PHYSIOLOGICAL STATUS AND POST-RELEASE MORTALITY OF SEA TURTLES RELEASED FROM GILLNETS IN THE LOWER CAPE FEAR RIVER, NORTH CAROLINA. Jessica E.

Transcription

1 PHYSIOLOGICAL STATUS AND POST-RELEASE MORTALITY OF SEA TURTLES RELEASED FROM GILLNETS IN THE LOWER CAPE FEAR RIVER, NORTH CAROLINA Jessica E. Snoddy A Thesis Submitted to the University of North Carolina Wilmington in Partial Fulfillment of the requirements for the Degree of Master of Science Department of Biology and Marine Biology University of North Carolina Wilmington 2009 Approved by Advisory Committee Andrew J. Westgate Thomas E. Lankford Amanda L. Southwood Chair Accepted by Dean, Graduate School

2 TABLE OF CONTENTS ABSTRACT... iv ACKNOWLEDGMENTS...v DEDICATION... vi LIST OF TABLES... vii LIST OF FIGURES... viii INTRODUCTION...1 Sea Turtle Biology...1 Nearshore Behavior and Movements...2 Fishing Threats...4 Diving Physiology and Stress of Capture...6 Project Rationale...9 CHAPTER 1. BLOOD BIOCHEMISTRY OF SEA TURTLES RELEASED FROM GILLNETS IN THE LOWER CAPE FEAR RIVER, NORTH CAROLINA, USA...12 ABSTRACT...12 INTRODUCTION...13 METHODS...16 Field Procedures...16 Blood Analysis...19 Statistical Analysis...21 RESULTS...22 DISCUSSION...23 Management Implications...28 ii

3 CHAPTER 2. MOVEMENTS OF JUVENILE SEA TURTLES RELEASED FROM GILLNETS IN THE CAPE FEAR RIVER, NORTH CAROLINA...38 ABSTRACT...38 INTRODUCTION...39 METHODS...42 Field Procedures...42 Analysis of Location Data...44 Assessment of Mortality...45 RESULTS...48 Movements and Habitat Utilization...48 Post-Release Mortality...49 DISCUSSION...52 CONCLUSIONS...86 REFERENCES...89 iii

4 ABSTRACT By-catch of sea turtles in the North Carolina coastal gillnet fishery has been implicated as a significant source of mortality, and numerous management measures have been taken to either minimize the detrimental effects of capture (gear modification and attendance requirements) or to reduce or prevent capture of sea turtles in fishing gear (time and area-based closures). Management decisions regarding sea turtle interactions and acceptable take levels for this fishery are currently based on analyses of fishing effort, observed bycatch of sea turtles, sea turtle strandings, and estimates of mortality rates for sea turtles due to fisheries interactions derived from these data. The purpose of this study was to directly evaluate the impact that entanglement in gillnets has on the physiological status and post-release behavior of sea turtles, and to use these data to refine our estimates of post-release mortality for the gillnet fishery. I conducted physical examinations and collected blood samples from eighteen sea turtles captured in shallowset gillnets in the lower Cape Fear River. Satellite and VHF radio transmitters were deployed on fourteen of these turtles so that I could monitor their post-release movements and document mortality events. I found that entanglement in gillnets resulted in severe disruptions of blood biochemistry. One confirmed post-release mortality and three suspected post-release mortalities were documented during the course of the study. Integration of physiology and behavior data with observer and stranding data shows promise as a means of refining mortality estimates associated with gillnet entanglement. iv

5 ACKNOWLEDGMENTS Special thanks go to my family and friends for supporting me through the years. Many thanks to my advisor, Dr. Amanda Southwood, for the opportunity to be a part of her lab and this research. I am forever grateful for her guidance and support. I would like to thank my lab mates, James Casey, Lisa Goshe, and Leigh Anne Harden, for their advice and the laughs. Field work would not have been possible without the expertise of Captain Jeff Wolfe. Many laughs will be remembered from long days on the river. Thanks to Elizabeth Brandon, Diana Bierschenk, and countless other volunteers in the field. Catherine McClellan provided valuable information on techniques and problems associated with tracking juvenile green turtles in nearshore environments, and Theresa Thorpe gave valuable information on where to find sea turtles in the river. Jean Beasley, Craig Harms, and Chris Butler provided blood samples from captive sea turtles. Thanks to David Owens and Gaëlle Blanvillain for help with corticosterone analysis. I thank Francine Christiano and Marion Landon for assistance with analyzing blood lactate concentrations. Dr. James Blum assisted with statistical analysis. North Carolina Sea Grant provided funding for this research. Finally, I would like to thank University of North Carolina Wilmington Department of Biology and Marine Biology, especially my committee members, for their guidance and support throughout my studies. v

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

7 LIST OF TABLES Table Page 1. Capture data for Kemp s ridley and green sea turtles entangled in shallow-set gillnets in the lower Cape Fear River during May - October Descriptive statistics for blood parameters measured in juvenile Kemp s ridley and sea turtles immediately following removal from gillnets (INITIAL samples) Results for paired t-test analysis of INITIAL and PRE-RELEASE values of blood parameters in Kemp s ridley and green sea turtles Mortality classification, total tracking duration and percent of high quality Transmissions received for all tracked greens and Kemp s ridley sea turtles...51 vii

8 LIST OF FIGURES Figure Page 1. Blood lactate, LDH, CPK, glucose, phosphorus and corticosterone for sea turtles entangled in gillnets from minutes Photograph of satellite and radio tag deployment on Kemp s ridley turtle Lk Capture locations and 25 and 50% volume contours of all filtered location data for all turtles captured in the lower Cape Fear River May October Percent of each location class transmissions received from all turtles released from gillnets Map of filtered location data for green turtle Cm a. Map of filtered location data for green turtle Cm b. Expanded view of filtered location data in lower Cape Fear River for green turtle Cm 13 Lk Map of filtered location data for Kemp s ridley turtle Lk Map of filtered location data for Kemp s ridley turtle Lk Map of filtered location data for green turtle Cm Map of filtered location data for green turtle Cm Map of filtered location data for green turtle Cm Map of filtered location data for green turtle Cm Map of filtered location data for green turtle Cm Map of filtered location data for green turtle Cm Map of filtered location data for green turtle Cm Map of filtered location data for green turtle Cm Map of filtered location data for green turtle Cm Map of filtered location data for Kemp s ridley turtle Lk viii

9 19a. Transmission pattern of entire track duration for Kemp s ridley turtle Lk b. Transmission pattern of entire track duration for Kemp s ridley turtle Lk c. Transmission pattern of entire track duration for green turtle Cm d. Transmission pattern of entire track duration for green turtle Cm e. Transmission pattern of entire track duration for green turtle Cm f. Transmission pattern of entire track duration for green turtle Cm g. Transmission pattern of entire track duration for green turtle Cm h. Transmission pattern of entire track duration for green turtle Cm i. Transmission pattern of entire track duration for green turtle Cm j. Transmission pattern of entire track duration for green turtle Cm k. Transmission pattern of entire track duration for green turtle Cm l. Transmission pattern of entire track duration for green turtle Cm m. Transmission pattern of entire track duration for green turtle Cm n. Transmission pattern of entire track duration for Kemp s ridley turtle Lk ix

10 INTRODUCTION Sea Turtle Biology Sea turtles are long-lived marine reptiles that develop and mature very slowly (Musick and Limpus, 1997). These characteristics have important implications for conservation of threatened and endangered sea turtle species. Juveniles that do not survive to sexual maturity will not reproduce and will not add individuals to the population, which could eventually cause a population decline. Sea turtles face many threats to their survival, including loss of nesting beaches and foraging grounds, egg and hatchling predation, pollution, and other anthropogenic factors such as boat strikes and interactions with fisheries (Lutcavage et al., 1997). Recently, there has been a growing concern about mortality associated with entanglement in commercial fishing gear, and the impact this may have on sea turtle populations (Gearhart, 2001; Griffin et al., 2006). Coastal fishing operations may have a large impact on juvenile sea turtles in particular, as they utilize nearshore developmental habitats. Sea turtles undergo long-distance migrations throughout their life, beginning with hatchling movements from breeding beach nests to oceanic habitats (Bolten and Balazs, 1995). The first few years of life at sea are referred to as the lost years because very little is known of sea turtles movements and behavior until they recruit into coastal habitats as juveniles (Bolten and Balazs, 1995). Juvenile sea turtles migrate into North Carolina s nearshore waters to forage in the spring when water temperatures (Tw) warm to approximately 13 C. When Tw begin to drop in the fall, turtles will emigrate from the North Carolina waters (Coles and Musick, 2000) and move into deeper oceanic or more

11 southerly waters. (Epperly et al., 1995a, Epperly et al., 1995b). These yearly migrations are largely dictated by sea turtles preferred Tw range.. Nearshore Behavior and Movements North Carolina waters serve as an important nursery ground for foraging and developing juvenile sea turtles, and also provide breeding habitat for adult sea turtles. The three species of sea turtles most frequently encountered in North Carolina are loggerheads (Caretta caretta), greens (Chelonia mydas) and Kemp s ridleys (Lepidochelys kempii) (Epperly et al., 1995a). Kemp s ridleys only nest in Rancho Nuevo, Mexico, but frequent North Carolina as juveniles to feed on crabs and small benthic invertebrates (Mortimer, 1995). Adult loggerheads nest in coastal North Carolina (Hawkes et al., 2005), and juveniles of this species feed on benthic invertebrates and fish (Mortimer, 1995). Adult green turtles also nest in North Carolina, and juvenile greens feed on algae in the shallow coastal waters (Mortimer, 1995). Movements of juvenile turtles in the summer months are usually limited to a small home range (~5 km 2 ) (Seminoff et al., 2002; Makowski et al., 2005). Site fidelity is based on availability of food and nightly resting spots (Brill et al., 1995; Makowski et al., 2005). Sea turtles display strong homing behaviors for particular areas and routinely pinpoint specific foraging sites. They also exhibit strong site fidelity from hundreds of kilometers away when migrating (Lohmann et al., 1997). Juvenile loggerheads that were displaced after capture in Pamlico Sound, North Carolina returned to the same capture location within the same season and in subsequent years (Avens et al., 2003). 2

12 Sea turtles display predictable daily movement and diving patterns during summer months. During the morning and late afternoon they are found in shallow waters, but move into deeper areas when Tw increases around mid-day. Mendonça (1983) found that juvenile greens utilize shallow seagrass flats for 70% of daylight hours in a Florida lagoon. It is typical for turtles to return to the same resting spots night after night (Mendonça, 1983; Southwood et al., 2003). In general, dives are longer during nighttime resting periods than during daytime foraging. Turtles are most active from dawn to dusk while foraging and they dive frequently and to shallow depths during those times (Southwood et al., 2003). The majority of research on movements and seasonal utilization of coastal North Carolina waters by sea turtles has focused on the Core and Pamlico Sound region. Aerial surveys of Pamlico and Core Sounds indicate that there is a seasonal migration of sea turtles into and out of this region and that nearly all turtles found inshore are juveniles (Epperly et al., 1995a; Epperly et al., 1995b). Read et al. (2004) found that green turtles in Pamlico Sound occupy very shallow waters (0-2 m), while loggerheads and Kemp s ridleys are found at more variable depths (0-6 m). Turtles were found in Tw of 8º C to 30º C, but spent 90% of the time in Tw greater than 14º C (Read et al., 2004). Juvenile sea turtles also use the lower Cape Fear River as a seasonal foraging habitat, but very little is known of their behavior and movement patterns in this area, with most of the information coming from interactions with fisheries. 3

13 Fishing Threats All sea turtles are protected by the Endangered Species Act of 1973, so there is a federal mandate to minimize the negative impacts of commercial fishing on populations. Commercial fishing in North Carolina peaks during the summer months, which coincides with the presence of large numbers of juvenile sea turtles in coastal waters (Gearhart, 2001; Read and Foster, 2004; Price, 2005). Fishing interactions with sea turtles have become a critical concern for fisheries managers, as several mass stranding events have occurred during peak fishing season since 1995 (Gearhart, 2001; Price, 2005). Sea turtles encounter a variety of commercial fisheries while in nearshore habitats, but encounters with coastal gillnets are thought to be the primary contributor to stranding events and sea turtle mortality (NC Division of Marine Fisheries, 2005). These fisheries target mostly monkfish and flounder, but bycatch of sea turtles in gillnets is a common occurrence (Federal Register, December 2002). A gillnet can be described as a mesh fence in the water column, usually running perpendicular to the shoreline. A float line is attached to the top of the net and a weighted lead line pulls the bottom of the net down to the sea floor. Sea turtles that swim into a gillnet may become entangled, making it difficult for them to surface to breathe. The degree of injury and physiological disruption for sea turtles entangled in gillnets can vary depending on the gear type. Large mesh gillnets (mesh size >5 inches) (Federal Register, December 2002) are most dangerous to sea turtles, as the holes are large enough to wrap around the turtle s head or limbs. Large mesh gillnets were historically used in fisheries that targeted sea turtles before the establishment of the Endangered Species Act (Federal Register, December 2002). In Pamlico Sound, North 4

14 Carolina, large mesh net is prohibited due to the high risk to sea turtles (Federal Register, December 2002). Small mesh gillnets (mesh size 5 inches) are considered a lesser threat to sea turtles because turtles are less likely to become entangled (Federal Register, December 2002). The impact of gillnet entanglement on sea turtle physiology and survival may also depend on the depth at which the gillnet is set and manner in which net is fished. Shallow-set gillnets are set in waters less than 3 m deep, are typically left to soak overnight, and range in length from 500 to 2000 yards (Gearhart, 2001; Federal Register, September 2002; Price, 2005). In contrast, deep-set gillnets are set at depths of 3 m or greater, may soak for up to 3 days at a time, and range in length from 2,000 10,000 yards (Gearhart, 2001; Federal Register, September 2002; Price, 2005). Deep-set gillnets pose a much larger threat to sea turtles, as the turtles are unable to surface to breathe while entangled. In-net mortality rates for deep-set gillnets are high (Gearhart, 2001), and use of this gear is now prohibited (Federal Register, September 2002). The majority of sea turtles caught in shallow-set nets are released alive (Gearhart, 2001). Lower rates of in-net mortality for shallow-set gillnets may be due to the fact that turtles have less difficulty reaching the surface to breathe while entangled, or that they have not been in the net for a long period of time. The National Marine Fisheries Service and NC DMF have implemented numerous mitigation measures to address the problem of sea turtle bycatch in the gillnet fisheries of Core and Pamlico Sounds, but until recent years the lower Cape Fear River has received less attention from regulatory agencies. Anecdotal evidence from local fishermen suggests that there has been an increase in the number of juvenile turtles in the 5

15 Cape Fear River over the last 5-10 years, which may be due to an increase in algae in the area (Jeff Wolfe, David Beresoff, pers. comm.). Juvenile greens, loggerheads and Kemp s ridleys are incidentally captured in gillnets in the lower Cape Fear River, with greens being caught most frequently. Incidental capture of sea turtles is highest between June and September (Thorpe and Beresoff, 2005). Thorpe and Beresoff (2005) documented 23 green turtles, 4 Kemp s ridleys and 6 loggerheads captured in 40 gillnet trips over the course of one summer. Fisheries managers now recognize that the lower Cape Fear River provides important seasonal foraging habitat for sea turtles during the summer months, and the NC DMF has placed restrictions on gillnetting in the lower Cape Fear River between June and August. Gillnet fisherman are required to attend their nets at all times and remove any turtles that become entangled, however there are no mesh size or depth restrictions (Pate, DMF, 2006). Due to this net-attendance restriction, the summer gillnet fishery in this area has essentially closed because fisherman are unwilling to stay with their nets throughout the 12 hours of a typical set. Management agencies are eager to learn more about sea turtle habitat utilization and behavior in this region so that they may refine mitigation measures. Diving Physiology and Stress of Capture Sea turtles can dive for prolonged periods and to great depths voluntarily because of their low reptilian metabolic rates, efficient blood oxygen transport mechanisms, and moderate tolerance to hypoxia (Lutcavage and Lutz, 1997). Typical voluntary dive times vary according to size, species and habitat. Larger turtles are able to stay submerged for 6

16 longer periods than juveniles, due to the nature of scaling of oxygen stores and metabolic rate with body size. Oxygen stores scale directly with mass, whereas mass specific metabolic rate scales in an allometric manner, with an exponent of approximately 0.75 rather than 1. Therefore, larger turtles have larger oxygen stores with relation to massspecific metabolic rate than do smaller turtles (Schmidt-Nielsen, 1984). Juvenile loggerheads routinely dive for min, juvenile green turtles typically dive for 9-23 min, and juvenile Kemp s ridleys normally dive for min (Lutcavage and Lutz, 1997). Studies of natural diving behavior indicate that juvenile turtles typically spend very little time at the surface (Gitschlag, 1996; Lutcavage and Lutz, 1997). This is likely due to the fact that, as with most air-breathing diving vertebrates, sea turtles rely primarily on aerobic metabolism while submerged (Kooyman, 1985). The lung is the major oxygen store for shallow diving turtles in coastal areas (Lutcavage and Lutz, 1997). For example, loggerheads, store 72% of oxygen in the lungs, and tissue oxygen stores are minor (Lutz and Bentley, 1985). Oxygen stores are rapidly replenished and CO 2 is eliminated during short surfacing intervals, and the number of breaths increases with increasing dive time (Lutcavage and Lutz, 1991). Sea turtles will generally surface for air before oxygen stores run out, but are able to cope with progressively decreasing blood oxygen stores through high blood oxygen affinity and strong blood buffering capabilities. Thus, blood ph remains stable and lactate levels also remain low during the course of routine aerobic dives (Lutcavage and Lutz, 1991; Lutz and Bentley, 1985). Although sea turtles can stay submerged for minutes during voluntary dives, forced submergence due to net entanglement can be lethal (Lutz and Bentely, 7

17 1985). Turtles caught in a net will struggle in attempts to escape and surface for air, and oxygen stores will be rapidly depleted. Tethered adult green sea turtles diving in attempts to escape depleted their oxygen stores within 15 min (Wood, 1984), and turtles forcibly submerged in nets for as little as 30 minutes may drown (Lutcavage and Lutz, 1997). Juvenile turtles are more susceptible to drowning than adults, due to their high mass-specific metabolic rates and relatively small oxygen stores (Berkson, 1966; Schmidt-Nielsen, 1984). Turtles submerged in nets for longer than an hour have a low chance of survival (Lutz and Bentley, 1985; Federal Register, 2004). Forcibly submerged turtles must resort to anaerobic metabolism when there is not enough oxygen to support aerobic metabolic pathways, and this results in a build-up of lactic acid and acidification of the blood (Lutz, 1997). Increased CO 2 levels (respiratory acidosis) may also contribute to a drop in blood ph during forced submergence (Stabenau et al., 1991; Stabenau and Vietti, 2003). Alteration in blood ph can damage proteins, disrupt cellular function, and affect the ability of blood to coagulate properly (Soslau, 2004). Physiological damage incurred due to net entanglement may affect the turtle s behavior and reduce its chances of survival post-release. Previous studies have noted that forcibly submerged sea turtles spend extended periods of time at the surface, presumably to recover and restore physiological homeostasis (Stabenau and Vietti, 2003). A sea turtle s recovery from lactic acid build up can take over 15 hours, depending on the severity of the acidosis (Lutz and Dunbar-Cooper, 1987). Lactate must be cleared from the blood stream and tissues by conversion back to pyruvate and subsequent processing by aerobic metabolic pathways, so ample oxygen is required to recover from metabolic 8

18 acidosis. Extended surface intervals would make turtles vulnerable to predators and increase the danger of being struck by a boat. Numerous other alterations in blood biochemistry may occur as a result of enforced submergence. For example, prolonged struggling in a net may result in injuries and muscle tissue damage, which could result in release of intracellular enzymes such as lactate dehydrogenase (LDH) and creatine phosphokinase (CPK) into the blood (Aguirre et al., 1995; Randall et al., 2002; Dahlhoff, 2004; Martínez-Amat et al., 2005; Moyes and Schulte, 2006). Damaged muscle tissue may also release intracellular ions, such as K+, Mg 2 +, and Ca 2 + into the blood (Moyes et al., 2006), and the controlled release of intracellular ions may occur as a counteractive measure against blood acidosis. Corticosterone concentration may increase in sea turtles that are forcibly submerged, due to induction of a systemic stress response (Gregory et al., 1996). Assessment of blood biochemistry at the time a turtle is captured may provide important information regarding the overall condition of the animal and the likelihood of post-release survival (Dahlhoff, 2004). Project Rationale Although most sea turtles are released alive from shallow-set gillnets, there are no existing data on the physiological condition and ultimate fate of turtles once they are released. Fisheries impacts on the sea turtle population are thought to be substantial, which is why bycatch reduction is a hotly debated subject amongst conservationists and fisheries managers (Magnuson et al., 1990; Griffin et al., 2006). Before 1995, an average of less than 200 turtles were found stranded on North Carolina s coasts per year. From 9

19 , an average of 401 turtles per year stranded in North Carolina (North Carolina Wildlife Resources Commission Sea Turtle Stranding Network Database 2005). The Pamlico Sound gillnet fishery was identified as the most likely cause of the turtle stranding events, due to the concentrated fishing effort in that region at the time of the strandings (Gearhart, 2001). Additionally, stranded turtle carcasses showed visible injuries consistent with gillnet fishery interactions, such as trauma to flippers and neck, and necropsies showed that the turtles were in otherwise good nutritional condition (Boettcher, 2000). Although it was likely that the increase in strandings was due to gillnet entanglements, other possibilities, such as an overall increase in the turtle population in this area, could not be discounted. The Division of Marine Fisheries currently bases sea turtle mortality estimates on fishing effort, stranding data, and observer coverage (Gearhart, 2001). These are only estimates of fisheries related mortality, and there is a need for refinement. Current fisheries management decisions are based on best available estimates and anecdotal evidence, but there is no direct documentation of the number of turtles that survive after release from gillnets. An assessment of the impacts of entanglement on sea turtle health and documentation of post-release mortality rates are necessary so that the current restrictions on the fishery are correctly justified and the economic impacts on fisherman are minimized. There is also a need to conduct basic research of the movement patterns and behavior of sea turtles in the lower Cape Fear River. There are numerous reports of sea turtle interactions with gillnets in this area, but little understanding of sea turtle habitat utilization and the potential for overlap with fishing operations. An understanding of sea 10

20 turtle movement patterns would allow managers to implement mitigation measures to reduce or prevent sea turtle interactions in this region. The overall goal of my research was to refine our understanding of the impacts of entanglement in gillnets on the physiology, behavior, and survivability of sea turtles. My specific research objectives were as follows: 1. Analyze blood chemistry of sea turtles released from gillnets to assess physiological impacts of entanglement. 2. Quantify post-release mortality of sea turtles released from shallow-set gillnets in the lower Cape Fear River using satellite telemetry. 3. Document seasonal movements of sea turtles in the lower Cape Fear River using satellite telemetry. 11

21 CHAPTER 1. BLOOD BIOCHEMISTRY OF SEA TURTLES RELEASED FROM GILLNETS IN THE LOWER CAPE FEAR RIVER, NORTH CAROLINA, USA Accepted for publication in the Journal of Wildlife Management ABSTRACT Mortality due to fisheries interactions has been implicated as a contributor to population decline for several species of sea turtle. The incidental capture of sea turtles in the coastal gillnet fisheries of North Carolina has received much attention in recent years, and mitigation measures to reduce sea turtle mortality due to gillnet entanglement are a high priority for managers and conservationists. Efforts to evaluate the effects of gillnet entanglement on sea turtle populations are complicated by the lack of information on health status of turtles released alive from nets and post-release mortality. I obtained blood samples from green and Kemp s ridley sea turtles captured in gillnets for minutes to assess the impacts of gillnet entanglement on blood biochemistry and physiological status. I measured concentrations of lactate, corticosterone, ions (Na +, K +, Cl -, P, Ca 2+ ), enzymes (LDH, CPK, AST), protein and glucose in the blood and also performed comprehensive physical examinations of turtles to document external indicators of health status (injuries, lethargy, muted reflexes). Statistical analyses were conducted to evaluate the effects of entanglement time on blood biochemistry and to look for correlations between blood biochemistry and results of the physical examinations. I observed a significant increase in blood lactate, LDH, CPK, phosphorus, and glucose with increased entanglement time. Alterations in blood biochemistry were generally associated with a decline in health status as indicated by results of the physical

22 examination. Although entanglement time plays an important role in determining the health status of sea turtles upon release from a gillnet, my results suggest that factors such as the depth and severity of entanglement may also have an effect on health status of turtles and the probability of post-release survival. INTRODUCTION Commercial fishing operations frequently overlap with sea turtle habitat, and unintended capture of sea turtles in fishing gear has become a problem of increasing concern for fisheries managers and conservationists (Magnuson et al., 1990; National Marine Fisheries Service and U.S. Fish and Wildlife Service, 1991; Santora, 2003; Lewison et al., 2004; Read et al., 2004; Cox et al., 2007). All sea turtles are protected by the Endangered Species Act of 1973, and the federal government has a mandate to assess and mitigate the impacts of commercial fisheries interactions on sea turtle populations (National Marine Fisheries Service and U.S. Fish and Wildlife Service, 1991). Interactions with gillnet fisheries have been implicated as a major source of mortality for loggerhead (Caretta caretta), Kemp s ridley (Lepidochelys kempii) and green (Chelonia mydas) sea turtles in coastal North Carolina, USA (Gearhart, 2001; Price, 2005). Throughout spring, summer and fall of 1999, 430 sea turtle carcasses washed up on the shores of North Carolina, accounting for 19% of the total sea turtle carcass strandings reported in the United States that year (Boettcher, 2000). Coastal gillnet fisheries were identified as a primary contributor to the mass sea turtle stranding event, due to the high 13

23 fishing effort that year and injuries on sea turtle carcasses consistent with gillnet entanglement (Boettcher, 2000; Gearhart, 2001). Since the mass stranding events of 1999, North Carolina Division of Marine Fisheries (NC DMF) and the National Oceanic and Atmospheric Administration (NOAA) Fisheries have implemented temporal and spatial fisheries closures to avoid interactions with sea turtles, as well as gear restrictions designed to minimize the impacts of entanglement on sea turtle health and survival (Federal Register, 2004; Thorpe and Beresoff, 2005). For example, use of gillnets is now restricted to shallow waters (< 3 m), as the incidence of in-net mortality for turtles caught in shallow-set gillnets is low compared with deep-set gillnets (3-6 m) (Gearhart, 2001; Price, 2005). Managers speculate that sea turtles entangled in shallow-set nets are still capable of reaching the surface to breathe, and therefore the risk of drowning in the nets is reduced (Gearhart, 2001). Although observer data and reports from fishermen indicate that sea turtles caught in shallow-set gillnets are typically released alive (Gearhart, 2001), the ultimate fate of these sea turtles is not known. Severe disruptions to normal physiological function and injuries sustained as a result of entanglement in fishing gear could lead to undocumented post-release mortality (Lutcavage and Lutz, 1991; Stabenau and Vietti, 2002; Harms et al., 2003). Previous studies have shown that sea turtles that experience hypoxia and restraint stress related to enforced submergence have significant alterations in blood biochemistry (Berkson, 1966; Stabenau et al., 1991; Gregory et al., 1996). Harms et al. (2003) observed a decrease in blood ph in loggerhead turtles submerged in trawls for up to 30 minutes. The controlled release of ions into the blood may occur as a counteractive 14

24 measure against blood acidosis. For example, a significant increase in blood K + concentration has been observed following capture and restraint in Kemp s ridley turtles (Stabenau et al., 1991; Hoopes et al., 2000). It is possible that K + ions are released by cells in exchange for H + ions to buffer changes in blood ph, although a K + /H + exchanger has not yet been identified in the cells of sea turtles (Rose, 1977; Lutz, 1997; Stabenau and Vietti, 1999; Hoopes et al., 2000). It is likely that the acidosis experienced by forcibly submerged sea turtles has both a respiratory and metabolic component. Stabenau et al. (1991) saw a 6-fold increase in lactate levels of Kemp s ridleys forcibly submerged in shrimp trawls for approximately 7.3 min. Kemp s ridley turtles captured in entanglement nets and temporarily restrained in holding tanks also experienced an increase in plasma lactate concentrations and alterations in blood ions indicative of acidbase adjustments (Hoopes et al., 2000). Increases in blood lactate concentrations and a concomitant decrease in blood ph suggest a shift towards reliance on anaerobic metabolic pathways, which could be the result of intense activity associated with escape attempts or hypoxia due to forced submergence. Prolonged anaerobiosis due to entanglement in fishing gear or restraint may leave sea turtles exhausted and vulnerable to other threats upon release from gear. There is also evidence that entanglement in fishing gear results in induction of a systemic stress response in sea turtles, which may persist following release depending on the extent of injuries suffered or stress experienced. Gregory et al. (1996) noted approximately a 3-fold increase above control values for plasma corticosterone (a hormone indicative of stress) in loggerheads that were forcibly submerged in a trawl for up to 30 minutes. 15

25 As with entanglement in trawls and other gear types, sea turtles entangled in gillnets may experience physiological disturbances related to restricted access to air, intense struggling, injuries to soft tissues, and induction of a systemic stress response (Lutz and Dunbar-Cooper, 1987; Stabenau et al., 1991; Gregory et al., 1996; Boettcher, 2000; Jessop et al., 2004). The goals of this study were to 1) investigate the physiological impacts of gillnet entanglement on juvenile sea turtles, and to 2) determine if entanglement time would be indicative of the degree of physiological disruption in gillnet captured juvenile sea turtles. I predicted that the degree of physiological disruption, as indicated by blood biochemistry, would increase with increased entanglement time and that severe disruptions in blood biochemistry would be associated with poor health status as ascertained by a physical examination. METHODS Field Procedures I captured sea turtles in 5.5 inch mesh gillnets set at depths of 1-2 m in the lower Cape Fear River, North Carolina during daylight hours (06:00-16:00) from May through October of This area consists of 3 bays enclosed by marsh to the east and a manmade rock wall which divides the bays from the river to the west. Average tide height at this location is approximately 1 m, and the rock wall is partially exposed at low tide. Gillnets remained in water for a maximum of 6 hours and were attended at all times so that I could record the time when turtles were captured and the length of time that turtles spent in the net (entanglement time). I captured 14 green turtles and 4 Kemp s ridley 16

26 turtles. Table 1 provides details on capture time and location, entanglement time, environmental conditions at the capture site, and morphometric data for each turtle. Turtles were entangled for an average of 82.3 min (range of min) and I closely monitored them for signs of distress while in the net. If turtles remained submerged for greater than 20 minutes or appeared to be in danger of drowning due to airway or swimming restriction, I immediately removed them from the net. Upon removal from nets, I brought the turtles on board the boat and restrained them in a 16 cm 43 cm padded plastic bin. The turtles were shaded from direct sunlight and were periodically sprayed with seawater. I immediately obtained a 4 ml blood sample (INITIAL sample) from the cervical sinus using heparinized vacuum tubes and a 21G x 1.5 needle (BD Vacutainer, Franklin Lakes, NJ, USA) and stored samples on ice (N = 12 for green turtles, N = 4 for Kemp s ridley turtles). INITIAL blood samples were not obtained from 2 of the 14 green turtles that I captured. Using calipers, I measured straight carapace length notch to notch (SCLnn) and straight carapace width (SCW),and calculated carapace area (cm 2 ) using the formula for the area of an ellipse: area (cm 2 ) = 1/2 (SCLnn) 1/2 (SCW). I obtained cloacal body temperature (T B ) using a digital thermometer (model 52 II, Fluke Corp., Everett, WA, USA) with a flexible veterinary probe at a depth of 4-25 cm. A passive induced transponder (PIT) tag was inserted above the left front flipper of each turtle that I captured for future identification. I examined turtles for net-inflicted external injuries and assessed reflex responses and activity levels using a protocol described by Sadove et al. (1998). Injuries were classified as minor (scrapes to skin or shell), moderate (shallow cuts to skin, bruising of skin), or severe (deep cuts that exposed muscle). I classified reflex responses to a gentle 17

27 touch to the tail, nose, and eyelid as good (immediate and strong flinch), delayed (slow or lethargic flinch), or absent. Activity was classified as high if the turtle frequently and vigorously struggled in attempts to escape. Turtles were classified as having moderate activity levels if they occasionally struggled vigorously with long periods of little to no movement in between. Activity was classified as low if turtles exhibited infrequent and weak struggling or no movement when onboard. I assigned a physical grade (A, B, C, D) based on the reflex response level, activity level, and presence/absence and severity of net-inflicted external injuries to each turtle. The physical grade criteria are as follows: A. High activity, all reflexes present and good, no injuries B. Medium activity, all reflexes present and good, minor injuries C. Medium activity, missing or delayed reflexes, moderate injuries D. Low activity, missing or delayed reflexes, severe injuries I assigned each grade based on exhibition of at least two of the three criteria. Turtles were on board the boat for an average of 57.7 min (range of min), and I subsequently released them within 10 meters of the capture site. Immediately prior to release, I obtained a second 4 ml blood sample (PRE-RELEASE sample) from 7 green turtles so that I could assess the effect of onboard restraint on blood parameters. INITIAL and PRE-RELEASE blood samples were stored on ice for minutes before centrifuging at 7,000 RPM for 10 minutes using a portable field centrifuge (Zip Spin, LW Scientific Inc., Lawrenceville, GA, USA). Plasma was stored in cryogenic tubes on dry ice and ultimately transferred to a 80ºC freezer. I analyzed blood biochemistry within 4 months of sample collection. 18

28 All procedures used for this study were approved by University of North Carolina Wilmington Institutional Animal Care and Use Committee (protocol # ) and The NOAA Fisheries Office of Protected Resources (permit # 1572). Blood Analysis Plasma concentrations of lactate dehydrogenase (LDH), creatine phosphokinase (CPK), aspartate aminotransferase (AST), Na +, K +, Cl -, P, Ca 2+, total protein, albumin, globulin, uric acid, urea nitrogen and glucose were analyzed by spectrophotometry at a veterinary diagnostic laboratory (Antech Diagnostics, Southaven, MS). I determined plasma lactate concentrations using a commercially available two-step lactate reagent kit (Pointe Scientific Inc., Canton, MI) and standard spectrophotometric techniques (Lambda 25 UV/Vis, PerkinElmer, Waltham, MA). I used lactate standards of 5 mmol L -1, 10 mmol L -1, 15 mmol L -1, and 50 mmol L -1 to generate a regression equation to describe the relationship between absorbance and lactate concentration ([Lactate] mmol L -1 = (abs )/0.0299, r 2 = ). All plasma samples were run in duplicate, and I used the mean of duplicate absorbance values to estimate plasma lactate concentrations using the standard regression. Buffer solutions and 15 mmol L -1 standard solutions were assayed simultaneously with plasma samples as a quality control measure. I analyzed corticosterone levels by radioimmunoassay (RIA) as previously described by Valverde (1996). For each sample, I extracted 25 to 250 µl of plasma with 4 ml of anhydrous diethyl ether, dried the tubes under nitrogen gas, and resuspended them with 1 ml of gel buffer (ph 7.0). I pipetted two 400-µL aliquots from the 1 ml of gel buffer, and placed the tubes at 4ºC overnight. The following day, all tubes were 19

29 incubated in a water bath for 30 minutes at 37ºC. At this point, I prepared tubes containing 400 µl of corticosterone standard solution with concentrations ranging from to 8 ng ml -1 in duplicate. Following incubation, I added 100 µl of costicosterone antibody (# B3-163, lot , purchased from Esoterix Laboratory Services, Calabasas Hills, CA) to all tubes (standards and samples), as well as 100 µl of tritiated corticosterone (~10000 cpm; PerkinElmer Life and Analytical Sciences, Inc, Boston, MA). The tubes were incubated overnight at 4ºC. The next day, I placed all tubes in an ice bath, and added 500 µl of dextran-coated charcoal to each assay tube except those used to determine total counts. All tubes were incubated for 15 minutes at 4ºC, and centrifuged at 2300 rpm at 4ºC for 15 minutes. I then poured the supernatant into scintillation vials and added 5 ml of Ecolume scintillation cocktail to each vial. The vials were counted for 60 seconds with a Wallac 1409 liquid scintillation counter. I then calculated corticosterone concentrations in ng ml -1 from the counts using the standard curve. The values were corrected by multiplying the volume extracted by the extraction efficiency and the fraction aliquoted from the reconstituted sample (40%). I calculated the extraction efficiency for each sample individually by extracting the same volume of plasma used to determine corticosterone concentration as described above ( µl), and by adding 100 µl of tritiated corticosterone (~10000 cpm) prior to the etherextraction. Extraction efficiencies ranged %. I used a loggerhead sea turtle control sample and extracted 4-5 times to evaluate intra and inter-assay variability, which was 4.3% and 12.9%, respectively. 20

30 Statistical Analyses In order to assess the effects of entanglement time and physical grade on each blood parameter for the INITIAL samples collected from green turtles, I used analysis of co-variance (ANCOVA). Only INITIAL samples (N = 12), as opposed to PRE- RELEASE samples were used for these analyses as this allowed me to assess the physiological disturbances attributable to gillnet entanglement without the potentially confounding effects of on-board restraint. Due to a low sample size (N = 4), I did not perform statistical analyses of Kemp s ridley data.. I examined the relationship between each blood parameter and predictors such as T B, T W, carapace area (as an indicator of body size), and salinity using Pearson s correlation. For predictors that were determined to be strongly (r > 0.50) and significantly (P < 0.05) correlated with a particular blood parameter, I initially included them in the ANCOVA model for that parameter. As a result, T B was used as a co-variate in the ANCOVA model for LDH, and carapace area was used as a co-variate in the ANCOVA models for Na +, Cl -, and glucose. Ultimately, carapace area contributed significantly to the ANCOVA model for glucose but none of the other co-variates contributed significantly to ANCOVA models. I used a paired t-test and applied a Bonferroni correction to compare INITIAL and PRE-RELEASE values for blood parameters in green sea turtles (N = 7). Significance was set at P < using the Bonferroni correction. I performed all analyses using Statistical Analysis Software (SAS) version 9.1 (Cary, NC, USA). 21

31 RESULTS The majority of turtles that were entangled in gillnets for less than 4 hours were classified as having physical grades of B and C (Table 1). Of the 18 turtles that I captured, 17 % of the turtles were classified as physical grade A, 33 % turtles were classified as physical grade B, 39 % of the turtles were classified as physical grade C, and 11 % of the turtles were classified as physical grade D. Table 2 presents descriptive statistics and standard deviations (SD) for blood parameters of INITIAL samples from gillnet-entangled green and Kemp s ridley turtles. Increased entanglement time and decreased physical grade accounted for an increase in plasma lactate (F 6 = 25.91, P = 0.001) (Fig. 1a), LDH (F 7 = , P = 0.009) (Fig. 1b), CPK (F 5 = 8.53, P = 0.017) (Fig. 1c), phosphorus (F 6 = 10.61, P = 0.010) (Fig. 1d), and glucose (F 7 = 8.44, P = 0.028) (Fig. 1e). I also found that entanglement time and physical grade did not account for trends in plasma albumin (F 5 = 5.01, P = 0.051), AST (F 6 = 0.25, P = 0.939), Na + (F 7 = 0.58, P = 0.750), K + (F 6 = 2.41, P = 0.177), Cl - (F 7 = 2.39, P = 0.209), Ca 2+ (F 6 = 3.67, P = 0.088), total protein (F 6 = 3.83, P = 0.081), globulin (F 6 = 2.11, P = 0.216), uric acid (F 6 = 4.83, P = 0.052), urea nitrogen (F 6 = 0.86, P = 0.577), or corticosterone (F 6 = 1.34, P = 0.383). Although the trend in increased corticosterone with increased time in net and decreased physical grade was not significant, I recorded very high corticosterone in gillnet entangled turtles (Table 2, Fig. 1f). I did not find any significant trend in any blood parameter between INITIAL and PRE-RELEASE samples from gillnet entangled green turtles. 22

32 DISCUSSION The main focus of this study was to investigate the effect of gillnet entanglement time on blood biochemistry and health status of sea turtles. I found that longer entanglement times resulted in pronounced disruptions in blood biochemistry and were associated with lower physical grades (Tables 1 and 2). In general, blood parameters that did not vary significantly with increased gillnet entanglement time fell within the range of published values for healthy, wild-caught green sea turtles (Bolten and Bjorndal, 1992; Aguirre et al., 1995; Hasbún et al., 1998), whereas parameters that were significantly impacted by gillnet entanglement time were in agreement with literature values for sea turtles exposed to stressors (Lutz and Dunbar-Cooper, 1987; Aguirre et al., 1995; Gregory et al., 1996; Hoopes et al., 2000; Jessop et al., 2002; Harms et al., 2003; Jessop et al., 2004; Jessop and Hamann, 2005). Entanglement times of as little as 30 minutes resulted in elevated plasma lactate. The highest lactate value recorded in my study (50.6 mmol L -1 ) was 2 8 times higher than maximum values for blood lactate reported in other studies of sea turtle entanglement in fishing gear ( mmol L -1 ) (Lutz and Dunbar-Cooper, 1987; Hoopes et al., 2000; Harms et al., 2003). The average blood lactate concentration for green turtles in our study (30.6 ± 3 mmol L -1, N = 12) was 9 times higher than average blood lactate concentration of rehabilitated captive green turtles not exposed to a stress protocol and just prior to release (3.4 ± 1.1 mmol L -1, N = 10) (C. Harms and J. Beasley, unpublished data). Although sea turtles have a high aerobic capacity to support sustained, long-distance swimming (Butler et al., 1984), they resort to anaerobic pathways during 23

33 intense burst activity. Intense struggling and forced submergence during entanglement likely result in a shift from aerobic to anaerobic metabolic pathways due to an imbalance between oxygen supply and demand. Increased reliance on glycolysis and lactic acid fermentation results in lactate accumulation in blood and tissues. Oxygen is required in order for lactate to be metabolized and cleared from the bloodstream. Previous studies have noted that sea turtles subjected to enforced submergence may require extended periods of time at the surface to rest, recover, and repay the oxygen debt incurred while forcibly submerged (Lutz and Bentley, 1985; Stabenau and Vietti, 2003). Extended time at the surface may leave recovering sea turtles vulnerable to other threats, such as boat strikes or shark predation. Studies investigating lactate loads and clearance rates for sea turtles captured in trawls or restrained in in-water cages have demonstrated that lactate clearance rates can vary between mmol L - 1 hr -1 for blood lactate concentrations of 5 14 mmol L -1 (Lutz and Bentley, 1985; Stabenau and Vietti, 2003). If sea turtles in my study cleared lactate at the fastest rates documented in the literature, it would take 4 15 hours to remove accumulated lactate from the bloodstream. It is likely that full clearance of the high lactate loads I observed would actually require more time, as clearance rates tend to decline with declining blood lactate concentrations. Lutz and Dunbar-Cooper (1987) calculated that clearance of only 3 4 mmol L -1 blood lactate could take as long as 20 hours due to the decline in clearance rates at low blood lactate concentrations. Although low to moderate levels of circulating lactate (< 5 mmol L -1 hr -1 ) may not impose a great physiological challenge to sea turtles, this situation may be problematic for turtles that experience repeat captures in fishing gear. The limited home range of 24

34 juvenile green and Kemp s ridley sea turtles in nearshore, coastal waters (Mendonca, 1983; Brill et al., 1995; Avens et al., 2003; Avens and Lohmann, 2004; Makowski et al., 2006) predisposes them to multiple encounters with fishing gear set within their home range. Additional enforced submergence events greatly increase the chance of in-net or post-release mortality, particularly if the turtle has not fully recovered from the physiological disruptions of the first entanglement. Stabenau and Vietti (2003) noted severe metabolic disturbances in juvenile loggerhead turtles forcibly submerged in trawls multiple times, and found that a surface recovery interval of 42 minutes following the first submergence was inadequate for blood lactate clearance. Although the ANCOVA model to assess the effects of entanglement time and physical grade on blood corticosterone levels was not statistically significant, I feel this blood variable warrants further comment given the widespread effects that corticosterone may have on the physiology of sea turtles (Gregory et al., 1996; Jessop et al., 2002; Jessop et al., 2004; Jessop and Hamann, 2005). Corticosterone is a glucocorticoid that is released into the blood by the adrenal glands as a response to various stressors. The release of corticosterone triggers behavioral and physiological adjustments to promote survival, while curtailing other non-essential processes preferentially partition energy stores towards survival (Jessop, 2001; Jessop et al., 2002). Continued stress associated with injuries sustained during entanglement or behavioral alterations may delay clearance of corticosterone and impact post-release survival, but I was not able to address this possibility in our study. I recorded a maximum corticosterone concentration of 51.8 ng ml -1 for INITIAL samples from green turtles (Table 2), and levels I observed were as high as or higher than 25

35 values reported in the literature for stressed turtles (2 25 ng ml -1 ) (Gregory et al., 1996; Jessop et al., 2002; Jessop et al., 2004; Jessop and Hamann, 2005). Previous studies of induction of the stress response in sea turtles demonstrated that maximum concentrations of blood corticosterone were reached within minutes of stress exposure (Jessop, 2001; Jessop et al., 2004; Jessop and Hamann, 2005). In my study, corticosterone appeared to level off at maximum concentrations within minutes of capture (Fig. 1f). This may explain why the ANCOVA model, which covered entanglement times that ranged from minutes, failed to detect a significant effect of entanglement time on blood corticosterone concentrations. The trend towards increased corticosterone with increased time in net was accompanied by an increase in blood glucose concentration, a classic signature of induction of a systemic stress response. Elevated blood glucose has been documented in previous studies of sea turtles exposed to capture and handling stress (Aguirre et al., 1995; Hoopes et al., 2000). Blood glucose levels of juvenile green turtles in my study ranged from mg dl -1 ( X ± SD = ± 34.9 mg dl -1 ), which is consistent with levels noted by Aguirre et al. (1995) for green turtles exposed to a capture stress protocol ( mg dl -1 ). The physiological stress response induced by gillnet entanglement may be exacerbated by injuries incurred while in the net. The significant increases in plasma LDH and CPK seen in gillnet-entangled turtles are indicative of muscle or tissue damage, as these enzymes may leak from ruptured cells into the blood stream (Aguirre et al., 1995; Killen et al., 2003; Morrissey et al., 2005). Several of the turtles in this study incurred soft tissue damage from the nets, as documented during the physical 26

36 examination, and possible cardiac muscle damage due to struggling and overexertion. I found average LDH in juvenile green turtles of U L -1, which was approximately 7 times the amount noted by Aguirre et al. (1995) in juvenile green turtles exposed to acute capture and handling stress for up to 4 hours. It is important to note that the significant increase in CPK and LDH with increased entanglement time and decreased physical grade is largely driven by values of green turtles with physical grade D (N = 2). These turtles had very high plasma CPK and LDH concentrations. When I removed these two turtles from the analysis, the ANCOVA model testing the effects of entanglement time and physical grade on LDH concentration was still statistically significant (P = 0.039), but the model for CPK concentrations was not significant (P = 0.915). The high concentrations of phosphorus in the blood that I observed may also indicate tissue damage, as inorganic phosphates leak out of damaged cells into the bloodstream (Bishop et al., 2004). Increased blood phosphorus may also indicate decreased kidney function and filtration. Previous studies on rabbits (Nastuk, 1947) and rats (Goranson et al., 1948) have shown an increase in blood inorganic phosphate levels associated with shock, potentially the result of an increase in the rate of high energy phosphate bond hydrolysis in the face of increased energy demands (McShan et al., 1945; Nastuk, 1947). Although my study focused on the effects of entanglement time on health status of sea turtles, other factors may also contribute to physiological stress during gillnet entanglement. For example, turtles entangled at the bottom of the gillnet (depths > 0.5 m) or turtles that have net tightly wrapped around their neck or flippers may be prevented 27

37 from reaching the surface to breathe and experience severe respiratory and metabolic disruptions after only a short entanglement time. Green turtle Cm 9 was entangled in the net for only 70 minutes, but at a depth greater than 0.5 m, which made it difficult to reach the surface. This turtle had a physical grade of C, very high lactate levels (38.2 mmol L -1 ), and the highest corticosterone levels observed in our study (51.8 ng ml -1 ). In contrast, green turtle Cm 12 was lightly entangled for 212 minutes at the top of the net (< 0.5 m). Although this turtle had one of the longest entanglement times in the study, it had a physical grade of B and low lactate (15.0 mmol L -1 ) and corticosterone (7.0 ng ml -1 ) levels compared to other turtles entangled for similar amounts of time. Turtles entangled at the top of the net, or only lightly entangled, may be able to endure long entanglement times with only mild to moderate disruptions in blood biochemistry due to relatively unimpeded access to air. Management Implications Currently, the North Carolina Division of Marine Fisheries enforces a mandatory gillnet attendance regulation on gillnet fisheries in the lower Cape Fear River, North Carolina during the summer months in an effort to minimize sea turtle entanglements and mortalities. This has essentially closed the fishery during this time, as fishermen are unlikely to remain with their nets during the typical soak period of 12 hours or more. I was hopeful that our investigation of the effects of entanglement time on the physiology of sea turtles would allow us to determine a maximum unattended gillnet soak time that could be implemented that would minimize the impacts on sea turtles that are captured. Due to the fact that there are variables other than entanglement time that contribute to the 28

38 severity of the impact of entanglement, it is very difficult to propose a safe soak time that would reliably minimize detrimental effects of entanglement on captured turtles. However, the health status of turtles at the time of removal from the net can be easily assessed using the on-board protocol described in this paper. Physical examinations to assess behavior, injuries and reflexes provided valuable insight into the physiological impacts of entanglement on sea turtles. Data obtained through a simple physical examination may help determine whether to release a turtle or take it to a rehabilitation facility following a gillnet encounter, thereby minimizing the potential for post-release mortality. 29

39 Table 1: Descriptive information for Kemp s ridley (Lk) (N=4) and green (Cm) (N=14) turtles captured in gillnets in the lower Cape Fear River, North Carolina from May - October Turtle ID Capture Date Capture Time Capture Location Salinity (ppt) Tw (ºC) Tb (ºC) SCL (cm) SCW (cm) Carapace Area (cm 2 ) Entanglement Time (min) Lk 1 6/6/ : N W A Lk 2 6/30/ : N W B Lk 3 7/3/ : N W A Lk 4 8/31/ : N W B X SD Cm 1 6/7/ : N W C Cm 2 6/8/ : N W D Cm 3 6/14/ : N W C Cm 4 7/4/ : N W C Cm 5 8/20/ : N W D Cm N 9/2/ : W B Cm 7 9/2/ : N C Physical Grade 30

40 W Cm 8 9/8/ : N W B Cm 9 9/12/ : N W C Cm 10 9/19/ : N W B Cm 11 9/26/ : N W A Cm 12 10/19/ N 13: W B Cm 13 10/19/ N 14: W C Cm 10/19/ N 18: W C X SD

41 Table 2: INITIAL blood parameters of green and Kemp s ridley sea turtles captured in the lower Cape Fear River, North Carolina, May - October Green Turtles Kemp s Ridley Turtles Initial (N=12) Initial (N=4) Mean Normal Blood Values Green Kemp s Ridley Blood Parameter Range X SD Range X SD Turtles Turtles LDH (U L -1 ) a, b g CPK (U L -1 ) c AST (U L -1 ) a, b g Na + (meq L -1 ) a, b e, g K + (meq L -1 ) a, 5.0 b 6.3 e, 3.6 g Cl - (meq L -1 ) a, b e, g P (mg dl -1 ) a, 7.9 b 6.8 g Ca 2+ (mg dl -1 ) a, 8.4 b 7.4 g Total Protein (g dl -1 ) a, 4.3 b 3.1 g Albumin (g dl -1 ) a, 0.9 b 1.3 g Globulin (g dl -1 ) a, 3.0 b 1.8 g Uric Acid (mg dl -1 ) a, 0.8 b Urea Nitrogen (mg dl -1 ) a, 1.0 b 73.7 g Glucose (mg dl -1 ) a, 86.6 b g Lactate (mmol L -1 ) c, 1.1 d 0.7 e CORT (ng ml -1 ) b 3.1 f 32

42 a Bolten and Bjorndal 1992, b Aguirre et al. 1995, c Butler et al. 1984, d Berkson 1966, e Stabenau et al. 1991, f Gregory and Schmid 2001, g Carminati et al

43 Table 3: Results of paired t-test analysis for INITIAL and PRE-RELEASE blood parameters of green sea turtles captured in the lower Cape Fear River, North Carolina, May - October 2007 (N = 6 for all parameters except for lactate and CORT, where N = 7). Time between INITIAL and PRE-RELEASE samples ranged from minutes. Blood Parameter Initial Pre-Release df t P X SD X SD LDH (U L -1 ) CPK (U L -1 ) AST (U L -1 ) Na + (meq L -1 ) K + (meq L -1 ) Cl - (meq L -1 ) P (mg dl -1 ) Ca 2+ (mg dl -1 ) Total Protein (g dl -1 ) Albumin (g dl -1 ) Globulin (g dl -1 ) Uric Acid (mg dl -1 ) Urea Nitrogen (mg dl -1 ) Glucose (mg dl -1 ) Lactate (mmol L -1 ) CORT (ng ml -1 )

44 a) [Lactate] mmol/l Entanglement Time (min) b) 6500 [LDH] U/L Entanglement Time (min) c) [CPK] U/L Entanglement Time (min) 35

45 d) [P] mg/dl e) Entanglement Time (min) 250 [Glucose] mg/dl f) Entanglement Time (min) [CORT] ng/ml Entanglement Time (min) 36

46 Figure 1: a) Blood lactate, b) LDH, c) CPK, d) phosphorus, e) glucose, and f) corticosterone of juvenile green sea turtles (N = 12) entangled in shallow-set gillnets for 20 to 240 minutes. Physical grade for each individual turtle is indicated by symbols: physical grade A ( ), physical grade B ( ), physical grade C ( ), physical grade D (Δ). 37

47 CHAPTER 2. MOVEMENTS AND POST-RELEASE MORTALITY OF JUVENILE SEA TURTLES RELEASED FROM GILLNETS IN THE LOWER CAPE FEAR RIVER, NORTH CAROLINA Submitted to Endangered Species Research (3/26/09) ABSTRACT The coastal waters of North Carolina, USA serve as an important seasonal foraging habitat for several species of endangered sea turtles, including the green sea turtle (Chelonia mydas) and Kemp s ridley sea turtle (Lepidochelys kempii). Sea turtle habitat utilization in the Cape Fear River region has not been well-documented compared with other coastal regions, but increased numbers of sea turtle interactions with fishing gear in this region suggest that sea turtles may be present in high abundance during the summer months. An understanding of sea turtle movement patterns is important for assessing the potential for seasonal overlap with fishing operations and developing appropriate mitigations strategies for reducing interactions. I used satellite telemetry to 1) monitor movements of juvenile green and Kemp s ridley sea turtles released from gillnets in the lower Cape Fear River and 2) assess the potential for using satellite telemetry to document post-release mortality of sea turtle released from gillnets in an inshore environment. Tracking durations for the fourteen sea turtles on which I deployed satellite transmitters ranged from 6 45 days. Twelve out of fourteen turtles released from gillnets stayed in the lower Cape Fear River throughout the tracking duration, and 50% of locations received via satellite fell within a 33 km 2 area that included the capture site. The region most utilized by turtles consisted of high salinity waters (35 39 ppt) of 1 5 meters depth. I observed an abundance of algae, fish and

48 invertebrates that could be food items for greens and Kemp s ridleys in the high-use region. I documented one confirmed post-release mortality and three suspected postrelease mortalities during the course of this study. INTRODUCTION Sea turtles face many threats to their survival, including loss of nesting and foraging habitat, egg and hatchling predation, pollution, and other anthropogenic factors such as boat strikes and encounters with recreational and commercial fishing gear (Lutcavage et al., 1997). Efforts to protect sea turtles on nesting beaches are wellestablished, but in-water threats remain a topic of great concern for sea turtle conservationists and policy-makers. Bycatch of sea turtles in commercial fishing gear has been identified as a significant source of mortality contributing to population declines (Magnuson et al., 1990; Lewison et al., 2004). Knowledge of sea turtle habitat and the potential for overlap with fisheries, as well as an understanding of the impacts of incidental entanglement on the behavior and survivability of sea turtles, are high priorities for management. Mitigation of fisheries interactions with juvenile sea turtles is of particular importance, as protection of this age class is thought to be critical to recovery efforts (Crouse et al., 1987; Read et al., 2004). North Carolina coastal waters serve as an important foraging ground for juvenile sea turtles in the summer months (Epperly et al., 1995a; Epperly et al., 1995b). The most common species found in North Carolina are loggerheads (Caretta caretta), greens (Chelonia mydas) and Kemp s ridleys (Lepidochelys kempii) (Epperly et al., 1995a). 39

49 While resident in coastal foraging habitats, sea turtle movements are typically limited to a home range of approximately 5 km 2, which is determined by availability of food sources and nightly resting sites (Brill et al., 1995; Seminoff et al., 2002; Makowski et al., 2005). Sea turtles display strong homing behaviors for particular areas and pinpoint specific daytime foraging and nighttime sleeping sites, which they return to within the same season and even between years (Lohmann et al., 1997; Avens et al., 2003). Sea turtles also display predictable diel movement patterns during summer months. Mendonça (1983) showed that juvenile green turtles at Mosquito Lagoon, Florida routinely shuttle between shallow foraging grounds during the morning and evening and deeper channels in mid-afternoon when surface Tw increases. Several studies have shown that green turtles return to the same resting sites night after night (Mendonça, 1983; Brill et al., 1995; Seminoff et al., 2002; Southwood et al., 2003). Entanglement of sea turtles in the gillnet fisheries of coastal North Carolina has become a critical concern for fisheries managers, as there have been mass stranding events that coincide with the peak season for this fishery (Federal Register, 2004; Gearhart, 2001; NCDMF, 2005; Price, 2005). The deep-water gillnet fishery in Pamlico Sound, NC was shut down in 2002 due to interactions with sea turtles (Federal Register, 2002) and fishing effort is now restricted to shallow waters (< 3 m) in this region, as this reduces the likelihood that captured turtles will drown in nets. Although the majority of sea turtles entangled in shallow-set gillnets are released alive (Gearhart, 2001), the ultimate fate of these turtles is unknown. Severe physiological disruptions and injuries incurred while entangled in gillnets could result in undocumented post-release mortality (Lutcavage and Lutz, 1991; Stabenau and Vietti, 2003; Harms et al., 2003). 40

50 Juvenile green and Kemp s ridley turtles are also captured in gillnets set in the lower Cape Fear River, NC (Thorpe and Beresoff, 2005). Due to an increasing number of sea turtle interactions with gillnet fisheries, North Carolina Division of Marine Fisheries (NC DMF) has placed restrictions on gillnetting in the lower Cape Fear River. The NC DMF requires full-time net attendance during the summer fishing season so that fishermen may immediately remove any turtles that become entangled (Pate, DMF, 2006). This restriction effectively closes the summer gillnet fishery in this region because fisherman are unwilling to stay with their nets throughout the 12 hours of a typical set. The implementation of strict fisheries regulations by the NC DMF reflects the concern that there are significant numbers of sea turtles in the lower Cape Fear River region during the summer. However, very little is known of the movement patterns of sea turtles in this area, with most of the available data coming from Marine Patrol observations, anecdotal reports from fishermen and a previous study by Thorpe and Beresoff (2005). Information on sea turtle movements and habitat utilization is vital for developing appropriate bycatch management strategies for the lower Cape Fear River. I used satellite telemetry to monitor the movements of juvenile green and Kemp s ridley sea turtles released from gillnets set in the lower Cape Fear River. My primary goals were to 1) document sea turtle movements and habitat utilization in this shallow, inshore environment, and 2) to investigate the potential for using satellite telemetry to document post-release mortality of sea turtles captured in a coastal gillnet fishery. The use of satellite telemetry to refine post-release mortality estimates for sea turtles released from pelagic longline fishing gear has been met with varying degrees of success (Chaloupka et al., 2004; Swimmer et al., 2006; Sasso and Epperly, 2007). Determining 41

51 the post-release fate (survival or mortality) of sea turtles using satellite telemetry is complicated by the fact that cessation of a satellite signal may be attributable to factors other than mortality, such as tag failure or tag loss due to shedding (Chaloupka et al., 2004; Hays et al., 2004). I reasoned that use of satellite telemetry to determine postrelease mortality of sea turtles would be more feasible in an inshore environment, as opposed to open ocean, because I would have the opportunity to locate and retrieve carcasses that stranded on land to confirm mortalities. To optimize my chances of locating stranded turtles, I deployed VHF radio beacons along with satellite transmitters. This study was conducted as part of a larger research initiative to assess the physiological and behavioral impacts for sea turtles entangled in shallow-set gillnets (Snoddy et al., in press; Southwood et al., 2008). METHODS Field Procedures Sea turtles were captured in mesh gillnets set at depths of 1-2 m in the lower Cape Fear River, North Carolina during daylight hours (06:00 16:00) from May through October of 2007 (Fig. 3). Gillnets remained in water for a maximum of 6 hours and were attended at all times, as per NC DMF regulations (Proclamation M ), so that I could document time of capture. A total of 18 sea turtles (14 green turtles and 4 Kemp s ridley turtles) were captured during the course of this study. Captured turtles remained in the net for up to 240 min, and were closely monitored for signs of distress while in the net. If a turtle remained submerged for greater than 20 minutes or appeared to be in 42

52 danger of drowning due to airway or swimming restriction, it was immediately removed from the net. Turtles that were tracked post-release (N = 14) were entangled for minutes ( X ± SD = 85.5 ± 67.7 minutes). Environmental variables (water temperature (T W ), air temperature (T A ), salinity) were recorded at the capture site, and GPS locations for capture sites were documented. I also identified algae and invertebrates found in the capture area that could be potential food sources for juvenile greens and Kemp's ridleys. Upon removal from nets, turtles were brought on board our boat and restrained in a 16 cm 43 cm padded plastic bin. Turtles were shaded from direct sunlight and periodically sprayed with seawater. I used calipers to measure the straight carapace length notch to notch (SCLnn) and straight carapace width (SCW), and marked turtles for future identification by inserting passive induced transponder (PIT) tags above the left front flipper. I used a two-part fast-setting marine epoxy (PowerFast, Powers Fasteners, Inc., New Rochelle, NY) to attach satellite transmitters (SPOT 5, Wildlife Computers, Redmond, WA, USA) (7.9 cm length x 4.9 cm width x 1.8 cm height, 90 g) and VHF radio beacons (SI-2F, Holohil Systems, Inc., Carp, Ontario, Canada) (3.5 cm length x 1.0 cm dia, 11 g) to 14 of the 18 turtles I captured (Fig. 2). The other four turtles were too small for transmitter deployment based on my size criteria. I did not deploy transmitters on turtles for which the total mass of transmitter and epoxy was greater than 5% of the turtle s mass in air, as calculated from a length-weight power regression (NOAA Beaufort Laboratory, unpublished data). Estimated masses of turtles captured ranged from 1-10 kg (6.3 ± 2.4 kg). Prior to transmitter attachment, the vertebral scutes of the carapace were cleared of barnacles, cleaned with acetone to remove biofouling, lightly sanded with sand paper, and given a final acetone rinse. Epoxy was used to secure the 43

53 VHF radio beacon to the third or fourth vertebral scute with the antenna facing toward the head of the turtle and laying flat on the carapace surface. The satellite transmitter was secured with epoxy to the first and second vertebral scutes of the carapace. The epoxy base for the transmitter was molded such that drag effects would be reduced. While epoxy was setting, I examined turtles for net-inflicted external injuries, tested reflex responses, and took a blood sample to assess the physiological impacts of entanglement (Snoddy et al., in press). Turtles were on board the boat for an average of 58 minutes (range 10 min-110 min) and were then released within 10 meters of the capture site. All procedures used for this study were approved by University of North Carolina Institutional Animal Care and Use Committee (protocol # ) and The NOAA Fisheries Office of Protected Resources (permit # 1572). Analysis of Location Data The satellite transmitters were programmed for a 24 hour duty cycle and transmitted location data to Service Argos network satellites when turtles surfaced to breathe. Transmitter positions were assigned to one of six location classes (LC 3, 2, 1, 0, A and B) by Service Argos based on the number of transmissions received, the number of satellites receiving transmissions, and the angle and speed of satellites relative to the transmitter at the time of transmissions. For location classes 3, 2, 1, and 0, the location accuracies are less than 150 m, 350 m, 1000 m, and greater than 1000 m, respectively. For location classes A and B, no location accuracy is assigned. The percentage of transmissions of each location class for all turtles combined was calculated (Fig. 4). 44

54 Location data were downloaded and analyzed using the Satellite Tracking and Analysis Tool (STAT) program available at (Coyne and Godley, 2005). Ideally, only high quality location class data (LC 3, 2, and 1) would be used in the tracking analysis, however the majority of positions I received were of low quality location class (LC A and B). I included all location data in our initial analysis and applied a multi-step filtering procedure to exclude implausible locations. Based on diagnostic data provided by Service Argos, I excluded any location that had a satellite pass time of less than 240 seconds. The remaining locations were then plotted sequentially on a map, and filtered based on speed and distance thresholds established in previous studies of sea turtle movement patterns in coastal environments (Read et al., 2004). Locations that were separated by distances that could not be covered at a swim speed of less than 5 km/hour were excluded from analysis, as were locations that would have required turtles to pass implausibly over land barriers. Because transmissions from land may indicate a stranding event, particularly high quality location class transmissions along the shoreline, all land-based transmissions were carefully analyzed. If low quality land-based transmissions were interspersed over time with transmissions from water, these points were excluded from analysis. Filtered data were mapped in ArcGIS (version 9.2) and Hawth's Tools extension was used to create 25 and 50% volume contours of filtered location data for all turtles combined. Assessment of Mortality Satellite transmissions received during the 30 days following release from gillnet were analyzed for patterns indicative of mortality based on 1) documented behavioral 45

55 patterns of green and Kemp s ridley turtles in nearshore environments, 2) behaviors associated with compromised health, and 3) knowledge of the process of decay and onshore stranding of sea turtle carcasses. I predicted that a mortality event would be reflected by satellite transmission patterns that deviated from previously documented patterns and were consistent with the process of decay and putrefaction (criteria described below). The 30 day monitoring period was chosen because turtles are exposed to numerous threats in their marine environment, and the more time that passes the more difficult it becomes to attribute mortality to the gillnet interaction. I reasoned that physiological and behavioral consequences of gillnet entanglement and vulnerability to other threats would be greatest in the first few weeks following entanglement. Previous satellite telemetry studies of sea turtles in coastal environments have demonstrated that short surfacing intervals (< 1 min) and low profile surfacing patterns typically result in receipt of low quality location class data (LC A or B) (Godley et al., 2003; Read et al., 2008). Sea turtles that are injured, fatigued, or have experienced large disruptions in blood biochemistry due to enforced submergence may require extensive amounts of time at the surface to recover (Lutz and Bentley, 1985; Stabenau and Vietti, 2003), which would be reflected by receipt of high quality locations (LC 3, 2, 1). I interpreted prolonged periods of numerous, high quality location class transmissions that occurred in the hours to days immediately following release as representative of a surface recovery period. I compared the percentage of high quality location class transmissions (LC 3, 2, 1) received during the first 24 hours following release to the percentage of high quality location transmissions received in the subsequent 72 hours for each turtle using an ANOVA and Tukey s test. Significance was set at P <

56 I predicted that mortality events would be reflected by alterations in the quality and quantity of location data transmitted via satellite. Specifically, I predicted that satellite transmissions would cease for several days following a mortality event as the carcass sank below the surface, but that frequent, high quality location class transmissions (LC 3, 2, 1) would resume for a brief period when putrefaction and buildup of gases caused the carcass to temporarily float back to the surface (Epperly et al., 1996; Committee on Sea Turtle Conservation, National Research Council, 1990). Increases in the quality and frequency of satellite transmissions along the shoreline were interpreted as a possible shore stranding event. In such cases, a VHF radio receiver (TR- 5, Telonics, 932 E. Impala Avenue Mesa, AZ, USA) and directional H antenna (RA-2AK, Telonics, 932 E. Impala Avenue Mesa, AZ, 85204) were used to search for the VHF radio beacon signal so that I could locate the carcass and verify mortality. Turtles that did not display transmission patterns indicative of a mortality event within 30 days of release were considered survivors. Turtles that displayed satellite transmission patterns indicative of mortality but for which I did not locate a carcass were categorized as suspected mortalities. Turtles that displayed satellite transmission patterns indicative of mortality and for which I located a carcass were categorized as confirmed mortalities. Carcasses that were located were examined for indications of boat strike, predation, and gut impaction. I compared the percentage of high quality locations (LC 3, 2, 1) for the entire track duration of suspected mortalities and confirmed mortalities (N = 4) to those of survivors (N = 10) using a student s t-test. 47

57 RESULTS Movements and Habitat Utilization Sea turtles were tracked for 17.0 ± 8.9 ( X ± SD) days following release from gillnets (range 6-42 days). Filtered location data indicate that the majority of sea turtles remained in the lower Cape Fear River for the tracking duration. A small percentage (16 %) of low quality locations that passed the filter criteria placed turtles in the ocean around the river mouth or off the eastern shore of the barrier island complex of southeastern North Carolina. Approximately 95% of filtered high quality location data (LC 3, 2, 1) for all but two turtles (Cm 3 and Cm 13) were limited to within approximately 2-3 km radius of the turtle s capture site. Fifty percent of all filtered locations were encompassed by a 33 km 2 area and 25% of all locations were encompassed by a 12 km 2 area, which included 10 of the 14 capture sites (Fig. 3). The 50% volume contour was bordered on the north by N, W, on the south by N, W, on the east by N, W, and on the west by N, W. Two turtles, Cm 3 and Cm 13, migrated out of the Cape Fear River following release from gillnets. Turtle Cm 3, captured on 06/14/07, remained in the lower Cape Fear River for three days following release and then exited the river and moved north along the North Carolina coastline for 10 days. The last transmission from turtle Cm 3 was received on 06/27/07 from the lower White Oak River near Swansboro, North Carolina (Fig. 5). Turtle Cm 13, captured on 10/19/07, exited the Cape Fear region 20 days after release and traveled south along the coasts of North Carolina and South 48

58 Carolina for 22 days before transmissions ceased (Fig. 6). The last transmission was received from east of the mouth of St. Helena Sound, South Carolina on 11/24/07. Post-Release Mortality Juvenile green and Kemp s ridley turtles released from gillnets were classified as confirmed mortalities, suspected mortalities, or survivors based on patterns observed in satellite transmissions (Fig. 19a 19n, and Table 4). The one turtle for which I directly documented mortality by recovering the carcass (Kemp s ridley Lk 2, captured 06/30/07) displayed a pattern of satellite transmissions that met my criteria for mortality. Between 06/30/07 and 07/04/07 I received 14 transmissions from this turtle. Following a LC B transmission on 07/04/07, there was a period of several days during which no signals were received. Transmissions resumed at 23:09 on 07/06/07, and all further transmissions were of high location class quality (Fig. 7, Fig. 19a). The rising tide likely stranded the carcass in the marsh, with high tide at 01:18 on 07/07/07. The carcass was located within 1 km of gillnet capture site on 07/07/07 by a wildlife enforcement officer during a mid-morning patrol of Oak Island. When the carcass was discovered, the tide was low but rising. The officer transported the satellite tag approximately 2 km to Southport, NC and contacted me that afternoon. I was receiving transmissions throughout this transport period and had initiated my search for the carcass. Investigation of the carcass yielded no evidence of boat strike, predation, or gut impaction, and this turtle was classified as a confirmed mortality. Three other turtles, one Kemp s ridley (Lk 4) and two green turtles (Cm 2 and Cm 3), displayed transmission patterns indicative of a mortality event. Carcasses were not 49

59 located for these turtles, so they were classified as suspected mortalities. Turtle Lk 4 was released after a 30 minute gillnet entanglement on 08/31/07. Several high quality location class data points were received from this turtle in the initial 2 days post-release, a pattern suggestive of a lengthy surface recovery period (Fig. 19b). Signals received over the course of the next several days revealed that the turtle moved 24 km up river from the capture site (Fig. 8). Turtle Lk 4 was the only turtle that ventured this far north into the river. High quality location signals were reported on the low to rising tide in the river north of Snow s Cut on 09/06/07, with more high quality signals received the following day (09/07/07). I checked repeatedly for the VHF radio beacon for this turtle over the course of these two days, but did not detect any signals. Satellite transmissions for this turtle ceased on 09/09/07, with the last location reported on the river side of Snow s Cut. I did not locate this turtle s carcass, but the up-river movements and increase in high quality transmissions towards the end of the tracking period led me to believe that a mortality may have occurred. Turtle Cm 2 was released after a 218 minute gillnet entanglement on 06/08/07. Prior to release, this turtle had demonstrated weakened reflex responses and low activity levels on-board the boat. Upon release from the boat, the turtle sank slowly beneath the water surface with no active flipper strokes. High winds and choppy seas prevented me from visually relocating and recapturing this turtle, however I picked up her VHF radio beacon within minutes of release. I received numerous high quality location class data points from this turtle during the 8 hours following release, which indicates that she was at the surface for an extended period of time. I continued to receive daily low quality transmissions from this turtle until 06/19/07. After this date, I received high quality 50

60 location class data intermittently for the next several months. Transmissions received on 06/30/07 (LC 3), 08/09/07 (LC 2), 08/18/07 (LC 3), and 10/26/07 (LC 1) were clustered along a partially submerged rock wall within 500 m distance of the site where I had captured the turtle. Although this area was checked frequently, I was unable to detect the VHF radio signal for this turtle or locate a carcass or shed transmitter. Intermittent transmissions likely reflect the exposure of the transmitter, either detached or still attached to a carcass, at low tide. The poor condition of this turtle, behavior of turtle at release, and pattern of satellite transmissions led me to categorize this turtle as a suspected mortality. Transmission patterns for turtle Cm 3 also led me to believe that this turtle had died post-release. Transmissions from this turtle ceased after a period of approximately 9 days (06/14/07 06/22/07) spent traveling northwards along the coast of North Carolina from her capture site in the lower Cape Fear to just off the coast of Emerald Isle close to Bogue Inlet. Transmissions resumed 4 days later on 06/26/07, and several high quality location class transmissions were received from within the lower White Oak River adjacent to Swansboro, NC. The pattern of signal disappearance and reappearance close to the shoreline several days later suggests that this turtle died and stranded temporarily along the shoreline due to tidal flow (Fig. 19d). I received the strongest signals on the rising tide, which may indicate that the carcass was washed ashore temporarily. I was unable to recover a carcass before transmissions permanently ceased. The remaining ten turtles on which I deployed satellite and VHF radio transmitters did not display transmission patterns indicative of a mortality event as defined by my criteria, and were thus classified as survivors (Fig. 19e-n). Satellite 51

61 transmissions received from survivors were predominantly of low quality location class, sometimes with intermittent high quality signals received throughout the monitoring period (Figs. 6a, 6b, 10 18). There was a significantly larger percentage of high quality transmissions for the entire tracking period for the confirmed and suspected mortalities (N = 4) compared to the survivors (N = 10) (P = 0.017). For several turtles classified as survivors (Cm 1, Cm 4, Cm 7, Cm 8), the 24 hour period immediately following release was characterized by receipt of several high quality location class transmissions. This pattern suggests that turtles were spending extended periods of time at the surface, potentially recovering from physiological disturbances incurred while entangled in gillnets (Snoddy et al., in press). The results from the ANOVA for all tagged turtles (N = 14) indicate that there was a significantly higher percentage of high quality transmissions (LC 1, 2, 3) within the first 24 hours following release compared with the subsequent 72 hours (P = 0.014). DISCUSSION One goal of this study was to investigate the potential for using satellite telemetry to document post-release mortality of sea turtles released from gillnets. Previous studies have attempted to use satellite telemetry to determine post-release mortality of sea turtles released from longline pelagic fishing gear. Swimmer et al. (2006) were only able to document one clear example of post-release mortality for 15 longline-captured turtles based on dive data. Sasso and Epperly (2007) inferred 3 mortalities out of 39 longlinecaptured turtles based on dive data, with several turtles of unknown fate due to problems 52

62 of tag retention, which I discuss later. For pelagic studies, confirmation of mortality is hindered by the difficulties of recovering a carcass in the open ocean. In this study, I was able to recover the carcass of the one confirmed mortality, Lk 2, which permitted me to verify the satellite transmission patterns indicative of mortality in a nearshore environment. The pattern observed for Lk 2 in the 5 days following release (Fig. 19a) reflects the sinking of the animal upon death (cessation of signals), subsequent resurfacing as gases accumulate in the animal due to putrefaction (reappearance of signals) and, in this case, an onshore stranding event with the rising tide. The carcass of Lk 2 was recovered with the satellite transmitter still attached to the carapace within 8-10 hours of stranding. Turtles that were suspected mortalities were classified as such because their satellite transmission patterns were similar to those of the confirmed mortality, with gaps in transmissions and/or intermittent, repetitive, high quality transmissions from shoreline locations. Tides play an important role in the ability to recover a carcass, and movement of carcasses on and off of the shoreline with the rising and falling tides may have impeded my ability to locate suspected mortalities. I cannot entirely discount the possibility that transmitters for the three suspected mortalities were shed early and washed ashore, given that I did not recover carcasses of these animals. There were, however, distinct differences in transmission patterns for suspected mortalities and survivors with short duration tracks, namely that transmissions from survivors were of consistently low quality location class and ceased abruptly. I documented one confirmed mortality and classified three turtles as suspected mortalities in this study. Based on these data, I estimate that that post-release mortality 53

63 of sea turtles released from shallow-set gillnets lies somewhere between 7.1 and 28.6%. It is important to acknowledge that these figures are for soak times of 4 hours or less, and nets are typically left to soak overnight in the coastal North Carolina gillnet fishery. Blood samples taken from green sea turtles entangled in gillnets show significant positive relationships between entanglement time and blood biochemical parameters indicative of restraint stress and hypoxia (i.e., lactate and glucose) (Snoddy et al., in press). Given the impact of entanglement time on physiological status of green sea turtles, longer entanglement times would be expected to result in an increase in recovery time and, potentially, post-release mortality rates. My assessment of mortality for turtles released from gillnets was hampered by short track durations. I had planned to monitor satellite transmissions for signs of mortality for 30 days following release from gillnets, but the maximum track duration was 42 days and 12 track durations were 23 days or less. For turtles classified as survivors, the satellite transmissions up until the time when transmissions ceased did not show a pattern indicative of mortality as defined by my criteria. Therefore, I concluded that the short track durations were due to premature shedding of the transmitters rather than mortality. Previous satellite telemetry studies conducted with juvenile green turtles in inshore waters of North Carolina have documented difficulties with transmitter retention (Read et al., 2008). Signals received from transmitters deployed on juvenile green turtles in Core and Pamlico Sounds, NC during the summer were typically of low quality location class (LC A and LC B) and track durations ranged from days, with an average track duration of 67.7 days (Read et al., 2008). Rapid growth rates of juvenile 54

64 green turtles that utilize coastal waters of North Carolina as a seasonal foraging habitat during the warm summer months may contribute to short transmitter retention times (Shaver, 1994). The epoxy bond with the carapacial scutes may become weakened as the scutes increase in diameter. Average transmitter retention time in my study was only 17.1 days, even shorter than retention times documented for juvenile green turtles in Core and Pamlico Sounds. This may be due to differences in habitat. Green turtles typically utilize seagrass habitats with minimal rough substrate in Core and Pamlico Sounds, whereas 10 of the turtles in my study were captured along a partially submerged rock wall where they may have been foraging on algae or invertebrates. Abrasion against the rocky substrate at this foraging site may have contributed to pre-mature shedding of transmitters. Transmitters deployed on green turtles in Core and Pamlico Sounds (Read et al., 2008) and in the lower Cape Fear River were retained for a longer duration when deployed late in the season (October November), just prior to fall migration. In my study, turtles Cm 12 and Cm 13 were tagged in mid-october and their track durations (25 and 42 days, respectively) were approximately 2 times longer than the average track duration for all turtles. The majority of location data obtained from sea turtles behaving normally in inshore areas are typically of low quality, due to surfacing behavior in this habitat (Godley et al., 2003; McClellan, pers. comm.). Turtles that are physiologically stressed from enforced submergence or intense struggling may spend more time recovering at the surface (Lutz and Bentley, 1985; Stabenau and Vietti, 2003), which would result in more frequent and high quality satellite transmissions. I documented a significantly higher percentage of high quality transmissions in the first 24 hours post-release compared with 55

65 the subsequent 72 hours, which supports the idea that turtles released from gillnets require extended surface recovery periods immediately following release. This behavior may render turtles more susceptible to shark predation or boat strike. Through the course of this study I gained insight into habitat utilization of juvenile sea turtles in the lower Cape Fear River. This area is dominated by marshes, small coves and bays, sand islands, and tidal creeks. The dominate marsh grasses are Spartina sp., Juncus roemerianus, and Salicornia sp., and the bottom substrate is a mud and sand mix. During the period when my study was conducted (May and October 2007) there was low rainfall and I recorded high salinities (32-39 ppt), consistently clear water, and Tw of 23.3 C to 32.3 C. As previously mentioned, 10 of 18 turtles were captured along a partially submerged, human engineered rock wall that divides two northern bays from the river to the west (Fig. 3). I identified various types of algae growing on the rock wall and in the bays that could serve as potential food sources for green turtles, including Enteromorpha, Gracilaria, Halymenia, Codium, Ulva and Sargassum species. Several types of algae present in the lower Cape Fear River have been observed in stomach contents of green turtles (Bjorndal, 1997). I also observed many different invertebrates living along the rock wall which could serve as a food source for juvenile Kemp s ridley turtles, including blue mussels (Mytilus edulis), mud snails (Ilyassoma obsoleta), blue crabs (Callinectes sp.), stone crabs (Menippe mercenaria), spider crabs (Libinia sp.) and lady crabs (Ovalipes ocellatus). Stomach contents of Kemp's ridley turtles stranded in Texas consisted of 93.6% crabs, 3.2% shrimp, 2.2% molluscs, 0.4% fish, and 0.3% vegetation (Shaver, 1991). Analysis of fecal content of Kemp's ridley turtles in the coastal waters of New York revealed that this 56

66 species preyed upon spider crabs (Libinia emarginatus), rock crabs (Cancer irroratus), lady crabs (Ovalipes ocellatus), blue mussels (Mytilus edulis), and various algae (Fucus sp., S. natans, Ulva, Z. marina) (Burke et al., 1993; Burke et al., 1994). Home ranges for sea turtles in nearshore habitats include both foraging sites and shelter sites where turtles rest during nocturnal hours. Their movements are typically limited to a 5 km 2 area (Brill et al., 1995; Seminoff et al., 2002; Makowski et al., 2005). In areas where food resources are patchily distributed, home ranges for green turtles tend to be larger than in areas where food resources are concentrated and abundant (Seminoff et al., 2002; Makowski et al., 2006). Home range for juvenile Kemp s ridley turtles (5 30 km 2 ) tends to be slightly larger than that observed for green turtles (Schmid et al., 2003). Previous studies to assess home range of sea turtles in nearshore habitats used radio and sonic telemeters with transmission ranges of 1-3 km. Satellite transmitters provide a less accurate estimate of location, but permit remote monitoring of sea turtle movements over a larger area. Godley et al. (2003) documented nearshore resident movements of green turtles on the coast of Brazil in July, and their data showed that resident movements were within a few square kilometers, despite low quality location data. Location data from the turtles that remained in the lower Cape Fear River after release from gillnets were of low quality location class, which likely resulted in an overestimation of range. Ninety-five percent of high quality locations that I received placed turtles within a km 2 area of their capture sites, which is within reason when compared to previous estimates of inshore sea turtle movements. The "hot spots" that were described by 25-50% of all location data encompassed areas of km 2, respectively, which included the capture sites of 11 of the turtles in this study (Fig. 3). 57

67 The data I obtained support the growing body of evidence that the Cape Fear River is an important seasonal foraging ground for juvenile green and Kemp's ridley turtles. My results also provide evidence that gillnet entanglement results in a prolonged surface recovery for sea turtles, and entanglement times of 4 hours or less result in postrelease mortality rates of %. Currently, seasonal gillnet attendance regulations in the lower Cape Fear River greatly restrict fishing activities during the summer months when sea turtles are present in large numbers (Pate, NCDMF, 2006). Results from my study suggest that these management decisions are justified and should continue to be enforced. 58

68 Table 4: Track duration and percentage of high quality locations (LC 3, LC 2, LC 1) for entire track duration for all turtles captured. Turtle ID Confirmed mortality Track Duration (days) % High Quality Locations Lk Suspected mortality Lk Cm Cm Survivor Lk Cm Cm Cm Cm Cm Cm Cm Cm Cm X ± SD 17.0 ±

69 Figure 2: Satellite and VHF radio tags deployed on turtle Lk 1 prior to release in lower Cape Fear River. 60

70 Figure 3: Squares are capture locations of all turtles captured (N = 18). Circles are filtered location data for all turtles that remained in Cape Fear River for track duration. Inner two circles represent 25% contours and outer circle represents 50% contour for all turtles that remained in the Cape Fear River for the track duration. 61

71 Percent of Total Transmissions LC3 LC2 LC1 LC0 LCA LCB LCZ Location Class Figure 4: Proportions of location classes of the total transmissions received for all turtles (N = 14) tagged in the lower Cape Fear River, North Carolina. 62

72 Figure 5: Filtered location data for turtle Cm 3, suspected mortality. Stars are LC 3 locations, triangles are LC 2 locations, squares are LC 1 locations and circles are LC A and LC B locations. 63

73 a) 64

74 b) Figure 6: a) All filtered location data for turtle Cm 13, b) expanded view of the lower Cape Fear region filtered location data for turtle Cm 13. Triangles are LC 2 locations, squares are LC 1 locations and circles are LC A and LC B locations. 65

75 Figure 7: Filtered location data for turtle Lk 2, confirmed mortality. Stars are LC 3 locations, triangles are LC 2 locations, squares are LC 1 locations and circles are LC A and LC B locations. 66

76 Figure 8: Filtered location data for turtle Lk 4, suspected mortality. Stars are LC 3 locations, triangles are LC 2 locations, squares are LC 1 locations and circles are LC A and LC B locations. 67