for reproductive activities that include egg production, nesting activities and internesting activities. If this energy is not available in reserves, leatherbacks need to consume at least 8.2 x 10 6 g of 92

107 prey during the nesting season for successful reproduction, based on gelatinous prey having an energy content of ~0.18 kj g (Doyle et al., 2007) and prey assimilation of 80% (as assumed in Wallace et al., 2006). Despite the difficulties in estimating prey mass from the archival T GT data, the mean prey ingestion rates (range: 0.05 ± 0.04 to 0.35 ± 0.4 prey ingestion h -1 ) indicate that energy reserves procured prior to the breeding season are critical for successful reproduction by leatherbacks from the St. Croix, USVI nesting population. CONCLUSION The use of bio-logging technology in this study revealed that leatherbacks from the St. Croix, USVI nesting population conduct extensive movements and forage opportunistically toward the end of the nesting season. Management strategies devised to protect leatherbacks should consider expanding the leatherback critical habitat boundary adjacent to SPNWR, which was designated in 1998 by the National Marine Fisheries Service (Fig. 50), based on the important interesting habitat identified in this study (Fig. 14). The technique of stomach temperature telemetry proved to be a useful tool for investigating the foraging behavior of leatherback turtles during the nesting season. Prey ingestion rates indicate that energy reserves acquired prior to the breeding season are critical for successful reproduction by leatherbacks from the St. Croix, USVI nesting population. Future studies aimed at understanding how leatherbacks accrue sufficient energy reserves for successful reproduction should consider remotely collecting leatherback T GT during migration periods to detect feeding events. 93

108 Table 1. Summary of the instruments deployed on leatherback turtles nesting at St. Croix, USVI. Turtle no. Curved carapace length (cm) Instruments deployed Tagging date Re-nesting date Internesting period (d) Tracking period (d) T GT monitoring period (d) (A) PPQ Mk10-AL & STP3 18-May May e, j (B) XXZ Mk10-AL & STP3 19-May b n/a f (C) XXZ Mk10-AL & STP3 20-May b n/a f (D) AAR Mk10-AL & STP3 21-May May e, j (E) AAR Mk10-AL & STP3 22-May a n/a f (F) PPQ Mk10-AL & STP3 22-May Jun e, j (G) AAV Mk10-AL & STP3 23-May a, c 1.0 e, j (H) AAR Mk10-AL & STP3 24-May Jun e, j (I) XXZ Mk10-AL & STP3 24-May Jun e, j (J) AAR Mk10-AL & STP3 12-May May h, j (K) VI Mk10-AL & STP3 14-May b 24.3 g, k (L) XXZ Mk10-AL & STP3 14-May Jun i, j (M) AAC Mk10-AL & STP3 15-May b 0.1 g, k (N) AAG Mk10-AL & STP3 16-May d 32.3 h, k (O) XXZ Mk10-AL & STP3 17-May d 23.1 h, k (P) KL Mk10-AL & STP3 18-May Jun a 10.8 h, k (Q) AAC Mk10-AL & STP3 18-May May g, j (R) Mk10-AL & STP3 19-May Jun g, j (S) AAQ Mk10-AL & STP3 20-May May g, j a Tracking period was ended by Mk10-AL shedding, while the turtle was at sea. b Tracking period was assumed to have ended because of Mk10- AL shedding or malfunction. c Mk10-AL was recovered from the ocean s floor by a scallop dredger. d Tracking period was assumed to have ended because of Mk10-AL battery life failure. e STP3 retention time was assumed to be approximately the same as the T GT monitoring period. f STP3 retention time was unknown because turtle s Mk10-AL was not retrieved and its Mk10-AL was not programmed to relay T GT data via satellite. g STP3 retention time was unknown because the turtle returned to sea with its STP3 and without an Mk10-AL to received T GT data. h STP3 retention time was unknown because the battery life of the STP3 may have occurred prior to pill excretion. i STP3 retention time was unknown because the turtle s Mk10-AL failed at sea while receiving T GT data and the turtle did not return to nest again at SPNWR. j T GT obtained as archival times-series data. k T GT data obtained only from satellite transmissions. 94

109 Table 2. Summary of satellite data relayed from leatherbacks from the St. Croix, USVI nesting population during 2007 and Turtle no. Dive depth Dive Duration No. of No. of records No. of dives No. of records No. of dives time-at-temp. records No. of Argos locations B Traveling speed (km/d) Max. internesting displacement from SPNWR (km) (A) PPQ ± (B) XXZ ± (C) XXZ ± (D) AAR ± (E) AAR ± (F) PPQ ± (G) AAV ± (H) AAR ± (I) XXZ ± (J) AAR ± (K) VI ± a 1660 (L) XXZ ± (M) AAC (N) AAG ± (O) XXZ ± (P) KL ± (Q) AAC ± (R) ± (S) AAQ ± Min. distance traveled (km) Note. Data are presented as the mean ± 1SD. a Incomplete internesting tracking period for turtle. Argos satellite locations (Class 3, 2, 1, 0, A) were filtered using the following parameters: speed >5 km/h, topography > 0, and turning angle < 25º. 95

110 Table 3. Summary of internesting diving behavior by leatherback turtles (n = 10) from the St. Croix, USVI nesting population. Turtle no. Total no. of dives V-shaped Dives (%) U-shaped Dives (%) Hourly no. of dives (dives) Dive depth (m) Dive duration (min) Surface time (min) Mean Max. Mean Max. Mean Max. (A) PPQ ± ± ± ± (D) AAR ± ± ± ± (F) PPQ ± ± ± ± (H) AAR ± ± ± ± (I) XXZ ± ± ± ± (J) AAR ± ± ± ± (L) XXZ ± ± ± ± (Q) AAC ± ± ± ± (R) ± ± ± ± (S) AAQ ± ± ± ± All turtles ± ± ± ± ± ± Note. Data are presented as the mean ± 1SD. Dive data analysis was performed on archival time-series data, which was recorded by platform terminal transmitters (PTTs) (model Mk10-AL, Wildlife Computers). 96

111 Table 4. Summary of internesting submergence time and dives with wiggle events by leatherback turtles (n = 10) from the St. Croix, USVI nesting population. A higher proportion of internesting daytime dives contained and wiggle event in comparison to those performed during the nighttime (t = 2.6, df = 9, P < 0.03). Dive data analysis was performed on archival time-series data, which was recorded by platform transmitter terminals (PTTs) (model Mk10-AL, Wildlife Computers). Turtle no. Time spent submerged (%) Dives with a wiggle event (%) Total Daytime Nighttime (A) PPQ (D AAR (F) PPQ (H) AAR (I) XXZ (J) AAR (L) XXZ (Q) AAC (R) (S) AAQ All turtles 73.9 ± ± ± ± 8.6 Note. Data are presented as mean ± 1SD.Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. 97

112 Table 5. Diel comparison of hourly number of dives, dive duration, maximum dive depth, and post-dive surface time by leatherback turtles (n = 10) from the St. Croix, USVI nesting population. Daytime dive were significantly less frequent and deeper than nighttime dives (t = 3.93, df = 9, P < 0.01; t = 5.23, df = 9, P < 0.001). Additionally, daytime time dives were significantly longer than those performed during the nighttime (t = 3.45, df = 9, P < 0.001), and post-dive surface times (min dive -1 ) during the daytime were longer than those performed during the nighttime (t = 5.15, df = 9, P < 0.001). Turtle no. Hourly no. of dives (dives hour -1 ) Maximum dive depth (m) Dive duration (min) Post-dive surface time (min) Daytime Nighttime Daytime Nighttime Daytime Nighttime Daytime Nighttime (A) PPQ ± ± ± 4 93 ± ± ± ± ± 0.1 (D) AAR ± ± ± ± ± ± ± ± 0.1 (F) PPQ ± ± ± ± ± ± ± ± 0.2 (H) AAR ± ± ± ± ± ± ± ± 0.1 (I) XXZ ± ± ± ± ± ± ± ± 0.1 (J) AAR ± ± ± ± ± ± ± ± 0.1 (L) XXZ ± ± ± ± ± ± ± ± 0.1 (Q) AAC ± ± ± ± ± ± ± ± 0.1 (R) ± ± ± ± ± ± ± ± 0.1 (S) AAQ ± ± ± ± ± ± ± ± 0.1 All turtles 3.3 ± ± ± ± ± ± ± ± 0.7 Note. Data are presented as the mean ± 1SE. Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. Dive data analysis was performed on archival time-series data, which was recorded by platform transmitter terminals (PTTs) (model Mk10-AL, Wildlife Computers). 98

113 Table 6. Summary of the internesting gastrointestinal tract temperature (T GT ) and ambient temperature (T A ) for leatherback turtles (n = 8) from the St. Croix, USVI nesting population. The turtles T GT were significantly higher than their T A (t = 7.77, df = 7, P < 0.001). Turtle ID no. Duration of T GT data monitoring (d) T GT (ºC) T A (ºC) (A) PPQ ± ± 2.0 (D) AAR ± ± 3.3 (G) AAV ± ± 1.2 (J) AAR ± ± 1.7 (L) XXZ ± ± 1.3 (R) AAC ± ± 2.2 (Q) ± ± 2.0 (S) AAQ ± ± 1.7 All turtles 8.7 ± ± ± 0.7 Note. Data are presented as the mean ± SD. Data analysis was performed on archival timeseries data collected after the turtles were at sea for 6 h. 99

114 Table 7. Summary of total number of dives, dive frequency, and maximum depth of dives by leatherback turtles (n = 8) from the St. Croix, USVI nesting population, while their gastrointestinal temperatures were being monitored using stomach temperature telemetry. Turtle ID no. No. of dives during T GT data Dive frequency(dives h -1 ) Maximum dive depth (m) Total Daytime Nighttime Total Daytime Nighttime (A) PPQ ± ± ± ± ± ± 51 (D) AAR ± ± ± ± ± ± 38 (G) AAV ± ± ± ± ± ± 33 (J) AAR ± ± ± ± ± ± 38 (L) XXZ ± ± ± ± ± ± 28 (R) AAC ± ± ± ± ± ± 64 (Q) ± ± ± ± ± ± 52 (S) AAQ ± ± ± ± ± ± 45 All turtles ± ± ± ± ± ± 33 Note. Data are presented as the mean ±SD. Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. Data analysis was performed on archival time-series data collected after the turtles were at sea for 6 h. 100

115 Table 8. Summary of dive duration, post-dive surface time and dive shape by leatherback turtles (n = 8) from the St. Croix, USVI nesting population, while their gastrointestinal temperatures were being monitored using stomach temperature telemetry. Turtle ID no. Dive duration (min) Post-dive surface time (min) Total Daytime Nighttime Total Daytime Nighttime V-shaped dives (%) (A) PPQ ± ± ± ± ± ± (D) AAR ± ± ± ± ± ± (G) AAV ± ± ± ± ± ± (J) AAR ± ± ± ± ± ± (L) XXZ ± ± ± ± ± ± (R) AAC ± ± ± ± ± ± (Q) ± ± ± ± ± ± (S) AAQ ± ± ± ± ± ± U-Shaped Dives (%) All turtles 12.4 ± ± ± ± ± ± ± ± 5.9 Note. Data are presented as the mean ±SD. Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. Data analysis was performed on archival time-series data collected after the turtles were at sea for 6 h. 101

116 Table 9. Summary of ingestion events documented in gastrointestinal temperature records of leatherback turtles (n = 7) from the St. Croix, USVI nesting population, using stomach temperature telemetry. Turtle ID no. Total no. of ingestion events Rate of ingestion events (ingestion h -1 ) Duration of ingestion events (min) (A) PPQ ± 35.6 (D) AAR ± 14.2 (G) AAV (J) AAR ± 39.8 (L) XXZ ± 36.7 (R) AAC ± 27.8 (Q) ± 22.8 (S) AAQ ± 27.0 All turtles ± ± 17.8 Note. Data are presented as the mean ±SD. Data for ingestion events came from analysis of archival time-series gastrointestinal temperature data collected after the turtles were at sea for 6 h. 102

117 Table 10. Summary of laboratory ingestion simulations conducted using artificial seawater and moon jellyfish (Aurelia aurita). The group comprised of 300 g of jellyfish with 200 g of seawater was the smallest mass of jellyfish used in the laboratory simulations that had significantly higher integral index values than group comprised of 650 g of seawater. Ingestion Simulation Groups 500 g of seawater 650 g of seawater 100 g jellyfish with 400 g seawater 200 g jellyfish with 300 g seawater 300 g jellyfish with 200 g seawater No. of trials Integral index value 500 g jellyfish 264 ± ± ± ± ± ± 28.3 Note. Data are presented as the mean ±1SD. The difference in mean integral index values of ingestion simulation groups were tested statistically using a Kruskal-Wallis followed by a post-hoc Mann-Whitney U-Test. The significance level (P < 0.017) for multiple comparisons was adjusted for each test using a Bonferroni method (α/n). Pairwise comparisons were made between the seawater group with the largest mean integral index value (i.e. 650 g of seawater) and groups that contained 200 g of jellyfish. 103

118 Table 11. Summary of integral index values and duration ingestions events characterized as prey in seven leatherback turtles from the St. Croix, USVI nesting population. All prey ingestions documented in this study were required to have an integral index value > 482 s -1 based on laboratory ingestion simulations. There was no significant difference between the mean prey ingestion event rates (ingestions h -1 ) documented during the daytime and the nighttime (t = 1.61, df = 6, P = 0.16). Turtle no. No. of prey ingestions Prey Ingestion Rate (prey ingestions/h -1 ) Prey ingestion integral index value Total Day Night Total Day Night Total Day Night Duration of prey ingestions (min) (A) PPQ ± ± ± ± 22.5 (D) AAR ± ± ± 10.2 (J) AAR ± ± ± 19.2 (L) XXZ ± ± ± ± 32.8 (Q) AAC ± ± ± ± 25.8 (R) ± ± ± 21.3 (S) AAQ Other ± ± ± ± 25.9 All turtles ± ± ± ± ± ± ± 12.7 Note. Data are presented as the mean ±SD. Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. 104

119 Table 12. Summary of the dive depth associated with ingestion events characterized as prey ingestions in leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval. The prey ingestion dives that occurred during the daytime were typically deeper than those that occurred during the nighttime, but the difference between the mean maximum dive depth of daytime prey ingestion dives was not significantly different in comparison to nighttime prey ingestion dive depth (t = 2.49, df = 3, P = 0.08). There was also no significant difference between the mean starting depths of daytime and nighttime prey ingestion events (t = 3.02, df = 3, P = 0.06). Turtle no. Maximum depth of prey ingestion dives (m) Overall Daytime Nighttime Depth at the start of prey ingestion events (m) Overall Daytime Nighttime Mean Max Mean Max Mean Max Prey ingestion start depth max. depth of prey ingestion dive (m) (A) PPQ ± ± ± ± ± ± ± 59 (D) AAR ± ± ± ± ± 42 (J) AAR ± ± ± ± ± 71 (L) XXZ ± ± ± ± ± ± ± 77 (Q) AAC ± ± ± ± ± ± ± 26 (R) ± ± ± ± ± 35 (S) AAQ ± ± ± ± ± ± ± 59 All turtles 187 ± ± ± ± ± ± ± 13.4 Note. Data are presented as the mean ±SD. Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. 105

120 Table 13. Summary of maximum depth, duration and the post-dive surface time of dive associated with prey ingestion events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval. Mean daytime prey ingestion dive durations were not statistically different from mean nighttime prey ingestion dive durations (t = 0.66, df = 3, P = 0.55). Post-dive surface times of prey ingestion dives were typically longer during the daytime than the nighttime, but the difference was not statistically significant (t = 1.9, df = 3, P = 0.15). Turtle no. Dive duration of prey ingestion dives (min) Prey ingestion dive post-dive surface time (min) Total Daytime Nighttime Total Daytime Nighttime (A) PPQ ± ± ± ± ± ± 0.2 (D) AAR ± ± ± ± (J) AAR ± ± ± ± 6.5 (L) XXZ ± ± ± ± ± ± 0.4 (Q) AAC ± ± ± ± ± ± 0.3 (R) ± ± ± ± (S) AAQ ± ± ± ± ± ± 1.3 All turtles 19.1 ± ± ± ± ± ± 0.4 Note. Data are presented as the mean ±SD. Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. 106

121 Table 14. Summary of ambient temperature (T A ) at maximum depth and post-dive surface times of dives associated with prey ingestion events by leatherbacks turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval. The T A at the maximum depth of prey ingestion dives was similar between daytime and nighttime prey ingestion dives (df = 3, P = 0.17); however, T A at the start of prey ingestion events were different between daytime and nighttime prey ingestion events (df = 3, P = 0.03). No difference was detected between the ΔT D of daytime and nighttime prey ingestion events (df = 3, P = 0.34). Turtle no. T A at maximum depth of prey ingestion dives (ºC) T A at start of prey ingestion events (ºC) ΔT D of prey ingestion events (ºC) Overall Daytime Nighttime Overall Daytime Nighttime Overall Daytime Nighttime (A) PPQ ± ± ± ± ± ± ± ± ± 0.1 (D) AAR ± ± ± ± ± ± (J) AAR ± ± ± ± ± ± 0.2 (L) XXZ ± ± ± ± ± ± ± ± ± 0.2 (Q) AAC ± ± ± ± ± ± ± ± ± 0.2 (R) ± ± ± ± ± ± (S) AAQ ± ± ± ± ± ± ± ± ± 1.2 All turtles 21.2 ± ± ± ± ± ± ± ± ± 0.2 Note. Data are presented as the mean ±SD. Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. 107

122 Table 15. Summary of dive depth and duration for V-shaped and U-shaped dives associated with prey ingestion events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval. Turtle No. V-shaped dives with prey ingestion events Note. Data are presented as the mean ±SD. U-shaped dives with prey ingestion events No. Total (%) Depth (m) Duration (min) No. Total (%) Depth (m) Duration (min) (A) PPQ ± ± (D) AAR ± ± (J) AAR ± ± ± ± 0.6 (L) XXZ ± ± (Q) AAC ± ± (R) ± ± (S) AAQ ± ± All Turtles ± ± ± ± ± ±

123 Table 16. Summary V-shaped and U-shaped dive phases associated with prey ingestion events by leatherback turtles (n = 7) from the St. Croix, USVI, nesting population during the internesting interval. The difference in mean proportions of prey ingestion dive phases for V-shaped dives were tested statistically using a one-way analysis of variance (ANOVA) test (significance level P = 0.05), followed by a post-hoc Tukey Test. Different letters (a, b) indicate (P < 0.008) differences among groups. A Kruskal-Wallis test detected no significant difference among the mean proportions of prey ingestion dive phases for U-shaped dives (Kruskal-Wallis, H 2, 28 = 6.23, P = 0.10). Turtle No. Dive phase for prey ingestion during V-shaped dives (%) Dive phase for prey ingestion during U-shaped dives (%) Descent a Bottom ab Ascent b Surface a Decent Bottom Ascent Surface (A) PPQ (D) AAR (J) AAR (L) XXZ (Q) AAC (R) (S) AAQ All Turtles 16.5 ± ± ± ± ± Note. Data are presented as the mean ±SD. Data analysis was performed on archival time-series data, which was recorded by platform transmitter terminals (PTTs) (model Mk10-AL, Wildlife Computers). A significance level of P < was set for each pair-wise comparison in the Tukey Test based on a Bonferroni correction method (α/n) to reduce the change of Type I errors. 109

124 Table 17. Summary of dives associated with prey ingestion events containing wiggle events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval. The majority of wiggle events associated with prey ingestion dives were documented during V-shaped dives (84.3%), but the proportion of the total number of V-shaped prey ingestion dives that contained a wiggle event was lower than the proportion of U-shaped prey ingestion dives that contained a wiggle event (t = 7.8, df = 3, P = 0.004). Turtle no. No. of prey ingestion dives containing a wiggle event Frequency of prey ingestion dives containing a wiggle event (%) Total V-shaped U-shaped Total V-shaped U-shaped (A) PPQ ± 11.7 (D) AAR ± 4.0 (J) AAR ± 4.7 (L) XXZ ± 4.5 (Q) AAC ± 8.1 (R) ± 14.5 (S) AAQ ± 1.8 All turtles ± ± ± ± 3.5 Prey detection start time wiggle event start time (min) Note. Data are presented as the mean ±SD. The frequency of prey ingestion dives containing a wiggle event (%) was calculated from the total number of V-shaped and U-shaped prey ingestion dives by individual turtles during their T GT monitoring period. 110

125 Table 18. Summary of bathymetry and sea surface chlorophyll-a concentrations for the locations of prey ingestion events by seven leatherback turtles from the St. Croix, USVI nesting population during the internesting interval. Turtle no. Sea depth for location of prey ingestion events (m) Distance btw. St. Croix and identified prey ingestion events (km) Total Minimum Maximum [Chla] mg/m 3 (A) PPQ ± ± ± (D) AAR ± ± ± (J) AAR ± ± 21.9 Onshore* ± 0.32 (L) XXZ ± ± ± 0.31 (Q) AAC ± ± 14.0 Onshore* ± (R) ± ± ± 0.28 (S) AAQ ± ± ± All Turtles 2551 ± ± ± Note. Data are presented as the mean ±SD. Onshore locations (*) of foraging events were caused by an apparent error of foraging events occurring between consecutive Argos locations that were spatially separated by St. Croix. 111

126 Table 19. Summary of unidentified/non-characterized ingestion events by seven leatherback turtles from the St. Croix, USVI nesting population during the internesting interval. Daytime and nighttime unidentified ingestion rates were similar (t = 1.28, df = 6, P = 0.25). Turtle no. No. of unidentified ingestion events Rate of unidentified ingestion events (ingestions/h -1 ) Total Day Night Total Day Night (A) PPQ (D) AAR (J) AAR (L) XXZ (Q) AAC (R) (S) AAQ All turtles ± ± ±0.03 Note. Data are presented as the mean ±SD. Daytime was represented by 05:00-18:59 h and nighttime was represented by 19:00-04:59 h. 112

127 Table 20. Summary of diving behavior and thermal conditions associated with unidentified/non-characterized ingestion events by seven leatherback turtles from the St. Croix, USVI nesting population. Turtle Maximum Depth at detection of unidentified Post-dive Freq. of dives with T A at max. T A at detection of unidentified Freq. of V- depth of dives (m) ingestion events (m) Duration of dives (min) surface time (min) wiggle events (%) depth of dives (ºC) ingestion events (ºC) shaped dives (%) (A) PPQ ± ± ± ± ± ± (D) AAR ± ± ± ± ± ± (J) AAR ± ± ± ± ± ± (L) XXZ ± ± ± ± ± ± (R) AAC ± ± ± ± ± ± (Q) ± ± ± ± ± ± (S) AAQ ± ± ± ± ± ± Freq. of U-shaped dives (%) All Turtles 171 ± ± ± ± ± ± ± ± ± 12.8 Note. Data are presented as the mean ±SD. 113

128 Table 21. Summary of the dive phase associated with unidentified ingestion events by leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval. Turtle No. Dive phase for unidentified ingestion events during V-shaped dives (%) Dive phase for unidentified ingestion events during U-shaped dives (%) Descent a Bottom b Ascent a Surface a Decent Bottom Ascent Surface (A) PPQ (D) AAR (J) AAR (L) XXZ (Q) AAC (R) (S) AAQ All Turtles 9.2 ± ± ± ± ± ± ± 37.8 Note. Data are presented as the mean ±SD. 114

129 Table 22. Summary of wiggle events associated with unidentified/non-classified ingestion events for leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval. Unidentified ingestion events occurred more frequently during V-shaped dives than U-shaped dives (t = 9.7, df = 7, P < 0.001), but there was no significant difference between the proportion of V-shaped and U-shaped unidentified ingestion dives that contained a wiggle event (t = 3.81, df = 2, P = 0.06). Turtle no. No. of unidentified ingestion dives containing a wiggle event Frequency of unidentified ingestion dives containing a wiggle event (%) Total V-shaped U-shaped V-shaped U-shaped (A) PPQ na (D) AAR ± 2.8 (J) AAR ± 1.1 (L) XXZ na (Q) AAC ± 1.9 (R) ± 5.7 (S) AAQ All turtles ± ± ± 0.8 Note. Data are presented as the mean ±SD. Detection of unidentified ingestion event wiggle event start time (min) 115

130 Table 23. Prey mass estimates for ingestion events characterized as prey for leatherback turtles (n = 7) from the St. Croix, USVI nesting population during the internesting interval. Turtle ID No. Prey mass estimated with T fa1.0 Prey mass estimated with T fa3.5 Mean Min. Max. Total Mean Min. Max. Total (A) PPQ ± ± (D) AAR ± ± (J) AAR ± ± (L) XXZ ± ± (Q) AAC ± ± (R) ± ± (S) AAQ ± ± Note. Date are presented as mean ± S.D. Prey mass was estimated using the following equation: M f = INT/(m *SHC f * (T A -T F )). The SHC f was equal to 4.17 J s -1 ºC -1 ; M f was the estimated prey mass; INT values for ingestion events were measured using the archival gastrointestinal temperature data; m was calculated as the linear relationship between INT and E (energy) for laboratory ingestion simulations with jellyfish and was equal to 0.31 ± ºC s -1 J -1 ; T A was equal to the asymptote of the prey ingestion; and T F was either the T A at 1.0 min (T fa1.0 ) or 3.5 min (T fa3.5 ) prior to the detection of the prey. 116

131 A) B) Figure 1. Photos of adult leatherback sea turtles (Dermochelys coriacea) at sea (A) and onboard a research vessel. The photos were taken by Kara Dodge, Ph.D. candidate, Large Pelagics Research Center, University of New Hampshire, US, while conducting research offshore of Massachusetts, US. 117

132 Figure 2. Map displays the location of Sandy Point National Wildlife Refuge (SPNWR), St. Croix, USVI. The dotted line delineates the 200 m isobath. 118

133 A) B) Figure 3. Photos of leatherback sea turtles nesting at Sandy Point National Wildlife Refuge (SPNWR), St. Croix USVI after platform transmitter terminals (PTTs) were attached to their carapace during this study. The PTTs (model, Mk10-AL, Wildlife Computers) were attached directly to the carapace between the medial and 1 st lateral ridge of turtles (n = 9) in 2007 (A) and on the dorsal medial ridge of turtles (n = 10) in 2008 (B). 119

134 Orthopedic-mini-anchors (OMAs) positioned between the medial and 1 st lateral ridge of the carapace A) Transmitter s arm-plates secured to the OMAs, using washers and a hairpin B) Figure 4. Photos of the platform transmitter terminal (PTT) attachment site for turtles during the 2007 field season. Four OMAs were inserted into pre-drilled holes between the medial and 1 st lateral ridge near the frontal end of the carapace (A). The shafts of the OMAs (~15mm), protruding from the turtle s carapace, were used to secure the transmitter to the turtle s carapace (B). The PTTs were model Mk10-AL, Wildlife Computers, Inc. 120

135 { } Stainless steel wires passing through the transmitter s arm-plate (bottom of transmitter in view) Tygon-coated stainless steel wires inserted through two pre-drill holes in the medial dorsal ridge of the carapace A) B) Figure 5. Photos of the attachment site for platform transmitter terminals (PTTs) during the 2008 field season. PTTs were attached to the dorsal medial ridge of turtles in Tygon-coated stainless steel wires were inserted through two pre-drilled holes in the medial ridge and through each arm-plate of the transmitter (A). The ends of the wires were twisted together to secure the transmitter to the turtle s carapace. PTTs rested on non-compressible putty (Equinox ) that was formed to the medial ridge to stabilize the tag (B). The PTTs were model Mk10-AL, Wildlife Computers, Inc. 121

136 Nylon webbing Upper jaw Lower jaw Flexible PVC tube threaded with a plastic, rigid PVC trocar Figure 6. A photo of the stomach temperature pill (STP) insertion method used in this study. The STP was inserted into the turtle s esophagus after oviposting was completed. Nylon webbing was used to hold open the turtle s beak-mouth. Several STP insertion procedures included a piece of hard foam (shown above in blue) that was placed between the turtle s upper and lower jaw as precaution for the turtle closing its mouth onto the PVC tube during the insertion procedure. One piece of nylon webbing used for holding down the turtle s lower jaw is not show in the photo above. The STPs were model STP3,Wildlife Computers, Inc. 122

137 A) B) C) Figure 7. Photos of stomach temperature pills (STPs) used in this study. Each STP detected temperature, using a series of 4 thermistors, and transmitted temperature data acoustically to a platform transmitter terminal (PTT) with archival capability. The pills had an internal/external titanium ring at their center that minimized the time of heat transfer between the pill s external medium and its theremistors. The ends of the original STPs (A) were enlarged using layers of gelatin coated with ethocel for deployments in 2007 (B), and with only layers of ethocel for deployments in 2008 (C). Ethocel is a dissolvable cellulose matrix commonly used in for making time-release medicine capsules. This method increased the cylinder diameter of the STP3s from 24 mm to ~40 mm. The STPs were model STP3, Wildlife Computers, Inc. 123

138 33 Turtle enters ocean 0 o Depth TGT /20/ : Depth (m) T GT ( C) 31 5/20/ :00 5/20/ :00 5/20/ : /21/ :00 Local Time Figure 8. Dive depth profile and gastrointestinal tract temperature (TGT) recorded for turtle #AAQ943 during her first 24 h at sea. Rapid fluctuations in the turtle s TGT occurred while TGT lowered/cooled steadily from 29.7ºC to temperatures below 28.5ºC. The fluctuations are indicative of ingestion events and were most likely a result of the turtle drinking seawater to cool/decrease her body temperature (i.e. thermoregulate) or to move the STP down the esophagus. 124

139 A) Stomach Temperature A) Integral TRIM C) B) Time Time Figure 9. Stomach temperature fluctuations previously observed in cormorants (A) and schematic diagrams of temperature fluctuations following ingestion events (B-C). Stomach temperature fluctuations recorded in cormorants (A) after receiving 120 ml of seawater at 4 C (dotted line) and 86 g of squid at 4 C (solid line). The area below the asymptote of a stomach temperature fluctuation is also known as the integral of the fluctuation. Integrals are measured from the start of the temperature drop to the when the temperature returns to the asymptotic value (B) or from the minimum temperature of the fluctuation to when the temperature returns to the asymptotic values (C). The magnitude of the temperature drop affects the size and shape of the integral value. Graph A is from Catry et al. (2004) and graphs B & C were modified from Wilson et al. (1995). 125

140 Figure 10. Photo of the platform transmitter terminal (PTT) attachment site for turtle #AAR264 immediately following the PTT removal process during the 2007 field season. Turtle #AAR863 remained at sea with its PTT for a period of 9.1 d. PTTs were retrieved from 5 of 6 turtles that returned to nest again at SPNWR in 2007 and no turtles displayed signs of chaffing or necrosis at the attachment site, including the area immediately surrounding the OMAs. 126

141 A) B) C) Figure 11. Photos of the platform transmitter terminal (PTT) attachment site for turtle #XXZ456 (A-B) and turtle #AAQ943 (C) after being at sea for 29.8 d and 8.0 d, respectively. PTTs were retrieved from 5 of 6 turtles that returned to nest at SPNWR in The PTTs remained securely attached to a turtle s medial dorsal ridge for 8-10 d, but shifted to the side and were not tightly secured to the turtles medial dorsal ridge after > 15 d at sea. At the site of the attachment, specifically the medial ridge drill holes, two turtles showed minimal signs of necrosis (B) three turtles showed minimal to no signs of necrosis (C). Photos A and B were provided by the WIMARCS, Inc. 127

142 A) B) Figure 12. Maps of tracks of turtle #AAR863 (A), #PPQ234 (A), #AAR591 (B), and #PPQ244 (B), during 8-10 d internesting intervals in Tracks for turtles were derived from linear interpolation of their Argos satellite locations (3, 2, 1, 0, and A). Argos locations were filtered with the following parameters: speed > 5km/hr; topo >1m; angle < 25º. Arrows indicate traveling directions for turtles and the dotted line delineates the 1000 m isobath 128

143 A) B) Figure 13. Maps of tracks for turtle #AAR264 (A), #XXZ142 (A), #AAQ943 (B) and #AAC261 (B) during 8-10 day internesting intervals in 2007 or Tracks of turtles were derived from linear interpolation of their Argos satellite locations (3, 2, 1, 0, and A), which were filtered using STAT (satellite tracking analysis tool) with the following parameters: speed > 5km/hr; topo >1m; angle < 25º. Arrows indicate traveling directions for turtles and the dotted line delineates the 1000 m isobaths. 129

144 Figure 14. Map of the kernel-estimated habitat utilization distributions (KHUD) of internesting locations (n = 64) recorded for leatherback turtles from the St. Croix, USVI nesting population during one of their internesting intervals (8-10 d) in 2007 or The dotted line delineates the 1000 m isobath. 130

145 A) B) Figure 15. Maps of tracks for turtle #XXZ465 (A) and #KL56 (B), recorded during 28.9 d and 14.7 d tracking period, respectively, between consecutive nesting events at SPNWR in Tracks of turtles were derived from linear interpolation of their Argos satellite locations (3, 2, 1, 0, and A), which were filtered using STAT (satellite tracking analysis tool) with the following parameters: speed > 5km/hr; topo >1m; angle < 25º. Arrows indicate traveling directions for turtles and the dotted line delineates the 1000 m isobath. 131

146 Figure 16. Map of track recorded for turtle # during its 16 d tracking period between consecutive nests at SPNWR in The turtle s track was derived from linear interpolation of its Argos satellite locations (3, 2, 1, 0, and A), which were filtered using STAT (satellite tracking analysis tool) with the following parameters: speed > 5km/hr; topo >1m; angle < 25º. Arrows indicate traveling directions for turtles and the dotted line delineates the 1000 m isobaths. 132

147 Figure 17. Map shows the location of the satellite transmission in received for turtle #AAC271 during its 0.1 d tracking period. The dotted line delineates the 200 m. 133

148 A) B) Figure 18. Maps of post-nesting tracks recorded for turtle #AAR530 (A), #XXZ126 (A) and #VI1400 (B). Tracks for turtles were derived from linear interpolation of their Argos satellite locations (3, 2, 1, 0, and A), which were filtered using STAT (satellite tracking analysis tool) with the following parameters: speed > 5km/hr; topo >1m; angle < 25º. Arrows indicate traveling directions for turtles and the dotted line delineates the 1000 m isobath. 134

149 Figure 19. Map of post-nesting migrations tracks for turtle #0XXZ481, #AAV935, #AAG914, #XXZ123, recorded during 2007 and Tracks of turtles were derived from linear interpolation of their Argos satellite locations (3, 2, 1, 0, and A), which were filtered using STAT (satellite tracking analysis tool) with the following parameters: speed > 5km/hr; topo >1m; angle < 25º. Arrows indicate traveling directions for turtles. 135

150 A) B) Figure 20. Maps of post-nesting tracks for turtle #XXZ481(A) and #AAV935 (B), recorded during Tracks for turtles were derived from linear interpolation of their Argos satellite locations (3, 2, 1, 0, and A), which were filtered using STAT (satellite tracking analysis tool) with the following parameters: speed > 5km/hr; topo >1m; angle < 25º. Arrows indicate traveling directions for turtles. Arrows indicate traveling directions for turtles and the dotted line delineates the 1000 m isobaths. 136

151 A) B) Figure 21. Maps of post-nesting tracks for turtle #AAG914 (A) and #XXZ123 (B), recorded during Tracks of turtles were derived from linear interpolation of their Argos satellite locations (3, 2, 1, 0, and A), which were filtered using STAT (satellite tracking analysis tool) with the following parameters: speed > 5km/hr; topo >1m; angle < 25º. Arrows indicate traveling directions for turtles 137

152 Depth Bins A) #PPQ234 > Depth Bins E) #AAR Frequency Frequency > Frequency G) #AAV Frequency F) #PPQ D) #AAR863 C) #XXZ126 B) #XXZ Frequency Frequency Frequency H) #AAR Frequency Depth Bins I) XXZ142 > Figure 22. Satellite-derived relative frequency distributions of maximum dive depth by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2007 when the turtles were within the Caribbean Sea. Turtles generally conducted dives to depths ranging from m. The depth data was remotely relayed via the Argos satellite system Frequency

153 Depth Bins A) #AAR264 > Frequency Depth Bins E) #XXZ123 > Depth Bins Frequency Frequency Frequency Frequency H) # G) #AAC261 F) #KL56 Frequency I) AAQ943 > D) #AAG914 C) #XXZ465 B) #VI Frequency Frequency Figure 23. Satellite-derived relative frequency distributions of maximum dive depth by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2008 when the turtles were within the Caribbean Sea. Turtles generally conducted dives to depths ranging from , but dives to depths shallow depths ranging from 3 to 10 meters were also common for several turtles Frequency

154 Dive Duration (min.) A) #PPQ234 > B) #XXZ Frequency 0.2 Dive Duration (min.) Frequency D) #AAR Frequency Frequency H) #AAR591 G) #AAV Frequency Frequency F) #PPQ244 E) #AAR530 > C) #XXZ Frequency Frequency Dive Duration (min.) I) #XXZ142 > Figure 24. Relative frequency histograms for dive durations by turtles tagged in Turtle generally made dives lasting min. Binned data was remotely relayed via the Argos satellite system by the turtle s platform transmitter terminals Frequency 140

155 Dive Duration (min) A) #AAR264 > B) #VI Dive Duration (min.) Frequency Dive Duration (min.) Frequency I) #AAQ943 > Frequency D) #AAG G) #AAC Frequency F) #KL56 E) #XXZ123 > C) #XXZ Frequency Frequency H) # Frequency Frequency Figure 25. Satellite-derived relative frequency distributions of dive durations by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2008 when the turtles were within the Caribbean Sea. Turtle generally made dives lasting min. Binned data was remotely relayed via the Argos satellite system by the turtle s platform transmitter terminals. 0.4 Frequency 141

156 o Temperature ( C) Time (%) o D) #AAR Time (%) Time (%) Time (%) F) #PPQ244 E) #AAR530 Temperature ( C) C) #XXZ126 B) #XXZ481 A) #PPQ234 > H) #AAR591 G) #AAV935 > Time (%) Time (%) Time (%) Time (%) o Temperature ( C) I) #XXZ142 > Figure 26. Satellite-derived time-at-temperature distributions by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2007 when the turtles were within the Caribbean Sea. Turtles generally spent most of their time at temperatures ranging from 26º to 32ºC Time (%) 142

157 Temperature (oc) A) #AAR264 > B) #VI Time (%) Temperature (oc) E) #XXZ Time (%) F) #KL56 25 D) #AAG Time (%) G) #AAC Time (%) H) # > Time (%) Temperature (oc) 25 C) #XXZ465 I) #AAQ943 > Time (%) Time (%) Time (%) Figure 27. Satellite-derived time-at-temperature distributions by leatherback turtles from the St. Croix, USVI nesting population, recorded during 2008 when the turtles were within the Caribbean Sea. Turtles generally spent most of their time at temperatures ranging from 26º to 32ºC. 50 Time (%) 143

158 6 12 B) Post-Dive Surface Time (min.) Number of Dives A) Local Time D) Dive Duration (min) C) Dive Depth (m) 1800 Local Time Local Time Local Time Figure 28. Diel dive data for leatherback turtles (n = 10) from the St. Croix, USVI nesting population, recorded during one of their internesting intervals. Hourly overall means are shown as closed circles and error bars represent 1 standard error of the mean (SE). Data for daytime (05:00-18:59 h) is highlighted in yellow and nighttime (19:0004:59 h) is highlighted in gray. Nighttime dives were more frequent (dive hour-1) (4.2 ± 1.1) and shallower (79.3 ± 33.0 m) than dives performed during the daytime (3.3 ± 0.9, t = 3.93, df = 9, P < 0.01; ± m, t = 5.23, df = 9, P < 0.001) (A and C). Daytime time dives were longer (13.1 ± 3.3 min) than those performed during the nighttime (12.0 ± 2.8 min, t = 3.45, df = 9, P < 0.001) (D), and post-dive surface times (min dive-1) during the nighttime were significantly shorter (2.7 ± 0.7 min) than those performed during the daytime (5.8 ± 2.2 min, t = 5.15, df = 9, P < 0.001) (C). 144

159 Depth (m) /21/ :00 5/21/ :00 5/21/ :00 Date/Time 5/21/ :00 Figure 29. Dive profile recorded for turtle #AAC261 on day 2 of her internesting interval. The arrows point to dives that contain a wiggle event or a rapid change in vertical direction (>1m). 145

160 TGT (oc) Depth (m) A) 5/18/ :00 5/19/ :00 5/20/ :00 Local Date Depth (m) o TGT ( C) 28.5 * B) 08:00 09:00 10:00 11:00 Local Time on 5/18/ :00 o TGT ( C) Depth (m) :30 16:30 17:30 18:30 Local Time on 5/19/2008 C) 19:30 Figure 30. Archival time-series gastrointestinal tract temperature (TGT) and dive profile recorded for turtle #PPQ234, during its TGT monitoring period. The turtle s TGT was monitored over a 2.4 d period (A). There were fluctuations in TGT (B-C) for turtle #PPQ234 that were indicative of ingestion events. The arrows indicate the start of ingestion events that were identified as prey ingestions based on laboratory ingestion simulations. 146

161 30 o TGT ( C) Depth (m) B) 5/22/ :00:00 A) C) 5/22/ :00:00 5/23/ :00:00 Local Date o TGT ( C) Depth (m) B) :30 12:30 13:30 14:30 Depth (m) TGT (oc) Local Time * * Wiggle event 17:30 Wiggle event 18:30 19:30 C) 20:30 Local Time Figure 31. Archival time-series gastrointestinal tract temperature (TGT) and dive profile record for turtle #AAR863 during its TGT monitoring period. The turtle s TGT was monitored over a 0.8 d period (A). There were sudden fluctuations in the turtle s TGT that were indicative of ingestion events (B-C). The arrows indicate the start of ingestion events identified as prey ingestions, determined by based on laboratory ingestion simulations, and asterisks indicate the start of unidentified ingestion events. 147

162 31 TGT (oc) 29 Depth (m) Turtle #AAV935 enters ocean 5/24/200700:00:00 5/24/200712:00:00 5/25/200700:00:00 Local Date Figure 32. Archival time-series gastrointestinal tract temperature (TGT) and dive profile recorded for turtle #AAV935 during its TGT monitoring period. There were no sudden fluctuations in the turtle s TGT that were indicative of ingestion events. 148

163 TGT (oc) Depth (m) B) 5/12/200800:00 A) C) 5/14/200800:00 5/16/200800:00 5/18/200800:00 5/20/200800:00 5/22/200800:00 Local Date o TGT ( C) Depth (m) B) 13:00 o TGT ( C) :00 15:00 16:00 Local Time on 5/12/ : Depth (m) C) :00 16:00 17:00 18:00 19:00 20:00 21:00 Local Time on 5/16/2008 Figure 33. Archival time-series gastrointestinal tract temperature (TGT) and dive profile recorded for turtle #AAR264, during its TGT monitoring period. The turtle s TGT was monitored over a 10 d period (A) during one of its internesting intervals. There were sudden fluctuations in the turtle s TGT that were indicative of ingestion events (B-C). The arrows indicate the start of ingestion events identified as prey ingestions, determined by based on laboratory ingestion simulations, and asterisks indicate the start of unidentified ingestion events. 149

164 31 TGT (oc) Depth (m) A) 5/14/200800:00 5/21/200800:00 5/28/200800:00 6/4/200800:00 6/11/200800:00 Local Date B) 08:00 09:00 10:00 11: :00 C) 12:00 13:00 Local Time on 5/15/ TGT ( C) 15:00 16: o Depth (m) Depth (m) 14:00 Local Time on 5/18/2008 o TGT ( C) 29.0 * Depth (m) Depth (m) : o 28.0 TGT ( C) o TGT ( C) D) :00 08:00 09:00 10:00 11:00 12:00 Local Time on 5/21/ :30 E) 11:30 12:30 Local Time on 6/05/ :30 Figure 34. Archival time-series gastrointestinal tract temperature (TGT) and dive profile recorded for turtle #XXZ465, during its TGT monitoring period. The turtle s TGT was monitored over a 28 d period (A) during one of its internesting intervals. There were sudden fluctuations in the turtle s TGT that were indicative of ingestion events (B-C). The arrows indicate the start of an ingestion event identified as a prey ingestion, determined by based on laboratory ingestion simulations, and the asterisk indicates the start of an unidentified ingestion event. 150

165 TGT (oc) Depth (m) B) 500 C) 5/19/200800:00:00 5/21/200800:00:00 A) E) D) 5/23/200800:00:00 5/25/200800:00:00 5/27/200800:00:00 Local Date * :30 o TGT ( C) 28.0 Depth (m) Depth (m) o TGT ( C) B) 08:30 09:30 10: :15 11:30 Wiggle event 14:15 C) 15:15 Local Time on 5/20/2008 Local Time on 5/19/ :00 Wiggle event 05:00 06: o 28.0 * TGT ( C) Depth (m) Depth (m) o TGT ( C) : D) 07:00 08: :00 Local Time on 5/21/ :00 E) 01:00 02:00 03:00 04:00 Local Time on 5/24/2008 Figure 35. Archival time-series gastrointestinal tract temperature (TGT) and dive profile recorded for turtle #AAC261 during its TGT monitoring period. The turtle s TGT was monitored over a 9.1 d period (A), during one of its internesting intervals. There were sudden fluctuations in the turtle s TGT that were indicative of ingestion events (BC). The arrows indicate the start of an ingestion event identified as a prey ingestion, determined by based on laboratory ingestion simulations, and the asterisk indicates the start of an unidentified ingestion event. 151

166 TGT (oc) Depth (m) C) B) D) 5/20/200800:00:00 5/23/200800:00:00 5/26/200800:00:00 5/29/200800:00:00 * Wiggle event :00 07:00 08:00 09:00 Local Time on 5/20/2008 * * o 28.0 Depth (m) D) 16:00 17:00 18:00 19: C) 11:00 12:00 13:00 Local Time on 5/22/ : :00 10:00 o 29.0 B) TGT ( C) 05:00 TGT ( C) 6/4/200800:00:00 o 28.0 Depth (m) Depth (m) o TGT ( C) Depth (m) 6/1/200800:00:00 Local Date 29.0 TGT ( C) A) E) * * * * * E) :00 Local Time on 5/22/ :00 05:00 06:00 07:00 08:00 Local Time on 5/23/2008 Figure 36. Archival time-series gastrointestinal tract temperature (TGT) and dive profile recorded for turtle # during its TGT monitoring period. The turtle s TGT was monitored over a 16.0 d period (A) during one of its internesting intervals. There were sudden fluctuations in the turtle s TGT that were indicative of ingestion events (B-C). The arrows indicate the start of an ingestion event identified as a prey ingestion, determined by based on laboratory ingestion simulations, and the asterisk indicates the start of an unidentified ingestion event. 152

167 31 TGT (oc) Depth (m) B) 500 C) D) 5/21/200800:00:00 5/23/200800:00:00 A) E) 5/25/200800:00:00 5/27/200800:00:00 Local Time 29.0 TGT ( C) :00 10:00 11: B) 08:00 13:00 14:00 15:00 Local Time on 5/20/ o TGT ( C) 29.5 o D) :30 10:30 11:30 12:30 Depth (m) TGT ( C) Depth (m) C) 12:00 12:00 Local Time on 5/20/ o 27 Depth (m) Depth (m) o TGT ( C) 29 13: :30 Local Time on 5/23/2008 E) 10:30 11:30 12:30 Local Time on 5/25/2008 Figure 37. Archival time-series gastrointestinal tract temperature (TGT) and dive profile recorded for turtle #AAQ943 during its TGT monitoring period. The turtle s TGT was monitored over a 16.0 d period (A) during one of its internesting intervals. There were sudden fluctuations in the turtle s TGT that were indicative of ingestion events (B-C). The arrows indicate the start of an ingestion event identified as prey ingestion, determined by based on laboratory ingestion simulations. 153

168 o Temperature ( C) Depth (m) /23/ :00 5/25/ :00 5/27/ :00 5/29/ :00 Date and Time (M/DD/YYYY Local Time) Figure 38. Gastrointestinal temperature (TGT) and ambient temperature (TA) and depth recorded for turtle #XXZ465 during its tracking period from 21 May 2008 at 18:00 to 29 May 2008 at 18:

169 10-1 Dive Frequency (Dives h ) Daytime ( hrs) Nighttime ( hrs) AV93 J) AAR26 L) XXZ46 ) PQ23 AR86 AC26 ) AAQ94 ( ( (A) P (D) A (G) A (R) A (S (Q Turtle ID No. Turtle ID # Mean Maximum Dive Depth (m) 200 Daytime ( hrs) Nighttime ( hrs) AV93 J) AAR26 L) XXZ46 PQ2 AR8 AC2 ) AQ9 ( ( (A) P (D) A (G) A (R) A (S) A (Q Turtle ID No. Turtle ID # Figure 39. Diel dive frequency and maximum depth by turtles, while their gastrointestinal temperatures were monitored. Bars represent the mean and the error bars represent the standard error of the mean. 155

170 25 Daytime ( hrs) Nighttime ( hrs) Dive Duration (min) AV93 J) AAR26 L) XXZ46 PQ23 AR86 AC26 ) AQ9 ( ( (A) P (D) A (G) A (R) A (S) A (Q Turtle ID No. Turtle ID # Daytime ( hrs) Nighttime ( hrs) Post-dive Surface Time (min) AV93 J) AAR26 L) XXZ46 AR86 AC26 ) ) AAQ94 PQ2 ( ( (A) P (D) A (G) A (R) A (S (Q Turtle ID No. Turtle ID # Figure 40. Diel post-dive surface time and dive duration by turtles, while their gastrointestinal temperatures were monitored. Bars represent the mean and the error bars represent the standard error of the mean. 156

171 No. of Ingestions 20 #PPQ234 #AAR863 #AAR264 #XXZ465 #AAC261 # #AAQ Days at Sea No. of Ingestions 20 #PPQ234 #AAR863 #AAR264 #XXZ465 #AAC261 # #AAQ Days at Sea Figure 41. Frequency of ingestion events identified in the gastrointestinal tract temperature (TGT) data for turtle #PPQ235, #AAR863, #AAR264, #XXZ465, #AAC261, # and #AAQ943, recorded during one of their internesting intervals. The majority of ingestion events were documented during internesting days 0-4. Colored bars represent ingestion events by individual turtles and colored squares correspond to the last day of when TGT were monitored for individual turtles. 157

172 -1 Integral Index Value (s ) Seawater ingestion simulations (g) Jellyfish ingestion simulations (g) (total mass = 500 g) Figure 42. Integral index values of various combinations of jellyfish and artificial seawater used for laboratory ingestion simulations. The integral index values for the ingestion simulation group comprised of 300 g of jellyfish with 200 g seawater and 500 g of jellyfish were significantly greater (P < 0.017) in comparison with laboratory simulation group comprised of 650 g of seawater (Mann-Whitney U-Test). The dotted line delineates an integral index value of 482 s-1, which was the mean integral index value for the ingestion simulation group comprised of 300 g of jellyfish with 200 g seawater, and was used as the minimum integral index value for identifying leatherback prey ingestions in this study. The jellyfish ingestion simulation groups identified as 100 to 300 g had a total mass of 500 g the difference in mass accounted for by seawater (e.g. 300 g jellyfish with 200 g of seawater equals 500 g). 158

173 Integral Index Value (s-1) Integral Index Value (s-1) A) #PPQ B) #AAR C) #AAR Integral Index Value (s-1) Integral Index Value (s-1) E) #AAC Integral Index Value (s-1) TGT Fluctuation F) #AAQ TGT Fluctuation F) # TGT Fluctuation TGT Fluctuation D) #XXZ TGT Fluctuation 0 8 TGT Fluctuation Integral Index Value (s-1) Integral Index Value (s-1) TGT Fluctuation Figure 43. Integral index values for sudden fluctuations in TGT of leatherback turtles (A-G) that were indicative of ingestion events. Integral index values 481 s1 were considered to be ingestions of prey and those having values <481s-1 could not be classified as prey ingestions. The red line marks an integral index value of 482 s-1. Bars represent individual TGT fluctuations considered for integral index value analysis. A total of 133 fluctuations in TGT had integral index values 481 s-1 and were identified as prey ingestions. 159

174 No. of prey ingestion events 20 #PPQ234 #AAR863 #AAR264 #XXZ465 #AAC261 # #AAQ No. of prey ingestion events Days at Sea 20 #PPQ234 #AAR863 #AAR264 #XXZ465 #AAC261 # #AAQ Days at Sea Figure 44. Frequency of sudden fluctuations in gastrointestinal temperature characterized as prey ingestion events by turtle #PPQ235, #AAR863, #AAR264, #XXZ465, #AAC261, # and #AAQ943, recorded during one of their internesting intervals. The majority (69%) of prey ingestion events were documented during internesting days 0-3. Colored bars represent ingestion events by individual turtles and colored squares correspond to the last day of when TGT were monitored for individual turtles. 160

175 Prey Preyingestion Ingestiontime Time -maximum Maximum depth time Depth Time(min) (min) Time post-maximum dive depth Time prior to maximum dive depth -15 PPQ234 AAR863 AAR264 XXZ465 AAC AAQ943 Turtle no. Figure 45. The difference in time between the start time of prey ingestion events and time at which turtle reached maximum depth (prey ingestion time time at maximum depth of ingestion dives) by seven leatherback turtles from the St. Croix, USVI nesting population during one of their internesting intervals. Prey ingestions typically started within -1.0 to 5.0 min from the time the turtles reached the maximum depth of their prey ingestion dives. Boxes contain lower to upper quartile data (25-75%) and the mean is represented by the small square inside the box. The line inside the boxes represents the median. Whiskers ends contain 99% of the data and outliers are shown as asterisks. 161

176 Descent Phase Bottom Phase Ascent Phase Surface Phase Freq. of Prey Ingestion Events (%) * A) 0 a ab b a 'V' 'U' Dive Shape 125 With Wiggle Event Without Wiggle Event Freq. of Prey Ingestion Dives B) 0 'V' 'U' Dive Shape Figure 46. Frequency of prey ingestion events recorded during each phase of the dive (descent, bottom, ascent and surface) for V-shaped and U-shaped dive types (A) and the frequency of wiggle events associated with V-shaped and U-shaped dive types (B) by seven leatherback turtles in the Caribbean Sea during their internesting interval. A significant difference (*) was detected between the proportion of dive phases associated with prey ingestions for V-shaped dives (one-way ANOVA, F3,24 = 11.1, P < 0.001; followed by a post-hoc Tukey test (α = 0.008)), but not for U-shaped dives (KruskalWallis, H 2,28 = 6.23, P = 0.10). The significance level for each Tukey test was adjusted using a Bonferroni correction method (α/n or 0.05/6). Different letters (a, b) indicate differences (P < 0.008) among groups (A). The majority of wiggle events associated with prey ingestion dives were documented during V-shaped dives (84.3%), but the proportion of the total number of V-shaped prey ingestion dives that contained a wiggle event was significantly less than proportion of and U-shaped prey ingestion dives that contained a wiggle event (t = 7.8, df = 3, P = 0.004). 162

177 Sea Surface Height Deviation (cm) Low: High: 17.0 Figure 47. Map of cubic interpolated tracks and locations of prey ingestion events for turtle #PPQ234 and #AAR863, recorded during each turtles gastrointestinal monitoring period, overlayed onto a sea surface height deviation image. Prey ingestion event locations were based on the start time of sudden fluctuations in the gastrointestinal track temperature of turtles that were characterized as prey ingestions. Sea surface altimetry data (Jason-1, TOPEX/Poseidon, ERS 1/2, and GFO) is one-day composite image for May 14, Dotted line delineates the 1000 m isobath. 163

178 A) B) Sea Surface Height Deviation (cm) Low: High: 20.0 Figure 48. Maps display cubic interpolated tracks and locations of prey ingestion events for turtle #AAR264 from 12-May-2008 to 21-May-2008 (A) and for #XXZ465 from 15May-2008 to 20-May-2008 (B), recorded during each turtles gastrointestinal monitoring period, overlayed onto sea surface height deviation image. Foraging locations were based on the start time of sudden fluctuations in the gastrointestinal track temperature of turtles that were characterized as ingestion events. Sea surface altimetry data (Jason-1, TOPEX/Poseidon, ERS 1/2, and GFO) is a one-day composite for 14-May Dotted line delineates the 1000 m isobath. 164

179 A) B) Sea Surface Height Deviation (cm) Low: High: 20.0 Figure 49. Maps display cubic interpolated tracks and locations of prey ingestion events for turtle #XXZ456 from 21-May-2008 to 27-May-2008 (A) and from 28-May-2008 to 03-June-2008 (B), recorded during the turtles gastrointestinal monitoring period, overlayed onto sea surface height deviation image. Foraging locations were based on the start time of sudden fluctuations in the gastrointestinal track temperature of turtles that were characterized as prey ingestion events. Sea surface altimetry data (Jason-1, TOPEX/Poseidon, ERS 1/2, and GFO) is a one-day composite for 21-May-2008 (A) and for 28-May-2008 (B). Dotted line delineates the 1000 m isobath. 165

180 A) B) Sea Surface Height Deviation (cm) Low: High: 20.0 Figure 50. Maps display cubic interpolated tracks and locations of prey ingestion events for turtle #XXZ456 from 04-June to 11-June-2008 (A) and turtle #AAC261 from 19-May-2008 to 28-May-2008 (B), recorded during each turtles gastrointestinal monitoring period, overlayed onto a satellite-derived sea surface height deviation image. Foraging locations were based on the start time of sudden fluctuations in the gastrointestinal track temperature of turtles that were characterized as prey ingestion events. Sea surface altimetry data (Jason-1, TOPEX/Poseidon, ERS 1/2, and GFO) is a one-day composite for 04-June-2008 (A) and 21-May-2008 (B). Dotted line delineates the 1000 m isobath. 166

181 A) B) Sea Surface Height Deviation (cm) Low: High: 20.0 Figure 51. Maps display cubic interpolated tracks and locations of prey ingestion events for turtle # from 20-May-2008 to 27-May-2008 (A) and for turtle #PPQ234 from 20-May-2008 to 28-May-2008 (B), recorded during each turtles gastrointestinal monitoring period, overlayed onto a satellite-derived sea surface height deviation image. Foraging locations were based on the start time of sudden fluctuations in the gastrointestinal track temperature of turtles that were characterized as prey ingestion events. Sea surface altimetry data (Jason-1, TOPEX/Poseidon, ERS 1/2, and GFO) is a one-day composite for 21-May Dotted line delineates the 1000 m isobath. 167

182 A) Sea Surface Height Deviation (cm) High: 0.20 High : 0.2 Low : -0.2 B) Low: C) Figure 52. Maps shows satellite derived sea surface height (SSH) anomaly data on May 14, 2008 (A), May 28, 2008 (B), and June 11, 2008 (C). Positive (red) and negative (blue) SSH anomalies were detected within < 200 km of St. Croix during the 2008 field season and moved westward. Altimetry data was collected by Jason-1, TOPEX/Poseidon, ERS 1/2, and GFO satellites. 168

183 Figure 53. Diffuse attenuation coefficient at 490 nm in the Eastern Caribbean Sea between 5/20/2008 and 5/27/2008, recorded by Moderate Resolution Imaging Spectrodiometer (MODIS) sensors aboard NASA s Aqua and Terra satellite platforms. Note the increase in diffuse attenuation within 200 km south of St. Croix indicating increased productivity in this area. 169

184 A) B) Figure 54. Maps of interpolated locations of tracked turtles in 2007 (A) and 2008 (B) when they displayed sudden fluctuations in their gastrointestinal track temperature that were characterized as prey ingestion events, overlayed onto satellite-derived sea surface concentrations of chlorophyll-a image (SeaWiFI). Prey ingestion events were recorded between May 17, 2008 to May 22, 2007 (A) and May 12, 2008 to June Chlorophyll a data (MODIS) are monthly composite images from May 16, 2007 to June 16, 2007 (A) and from May 16, 2008 to June 16, 2008 (B). 170

185 Figure 55. Internesting foraging hot-spots for gravid leatherback turtles from the St. Croix, USVI nesting population, based on records of sudden fluctuations in the turtles gastrointestinal tract temperature that were characterized as prey ingestion events in 2007 and ` 171

186 Descent Phase Bottom Phase Ascent Phase Surface Phase Freq. of Unidentified Ingestion Events (%) * a a b a 'V' 'U' Dive Shape Figure 56. Frequency of sudden fluctuation in gastrointestinal tract temperature in seven leatherback turtles between nesting events at St. Croix, USVI recorded during each phase of the dive (descent, bottom, ascent and surface) for V-shaped and U-shaped dive type that were unidentified/non-characterized ingestion events. A significant difference (*) was detected between the proportion of the dive phases associated with unidentified ingestion events for V-shaped dives, but not for U-shaped dives (Kruskal-Wallis test (α = 0.05) followed by a post-hoc Mann-Whitney test. The significance level for each MannWhitney test was adjusted using a Bonferroni correction method (α/n or 0.05/6), and different letters (a, b) indicate differences (P < 0.008) among groups. 172

187 INT ( C s ) Energy (J) Figure 57. The relationship between INT and energy for laboratory ingestion simulations with jellyfish (closed circles) and seawater (open circles). The linear regression coefficients (i.e. slope) for ingestion simulations with jellyfish (INT = 0.31E , r2 = 0.98, n = 6) and seawater (INT = 0.02E , r2 = 0.50, n = 16) were significantly different (F test, P < 0.001). 173

188 Figure 58. Map showing the current critical for leatherbacks in waters adjacent to SPNWR, St. Croix, U.S. Virgin Islands, which was redesignated by the National Marine Fisheries Service in Map was obtained from the National Oceanic and Atmospheric Administrator s National Marine Fisheries Service (NOAA Fisheries Service) website 174