MOVEMENT PATTERNS, FORAGING ECOLOGY AND DIGESTIVE PHYSIOLOGY OF BLACKTIP REEF SHARKS, CARCHARHINUS

Transcription

1 MOVEMENT PATTERNS, FORAGING ECOLOGY AND DIGESTIVE PHYSIOLOGY OF BLACKTIP REEF SHARKS, CARCHARHINUS MELANOPTERUS, AT PALMYRA ATOLL: A PREDATOR DOMINATED ECOSYSTEM A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERITY OF HAWAI I IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN ZOOLOGY DECEMBER 2008 By Yannis P. Papastamatiou Dissertation Committee: Kim Holland, Chairperson Jeff Drazen Steve Karl Christopher Womersley Tadashi Fukami

2 UMI Number: INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. UMI Microform Copyright 2009 by ProQuest LLC All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest LLC 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, MI

3 We certify that we have read this dissertation and that, in our opinion, it is satisfactory in scope and quality as a dissertation for the degree of Doctor of Philosophy in Zoology. DISSERTATION COMMITTEE Chairperson ii

4 ACKNOWLEDGMENTS I want to particularly thank my chair, Dr Kim Holland for taking me into his lab, and for teaching me some of the tricks of the trade. I would like to thank Dr s Jeff Drazen, Steve Karl, Christopher Wormersley, and Tadashi Fukami for serving on my committee and always being available for advice when I needed it. I want to thank Dr Carl Meyer for providing me with a large number of opportunities (and employment) while a graduate student. I would also like to thank Dr Joanne Leong and Jane Ball for supporting me for a number of years and giving me the freedom to go to remote locations and do research. I would like to thank my colleagues Christopher Lowe, Alan Friedlander and Jenn Caselle, without whom I would have never completed the work at Palmyra. For helping me in the field in Palmyra I would like to thank T. Clark, N. Whitney, L. Davis, L. Max, M. Sheehy, and G. Goodmanlowe. For all experiments with captive sharks I would like to thank my army of volunteers E. Aus, A. Stankus, T. Tinhan, M. Burns, J. Coloma, E. Grau, C. Clarke, W. Connel, and J. Nakaya. For advice, discussions and arguments I would like to thank my lab mates N. Whitney, T. Clarke, D. Itano, J. Dale, T. Daly Engle, and M. Hutchinson. I would also like to acknowledge J. Dale and A. Taylor for statistical advice. L. Wedding and K. Anthony designed the maps of Palmyra. Finally I would like to thank Alexi and Caroline Papastamatiou, and Lori Davis for putting up with rather a lot from me, but still always being there when I needed it. Most importantly, I would like to dedicate this dissertation in the memory of Dimitri iii

5 Papastamatiou for pushing me to do something I knew I wanted to do, but didn t realize it at the time. Funding was provided by the National Geographic Society (grant # ), a UH Sea Grant Project Development fund, PADI Project Aware, and a grant from the Fish Aggregating Devices as Instrumented Observatories of Pelagic Life (FADIO) under DG Research of the European Commission. All experiments were approved by the University of Hawaii Animal Care Committee. iv

6 ABSTRACT Apex predators may have a strong regulatory function in marine ecosystems through both density and behaviorally mediated effects. Understanding the ecological impacts of apex predators is particularly important in predator dominated ecosystems where intra-specific competition may be high. While a number of techniques are available for quantifying predator movement patterns and distribution, little is known of the causative factors that regulate these behaviors. One important aspect of predator behavior is foraging, and an important regulating aspect of foraging is digestion. To advance our understanding of the interrelationship between gastric function and foraging behavior, I tested two types of data loggers for deployment in shark stomachs. One type of tag measured stomach acidity, the other the motility of the stomach wall. Both types of tags were deployed in the stomachs of captive free-swimming blacktip reef sharks to determine the effects of feeding and fasting on gastric digestive function. Gastric ph was maintained low during long periods of fasting, suggesting continuous secretion of acid. Gastric motility was higher for meals of mackerel than for similar sized meals of squid with maximum motility occurring at meal sizes of 1 % body weight. Based on diel patterns of gastric motility and ph, I predict that blacktips will feed daily and preferably forage during times of low water temperature. Palmyra Atoll is a remote, predator dominated ecosystem, and has a large population of blacktip reef sharks. Blacktips at Palmyra are smaller than those at other locations, which may be the result of food-limited growth due to intra-specific competition. Palmyra consists of two lagoons (east and west), and abundance of sharks appears to be similar in both lagoons. Active and passive tracking was used to study the v

7 movement patterns of the sharks at Palmyra. Sharks in the west lagoon utilized small home ranges over scales of days to weeks. Adult sharks selected ledge habitats, while smaller individuals selected sand-flats, and small pups were found in very shallow waters. Fractal analysis revealed that sharks used patches that were 3 17 % of the spatial scale of their home range, and that sharks move with a directed walk while in patches but move randomly between patches. Sharks in the west lagoon showed strong site fidelity with some individuals being detected there for over 3 years. Sharks showed little movements between lagoons and sharks in the east lagoon had shorter residence times. Sharks in the west lagoon had higher body condition indices than those in the east lagoon and stable isotope analysis revealed that trophic structure was different between the two lagoons. Conditions differ between the two lagoons which may be driving differences in foraging success. This study reveals the importance habitat can play in the movement patterns, home range and foraging success of sharks and suggests that intra-specific competition could be a strong regulator of apex predator populations in pristine predator dominated ecosystems. Studies of the digestive system revealed that physiology may also regulate some aspects of movement patterns, although field studies will be required to test these hypotheses. vi

8 TABLE OF CONTENTS ACKNOWLEDGEMENTS... iii ABSTRACT... iv TABLE OF CONTENTS... vii LIST OF TABLES... ix LIST OF FIGURES...10 CHAPTER I: Introduction...12 CHAPTER II: The response of gastric ph and motility to fasting and feeding in free swimming blacktip reef sharks, Carcharhinus melanopterus 21 ABSTRACT...21 INTRODUCTION...22 MATERIALS AND METHODS...24 RESULTS...31 DISCUSSION...44 CHAPTER III: A new acoustic ph transmitter for studying the feeding habits of free ranging sharks 53 ABSTRACT...53 INTRODUCTION...53 MATERIALS AND METHODS...55 RESULTS...59 DISCUSSION...59 CHAPTER IV: Scale dependent effects of habitat on movements and path structure of blacktip reef sharks, vii

9 Carcharhinus melanopterus, at Palmyra Atoll: a predator dominated ecosystem 67 ABSTRACT...67 INTRODUCTION...68 MATERIALS AND METHODS...70 RESULTS...81 DISCUSSION...97 CHAPTER V: Distribution, size frequency, and sex ratios of blacktip reef sharks, Carcharhinus melanopterus, at Palmyra Atoll: a predator dominated ecosystem ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CHAPTER VI: Habitat influence residence time and foraging success of blacktip reef sharks at a predator dominated coral ecosystem ABSTRACT INTRODUCTION MATERIALS AND METHODS.137 RESULTS 144 DISCUSSION..160 Chapter VII: Conclusion..168 REFERENCES viii

10 LIST OF TABLES Table Page 1. Summary information for blacktips used in experiments with ph and motility data-loggers Multiple regression of square root transformed 7 h motility against stomach temperature, meal size, and meal size Summary information for sharks used in ph transmitter experiments Blacktip reef sharks actively tracked at Palmyra Atoll Home range and movement statistics for blacktip reef sharks actively tracked at Palmyra Atoll Summary of acoustic monitoring data for 9 blacktip reef sharks tagged in the west lagoon of Palmyra Atoll Percentage detections by VR2 receivers at 8 locations throughout the Palmyra lagoons, for 6 acoustically tagged sharks Linear length-length relationships for blacktip reef sharks at Palmyra Atoll CPUE and sex ratios for sharks caught on sand-flats within the west and east lagoons at Palmyra Atoll Summary information for sharks tagged with acoustic transmitters in the Palmyra lagoons Percentage detections at individual receivers by blacktip reef sharks Null model pair-wise spatial overlap indices for acoustically tagged blacktip reef sharks 154 ix

11 LIST OF FIGURES Figure Page 1. Continuous measurements of gastric ph and temperature in free swimming blacktip reef sharks Continuous measurements of gastric motility and temperature in free swimming blacktip reef sharks Examples of lag in motility following feeding in two free swimming blacktip reef sharks Effect of meal size on mean motility in blacktip reef sharks during the seven hours following consumption of mackerel, measured using a motility data-logger Spectral analysis (FFT) of gastric motility data from blacktip reef sharks FFT of gastric motility from sharks 2, 1, 3, and An acoustic ph transmitter with transducer and ph electrode labeled Continuous measurements of gastric ph and temperature in free-swimming blacktip reef sharks, as measured with a ph transmitter Linear regression analysis between meal size and area under the ph-time curve from blacktip reef sharks Map of the Line Islands relative to the Central Pacific Ocean Home range of six blacktip reef sharks at Palmyra Atoll Habitat selection by blacktip reef sharks at Palmyra Atoll Changes in VFractal with scale for movements of all blacktip reef sharks combined Fractal analysis of blacktip reef shark movement patterns at Palmyra Atoll Seasonal movements of two blactip reef sharks from the west to the east lagoons Examples of long term movements of two acoustically tagged blacktip reef sharks

12 17. Location of the Line Islands in the Central Pacific Ocean, and Palmyra s location within the Line Island chain Size frequency histogram for 254 blacktip reef sharks caught at Palmyra Atoll Length-weight regression for juvenile blacktip reef sharks at Palmyra Atoll Relationship between total length and clasper length for male blacktip reef sharks at Palmyra Atoll Location of Palmyra Atoll within the Central Pacific Ocean and the location of VR2 receivers within the lagoons Detection matrix for individual sharks by VR2 s throughout the Palmyra lagoons Non-metric multidimensional scaling ordination of spatial overlap between sharks Scatter plots and FFT spectral analysis of shark movements at Palmyra Atoll Effect of shark total length on body condition indices and stable isotopes

13 Chapter I Introduction Apex predators may regulate ecosystems through top down control via density and behaviorally mediated interactions between predator and prey species (Brown et al., 1999, Dill et al., 2003, Preisser et al., 2005). Due to the low abundance and high mobility of most apex predators, empirical data supporting theoretical predictions of predator regulation are lacking, particularly marine species which also live in a concealing medium (e.g. Heithaus et al., 2002, Dill et al., 2003). Understanding the full ecological impacts of apex predators requires a comprehensive knowledge of the predator s diet, movement patterns, distribution, density, population dynamics and life history characteristics. Furthermore, one of the main regulators of apex predator populations is thought to be intra-specific competition, which suggests that the ecological effects of the predators may be density dependent (e.g. Estes et al., 2003). It is therefore important to study the ecological impacts of predators under base-line conditions where predator population sizes have not been reduced by anthropogenic influences. Sharks as apex predators Many species of sharks are apex predators and may therefore be important regulators of marine ecosystems. However, empirical data validating these assumptions are particularly lacking, and the few studies that have investigated this issue suggest that there may be geographical and species specific differences in the extent of the shark s importance in the ecosystem (e.g. Stevens et al., 2000, Bascompte et al., 2003, Dill et al., 2003). Due to their life history characteristics, shark populations worldwide are declining 12

14 due to over-fishing making it even more crucial that we understand the role of sharks in marine ecosystems because, if in fact their role as apex predators exerts a disproportionate influence over the structure of ecosystems, their removal may have far reaching and rapid impacts (Stevens et al., 2000, Myers and Worm 2003). Marine Protected Areas (MPA s) are one technique attracting much attention as a management tool to protect populations of marine apex predators (e.g. Chapman et al., 2005, Garla et al., 2006, Meyer et al., 2007). The efficient design of MPA s requires understanding which habitats sharks select and how and why they behave in those habitats. There is considerable overlap in quantifying the ecological impacts of shark populations and using that information to design effective MPA s. Acoustic telemetry and its utilization in ecology In the last couple of decades, there have been many advances in acoustic telemetry techniques that can be used to quantify space and habitat utilization of sharks over several spatial scales (e.g. Morrisey and Gruber 1993a, Holland et al., 1993, Weng et al., 2005, Garla et al., 2006). Active tracking involves attaching a small transmitter to a shark, which emits a high frequency acoustic signal approximately every second that can be detected by an underwater receiver attached to a boat or kayak, and which allows the shark to be followed continuously. Active telemetry can quantify fish movements and habitat utilization at high spatial resolution, as the researcher knows exactly where the shark is while it is being tracked. However, tracking can only continue over a scale of days, and therefore movements can only be quantified over short time scales (Morrisey and Gruber 1993a, b, Holland et al., 1993). Passive tracking, where a network of underwater listening stations detect transmitter equipped sharks when they come within 13

15 range of the various receivers, enables shark movements to be studied over scales of years. However, it is not possible to know where the shark is when it is not being detected and so passive telemetry cannot measure habitat utilization with high spatial precision (e.g. Heupel et al., 2004, Garla et al., 2006). A combination of active and passive tracking can therefore quantify movements with both high spatial and temporal resolution. Considerably harder, is being able to understand why animals select the habitats they use and why they exhibit certain behaviors when within those habitats. It is becoming increasingly clear that quantifying physiological processes in free ranging animals is increasing in popularity and can partly help to explain some of the behaviors seen in wild animals (Secor and Nagy 1994, Papastamatiou and Lowe 2004, Burns et al., 2005). Physiological measurements of free-ranging animals have been used to study activity patterns, bioenergetics, feeding habits, and digestion of a wide range of animals (Peters 1997a, Gremilliet et al., 2000, Lowe and Goldman 2001, Itoh et al., 2003). One of the most important behavioral decisions an apex predator has to make is with regards to feeding, and foraging can partially explain the distribution patterns of many predators. In endothermic marine predators, changes in stomach temperature can be used to quantify feeding events (e.g. Gremilliet et al., 2000, Itoh et al., 2003). Unfortunately, stomach temperature cannot be used to detect feeding events in ectothermic predators, and currently we can only speculate as to when a top level fish predator is feeding, based on changes in movement patterns. Conceptually, the physiology and constraints of the digestive system can partially be used to explain an animals foraging behavior. 14

16 Furthermore, quantifying the conditions that lead to optimal digestion can improve optimal foraging models and make predictions of the behavior of the animal in the wild. Digestive physiology of sharks Appetite in vertebrates is thought to be regulated by a variety of visceral and systemic regulatory pathways, integrated with sensory information processed in the brain (Mayer 1994, Sims et al., 1996). In elasmobranchs, there is an inverse relationship between the evacuation of stomach contents and return of appetite and it has been hypothesized that mechanoreceptors in the stomach wall respond to distention, and initiate a graded nervous response leading to regulation of appetite (Sims et al., 1996). Therefore, understanding the physiological regulation of the elasmobranch digestive system will improve our understanding of shark feeding and associated behaviors in the wild. Many species of shark consume their prey whole, yet the morphology of the intestine only allows the entrance of semi-liquid material (Andrews and Young 1993, Motta 2004). Therefore the stomach is responsible for the complete breakdown of whole prey into semi-liquid chyme (Barrington 1942). Until recently, little was known about gastric digestion in sharks other than species-specific measurements of gastric evacuation rates. However, data-loggers have been used to quantify digestive processes in the stomach of a couple of shark species (Papastamatiou and Lowe 2004, 2005, Papastamatiou 2007). Prey digestion is accomplished by the secretion of concentrated hydrochloric acid, protease enzymes, and contractions of the stomach wall (Holmgren and Holberg 2005, Papastamatiou 2007). Shark species differ in their diet and feeding habits, and therefore there should be inter-specific differences in the patterns of gastric 15

17 acid secretion and motility. For example, it has been noted that some species continually secrete gastric acid when fasting, while others temporarily cease acid secretions (Papastamatiou and Lowe 2004, 2005, Wood et al., 2007), which may be related to the feeding frequency of the species in the wild (Papastamatiou and Lowe 2005, Papastamatiou 2007). Validating this will require studying the digestive system of a number of different species, and improving techniques used to quantify feeding behavior of wild sharks. In addition, measuring digestive variables in free-ranging sharks can be used as a proxy for feeding in the field. Specific changes in gastric ph occur every time a shark feeds; therefore continuous measurements of ph of sharks in the field will indicate when the animal is actually ingesting prey (Papastamatiou and Lowe 2004, 2005). Finally, quantifying physiological processes of the stomach can be used to make prediction with regards to what behavior and habitats can optimize digestion. For example, how does the digestive system respond to changes in ambient temperature? How does the stomach respond to different meal types and sizes? Should the species consume as much as possible when foraging, or is there an optimal meal size? Blacktip reef sharks and Palmyra Atoll The blacktip reef shark, Carcharhinus melanopterus, was chosen as a model species as it is a common and abundant shark on coral reefs, is large enough to be fitted with data-loggers and transmitters, and is relatively easy to maintain in captivity. Blacktip reef sharks are one of the most abundant shark species at many atolls and islands in the Pacific and Indian Ocean (Hobson 1963, Stevens 1983, Sandin et al., 2008). Dietary studies in other locations show that blacktips are tertiary predators, and due to the high population densities in many locations, could exert top-down control on coral 16

18 ecosystems (Stevens 1983, Cortes 1999). The only analysis of movement for this species has been for individuals tracked over very short time periods (maximum 7 h), although results suggest that blacktip reef sharks move over limited areas (Stevens 1983). Thus they are an ideal study animal for both ecological and logistical reasons. I had the opportunity to study blacktip reef sharks at Palmyra Atoll. Palmyra is located in the central Pacific Ocean and has been a US National Wildlife Refuge since Before this time, Palmyra was largely uninhabited, although it was occupied by the US Navy during World War 2. Dredging and construction substantially changed the structure of the atoll, but the military left at the end of WW2, and the atoll was largely uninhabited from then until Palmyra s designation as a refuge. As a consequence, the lack of anthropogenic influences has lead to a high abundance of apex predators, where sharks make up nearly 35 % of the fish biomass (Sandin et al., 2008, DeMartini et al., 2008). Therefore, Palmyra is one of the few locations on earth where base-line numbers of apex predators can be found and subsequent intra-specific competition studied. I therefore conducted a study of the digestive physiology, movement patterns and foraging ecology of blacktip reef sharks at Palmyra Atoll. Goals and objectives The overall goal was to understand the ecological impacts of blacktip reef sharks at Palmyra Atoll, a location where, because of their apparently high densities, sharks should be under high levels of intra-specific competition, and the influence of these predators on the reef ecosystem should be maximal. This was to be accomplished through a detailed analysis of movement patterns and space utilization, foraging ecology and trophic relationships, and basic life history characteristics. Furthermore, an 17

19 experimental study of gastric digestion in blacktips under captive conditions was conducted from which I could make predictions about shark behavior and how this could optimize digestion. Although I was not able to directly test these predictions, I was able to gather evidence and provide testable hypotheses on the effect digestive physiology may have on regulating shark behavior in the wild. The various components of the overall study are described in the following chapters Chapter 2 describes an experimental study of gastric digestion in free-swimming captive black tip reef sharks. I fitted captive sharks with gastric data-loggers that measured gastric ph, motility and temperature. I determined how meal size and type influenced gastric digestion, and also quantified natural cycles in gastric digestion. Chapter 2 has been published as Papastamatiou YP, KN Holland, SJ Purkis The response of gastric ph and motility to fasting and feeding in free-swimming blacktip reef sharks, Carcharhinus melanopterus. J. Exp. Mar. Biol. 345: Chapter 3 describes the testing of a custom designed ph acoustic transmitter. The transmitter was tested with captive blacktip reef sharks, and sharks were fed meals of fish at different sizes. The goal was to determine if a ph transmitter could quantify feeding frequency and meal size in free-ranging blacktip reef sharks. Chapter 3 has been published as Papastamatiou YP, CG Meyer, KN Holland A new acoustic ph transmitter for studying the feeding habits of sharks. Aquat. Liv. Res. 20(4):

20 Chapter 4 describes Scale dependent effects of habitat on movements and foraging strategies of blacktip reef sharks at Palmyra Atoll. I used active telemetry to measure short term movements, behavior and habitat utilization of blacktip reef sharks at Palmyra Atoll. Chapter 4 has been accepted for publication in Ecology. Chapter 5 describes Distribution, size frequency, and sex ratios of blacktip reef sharks at Palmyra atoll. I fished for sharks throughout the Palmyra lagoons to describe the distribution of sharks, and how sex ratios may vary spatially. I also quantified the size frequency of sharks at the atoll and determined size of maturity of male sharks. Chapter 5 is currently in review in journal of Fish Biology. The final chapter, Chapter 6, describes Habitat influences residence time and foraging success of blacktip reef sharks at Palmyra Atoll. I used passive telemetry to look at movements of sharks between two lagoons over a time scale of several years. I also used stable isotopes and a body condition index to determine if trophic structure and foraging success differed for sharks between the two lagoons. My goal was to correlate movements and behavior with foraging success. Collectively, I made prediction of shark foraging behavior based on the physiology of the digestive system. I obtained evidence that some of these predictions may be true, although I was not able to verify them. I also quantified the movements of a population of sharks with both high spatial and temporal resolution, and provide the first 19

21 evidence showing that differences in habitat use and movements can affect foraging success. 20

22 Chapter II The response of gastric ph and motility to fasting and feeding in free swimming blacktip reef sharks, Carcharhinus melanopterus ABSTRACT In many fish and reptiles, gastric digestion is responsible for the complete breakdown of prey items into semi-liquid chyme. The responses of the stomach to feeding and to periods of fasting are, however, unknown for many lower vertebrates. We inserted data loggers into the stomachs of free-swimming captive adult blacktip reef sharks (Carcharhinus melanopterus) to quantify gastric ph, motility and temperature during fasting and following ingestion of food. Gastric acid secretion was continuous, even during long periods of fasting, with a mean ph of 1.41 ± 0.40 (± 1SD) when the stomach was empty. Stomach contractions were greater following meals of mackerel than for those of squid. Gastric motility following feeding on mackerel, was positively influenced by ambient temperature, and followed a quadratic relationship with meal size, with maximum motility occurring after meals of % body weight. Diel changes in gastric motility were apparent, and were most likely caused by diel changes in ambient temperature. Gastric digestion in blacktip reef sharks is affected by both biotic and abiotic variables. We hypothesize that behavioral strategies adopted by sharks in the field may be an attempt to optimize digestion by selecting for appropriate environmental conditions. 21

23 INTRODUCTION Gastric digestion in carnivorous vertebrates is responsible for the breakdown of ingested prey items into semi-liquid chyme. The role of the stomach is particularly important in lower vertebrates such as fish and reptiles, many of which ingest their prey whole with little mastication (e.g., Secor 2003, Motta 2004). Two components to gastric digestion occur: chemical digestion accomplished by the secretion of concentrated hydrochloric acid (HCl) and digestive enzymes, and mechanical digestion accomplished by muscular contractions of the stomach wall (Mayer 1994, Holmgren and Holmberg 2005). Elasmobranch fishes are one of the first group of carnivorous vertebrates to have evolved a functional stomach and (based on the identification of H + -K + ATPase in acid secreting cells) probably one of the first to have evolved an acid secreting stomach (Smolka et al., 1994). In addition, the morphology of the stomach permits only the passage of semi-liquid chyme into the intestine, yet many species of elasmobranch ingest their prey whole, highlighting the importance of the stomach to food breakdown (Andrews and Young 1993, Motta 2004). Elasmobranchs are capable of secreting highly acidic gastric fluids (down to ph 0.4, Papastamatiou and Lowe 2004, 2005). Distention of the stomach wall as food enters is the initial stimulus for increased acid secretion (Smit 1967), followed by the action of secretagogues such as gastrin and histamine, although the interactions between hormones and acid secretion are not well known (Hogben 1967, Vigna 1983). There are interspecific differences among elasmobranchs in the response of gastric acid secretion to fasting, with some species continuously secreting acid while others periodically cease 22

24 secretions during fasting (Barrington 1942, Papastamatiou and Lowe 2004, 2005). Secreted HCl aids in the physico-chemical breakdown of the hard parts of prey and contributes to enzymatic digestion by converting the inactive zymogen pepsinogen into the proteolytic enzyme pepsin (Guerard and Le Gal 1987, Holmgren and Nilsson 1999). Some elasmobranchs are also capable of secreting chitinase enzymes, which also have optimal function at low ph, and break down chitin-containing exoskeletons (Fange et al., 1979). To date, gastric motility has only been measured in euthanized or anaesthetized elasmobranchs, although results suggest that gastric motility is under the control of both nervous and hormonal mechanisms (Andrews and Young 1993, Holmgren and Nilsson 1999, Buddington and Krogdahl 2004). A variety of neurotransmitters have been identified in elasmobranch gut neurons (Nilsson and Holmgren 1988) and it appears that there is both nervous inhibition and excitation of the stomach muscles (Campbell 1975, Andrews and Young 1993). Elasmobranchs are known to have relatively slow gastric evacuation rates (Wetherbee et al., 1990), and electrical stimulation of the splanchnic nerve in lesser spotted dogfish, Scyliorhinus canicula, induced gastric contractions but peristalsis did not move the stomach contents into the small intestine (Andrews and Young 1993). Presently, it remains unclear whether gastric motility in elasmobranchs only functions to mix food items and to pass chyme out of the stomach, or if motility is also involved in mechanical trituration. Gastric evacuation rates (and presumably motility) in elasmobranchs are influenced by a variety of factors including: meal size, surface area of ingested prey, prey lipid composition, the presence of skeletal or chitin containing hard-parts, and feeding periodicity (Wetherbee et al., 1990, Schurdak and 23

25 Gruber 1989). In summary, it is probable that species specific differences in stomach motility and patterns of acid secretion are shaped by species specific diet and feeding strategies. Presently, very little is known of the response of gastric acid secretion and motility following feeding and during fasting in free-swimming elasmobranchs. Obtaining such data under semi-natural conditions is important as it enables the physiological response of the stomach to be put into an ecological context, and subsequently applied to the study of the feeding strategy and optimal foraging behavior of the animal in the wild. Our goals were to quantify changes in gastric ph, temperature and motility in a captive free-swimming elasmobranch, the blacktip reef shark (Carcharhinus melanopterus), using autonomous data-loggers under semi-natural conditions. The blacktip reef shark was chosen as a model species because it is an abundant predator on coral reefs in tropical and semitropical regions of the Pacific and Indian Oceans (Compagno et al., 2005), is large enough to retain gastric data-loggers, and feeds and behaves normally while in captivity. Our specific objectives were to: (1) determine the post-prandial changes in gastric ph and motility in free-swimming captive blacktip reef sharks; (2) quantify the influence of meal size, meal type, and temperature on gastric motility; (3) determine the response of ph and motility during periods of fasting; and (4) determine if there were any diel changes in the profiles of gastric ph and motility. Because many species of shark are considered nocturnal foragers (Wetherbee et al., 1990), we hypothesize that diel differences in gastric digestion will occur. MATERIALS AND METHODS 24

26 Study animals Tests were conducted with five captive adult blacktip reef sharks (Carcharhinus melanopterus, Quoy & Gaimard 1824), total length ± 6.8 cm (mean ± 1 SD) and mass 21.8 ± 3.1 kg (Table 1). All sharks were maintained at the Hawaii Institute of Marine Biology in a sectioned off lagoon (120 x 20 m) consisting of coral rubble, coral, and sand, with a maximum depth of 3 m. The lagoon is tidally flushed and contains a fish and invertebrate community typical of Kaneohe bay, Oahu, Hawaii. Prior to testing, sharks were fed to satiation two to three times a week with mackerel (Scomber spp.). Animals used in experiments were moved into a smaller rectangular section (approximately 10 x 20 m), with similar habitat characteristics as the rest of the lagoon. No more than two sharks were maintained in the testing area at any one time. Sharks were acclimated to the test area until they resumed feeding, after which they were fasted for a week before the experiments began. Sharks were fitted with one of two types of data-logger measuring either stomach ph or gastric motility. Gastric ph To measure gastric ph and temperature in free-swimming blacktip reef sharks, we used autonomous ph/temperature data-loggers (earth & Ocean Technologies, Kiel Germany). The data-loggers are cylindrical (11 x 2 cm), weigh approximately 80 g in air and consist of a ph micro-glass electrode, a reference electrode with a free-diffusion liquid junction and a 12-bit data-logger encased in a titanium shell (Peters 1997 a, b). The reference electrodes are designed to compensate for any pressure changes associated 25

27 Table 1 Summary information for adult blacktip reef sharks (Carcharhinus melanopterus) used in experiments with ph and motility data-loggers. Shark # TL (cm) Mass (kg) Sex Min ph Max ph Mean ph F F M F F

28 with diving (Peters 1997b). A sensor on the data-logger also measured temperature (resolution of 0.1 ºC). Before deployment, the ph data-loggers were programmed to record ph and temperature every 30 sec and were calibrated in NBS standard ph buffers (1.68, 4.01, 6.86, and 10.01). To deploy the ph data-loggers, we netted a shark and inverted it in a stretcher to induce tonic immobility (see Papastamatiou and Lowe 2005). Additional anesthesia was induced by inserting a 2 cm diameter siphon into the mouth and applying a solution of MS 222 (0.15 g l -1 ) to the gills. It took between 5 10 min before the shark was anaesthetized to a level of immobility, after which we inserted a lubricated 3 cm diameter PVC pipe through the mouth into the stomach. The ph data-logger was dropped down the pipe with the ph sensor pointing towards the caudal fin (i.e. at the base of the cardiac portion of the stomach), followed by pieces of bait fish to prevent premature regurgitation of the data-logger. We then removed the pipe and measured and sexed the shark before reviving it by manually swimming the animal through the lagoon water. Each shark revived within approximately min, after which it was observed for an additional 15 min to ensure normal swimming behavior. We determined shark mass using the lengthweight regression Weight= * 10-6 (Total length) 3.39 (Stevens, 1984). During the period that the ph data-logger was retained in the stomach, we fed each shark meals of mackerel (Scomber spp.) at a variety of ration sizes (Table 1). Two sharks were also fed meals of reef fish (various Acanthurus and Chaetodon species), and one shark was also fasted for 12 d. The data logger was deployed in each shark only once. The ph data-loggers can record accurate ph data for up to 16 d, depending on electrolyte outflow rate (see Peters 1997b). If the shark had not regurgitated the data- 27

29 logger within 16 d, then we restrained the shark as described above and used a magnetic device to remove it from the stomach. After retrieval, the data-loggers were re-calibrated with the same NBS standard ph buffers used prior to deployment. The data were then downloaded and analyzed using phg 2.0 software (Jensen Software Systems), which interpolates and corrects ph data for any drift of the electrode and also for changes in stomach temperature (Peters 1997a). Error analysis of ph electrode performance was determined using the ph drift model described by Peters (1997a). We determined titration time for each meal that each shark consumed, with titration time defined as the time taken for ph to return to 2.0 (baseline) (Gardner et al., 2002). To determine the time of onset of a response, we first established a baseline by analyzing gastric acidity in the two hours prior to feeding. This period was divided into 10 min blocks and onset of a response (P 1 ) was defined as the first of two consecutive 10 min intervals where ph was < 2.0 for only 5 % of the time. We analyzed the 24 h period following feeding in the same way and defined the end of the response (P 2 ) as the first of two consecutive 10 min intervals where ph was > 2.0 for less than 10 % of the time. Titration time was calculated as P 2 - P 1. We used linear regression analysis to quantify the relationship between meal size and titration time. For each meal, we also measured the area under the ph curve using ArcView GIS (ver 3.2). A linear regression was used to compare meal size to area under the ph - elapsed time curve. Gastric motility We measured gastric motility using a motility/temperature data-logger (14 x 1.9 cm, length x diameter, 45 g in air, earth & Ocean Technologies, Kiel Germany). The 28

30 sensor consists of a piezoelectric film encased in a flexible silicon bulb, connected to an 8-bit data-logger. Movement of the piezoelectric film generates a voltage, the size of which is a function of the extent and speed of deflection (Peters 2004). The motility sensor provides a cumulative measure of stomach muscle activity over time. In our case, the data-logger was programmed to record stomach motility every 15 sec. A temperature sensor coupled to the data-logger also enabled simultaneous measurements of stomach temperature (resolution 0.1 ºC). The stomach motility and temperature (SMT) data-loggers were deployed as described above for the ph data-loggers. However, the former were deployed with the sensor pointing towards the mouth. To evaluate any spatial differences in gastric motility within the stomach, one shark had the SMT data-logger deployed with the motility sensor pointing towards the caudal fin. During deployment, sharks were fed squid (Loligo spp.) or mackerel (Scomber spp.) at a variety of ration sizes. SMT data-loggers were either regurgitated by the shark or we retrieved them as described. After retrieval, data from the SMT data-loggers were downloaded and analyzed using phg 2.0 software (Jensen Software System). We used a General Linear Model (GLM) to evaluate the effects of temperature, meals size and meal type on gastric motility. In all cases, motility was the dependent variable while meal size, temperature, and meal type were covariates. Meal size and temperature were also set as interactive variables. Two measures of motility were used: (1) the mean over the first 7 h after feeding, and (2) the mean over the first 24 h after feeding. We used 7 h in addition to 24 h because there appeared to be an approximate 7 h delay between feeding and the onset of the strongest contractions and we wanted to test if 29

31 there were differences in motility related to feeding during the 7 h lag period (see results). Because the GLM showed that motility differed between mackerel and squid, we used multiple regression analysis for mackerel and squid meals separately. Motility values were not normally distributed, so we applied a square root transformation. The effect of meal size on motility appeared to be best described by a quadratic equation, so meal size was squared. In all cases, the residuals from the GLM and regressions were examined to ensure that all assumptions of the models were met. All GLM and multiple regression analysis were performed using Minitab (ver. 14). Due to the logistics associated with maintaining large adult sharks in captivity, we only had five sharks with which to deploy data-loggers (Table 1). As a consequence, we deployed the motility logger in each shark on two separate occasions (with the exception of shark # 2, in which it was only deployed once). Although this constitutes a degree of pseudo-replication, we visually checked the distribution of all data points to ensure that statistical analyses were not strongly influenced by data from one individual (by examining maximum and minimum data points). We used time series analysis to determine if there were any cyclical patterns in gastric motility, applying a Fast Fourier Transformation (FFT) which converts time-series data into frequencies, thereby facilitating the identification of temporal periodicity in the dataset. The FFT produces a power spectrum with the power of each frequency being dependent on how well the data fit the sinusoidal wave of that particular frequency (Chatfield 1996). The time period of the event could then be calculated as the inverse of frequency, with each block of data equivalent to 15 sec (the sampling rate of the datalogger). For example, there are 5760 data blocks (each equivalent to 15 sec) in a 24 h 30

32 period, which translates to a frequency of All motility data were smoothed using a Hamming window before running the FFT (Chatfield, 1996). FFT analysis was performed using Statistica (ver.7). RESULTS Gastric ph Drift of the ph electrodes were generally low, with resolution varying between ph units and error between (Table 1). Regardless, the blacktip reef sharks maintained an acidic stomach at all times (maximum ph: 5.3, Fig.1, Table 1). During periods of fasting (> 48 h after feeding), gastric ph was 1.41 ± 0.40 (mean ± 1SD). In all sharks, feeding caused a rapid increase in gastric ph (decrease in acidity) of 1.66 ± 0.41 units to a peak value of 3.15 ± 0.41, followed by a gradual decrease back down to baseline (more acidic) levels. The rate of increase in ph following feeding (0.027 ± ph units/min) was faster than the subsequent decrease ( ± ph units/min, t test paired sample for means, t = 3.77, p = 0.007). There was no significant effect of meal size on titration time (p = 0.26, F = 1.93) or area under the ph/time curve (p = 0.34, F = 1.25). However, the regression was strongly influenced by one outlier point. When this point was removed, meal size (expressed in g) affected both titration time (p = 0.04, F = 21.7, r 2 = 0.87) and area (p = 0.04, F = 36.0, r 2 = 0.92). Shark #3 was fasted for 12 d and showed ph profiles with two separate phases (Fig. 1c). For the first seven days of fasting, ph remained relatively stable between 1.4 and 2.1, but after day seven, ph started to fluctuate between 5.3 and 0.4 even though no 31

33 Figure 1. Continuous measurements of gastric ph and temperature in free swimming blacktip reef sharks (Carcharhinus melanopterus). Lower line is gastric ph, upper line is gastric temperature. Arrows indicate time of feeding, and the number above each arrow represents meal size expressed as % BW.? indicates that a meal of unknown size was consumed. Meal codes are M for mackerel (Scomber spp.), RF for reef fish (Acanthurus, Chaetodon), and S for squid (Loligo spp). Data from individual sharks are shown in separate panels: (a) shark #1, (b) shark #2, (c) shark #3 (fasted for entire duration of deployment), (d) shark #4. 32

34 10 a b ph RF 0.5 M Temperature (C) ph 6 4? 1.1 M 1.1 RF Temperature (C) Time (d) Time (d) 0 c d ph Temperature (C) ph M 0.88 M 0.88 S 0.3 M Temperature (C) Time (d) Time (d) 33

35 feeding occurred. FFT analysis was used to analyze both these phases in shark #3. As expected, no major peaks in the density spectrum were observed during the first phase (when ph remained constant). The second phase produced two peaks however; one at 27.8 h and one at 41.7 h. We interpret this as a diel fluctuation in ph, with ph being lowest between in the morning and highest during the late afternoon. Gastric motility Motility appeared to be reduced during the first two days of deployment and consequently motility data were only used from meals given to sharks > 2 d after deployment of the logger (Fig.2). All sharks showed a delay of 7 12 h following feeding, before the onset of strong contractions (Fig. 3). The results of the GLM showed that both meal type (F=13.79, p=0.006) and stomach temperature (F=6.44, p=0.035) affected motility during the 7 h post-prandial period, while only meal type (F=9.16, p=0.019) affected motility during the first 24 h post-feeding. Meals of mackerel elicited stronger contractions (0.60 ± 0.37 relative units) than meals of squid (0.21 ± 0.07 relative units, F=13.79, p=0.006). Multiple regression analysis for mackerel meals showed that meal size 2, and stomach temperature affected motility during the 7 h following feeding (r 2 =82.3, F=10.31, p=0.043, Table 2), but not over the 24 h following feeding (F=3.95, p=0.145). The effect of meal size on 7 h post-prandial motility was best described by a quadratic equation (r 2 =0.53, Fig. 4). Temperature and meal size did not affect motility during the first 7 or 24 h following consumption of squid meals (F=0.68, p=0.572). The FFT spectra showed a motility peak for all sharks at a frequency of approximately , regardless of whether the shark ate during that time period (Fig.5). 34

36 Figure 2. Continuous measurements of gastric motility and stomach temperature in free swimming black tip reef sharks (Carcharhinus melanopterus). The upper trace shows stomach temperature. Data from individual sharks are shown in separate panels: (a) Shark #5 (In this instance the data logger was deployed with sensor pointing towards caudal fin whereas all other sharks had sensor deployed pointing towards mouth.), (b) shark #4, (c) shark #3 (fasted for entire deployment), (d) shark #2 (The gap in data set was due to data-logger failure). Arrows indicate time of feeding, and number above arrow represents meal size expressed as a percentage of body mass.? indicates a meal of unknown size was consumed. Meals codes are M for mackerel (Scomber spp.), and S for squid (Loligo spp.). High contractions towards end of deployment seen in panel s c, and d are most likely attempts to regurgitate data-logger. 35

37 a b Motility (relative units) M 0.8 M Temperature (C) Motility (relative units) M 1.0 M Temperature (C) Time (d) Time (d) c d Motility (relative units) Temperature (C) Motility (relative units) M 1 S? S Temperature (C) Time (d) Time (d) 36

38 Figure 3. Examples of lag in motility following feeding in two free swimming blacktip reef sharks (Carcharhinus melanopterus). Upper line in each graph is stomach temperature. Results from sharks #2 and #4 show raw data (a, c) and running average (b, d). F indicates time of feeding, whereas C shows time of strong contractions. 37

39 a b Motility (relative units) Motility (relative units) F C c 120 F Time (h) C 40 shark Temperature (C) Temperature (C) Mean motility (relative units) Mean motility (relative units) d Time (h) Time (h) Time (h) 38

40 Table 2. Multiple regression of square root transformed seven hour motility against stomach temperature, meal size and meal size 2. Data are from blacktip reef sharks fed meals of mackerel (Scomber spp.). Predictor Coef SE Coef T P Constant Meal size Temp Meal size S=0.105 R 2 = 91.2% R 2 (adj.)=82.3% 39

41 Figure 4. Effect of meal size on mean motility in blacktip reef sharks (Carcharhinus melanopterus) during the seven hours following consumption of mackerel (Scomber spp.), measured using an motility data-logger. The solid line is a curve fitted using a quadratic equation (r 2 = 0.53, p= 0.029). Maximum motility occurs after sharks consume meals of % of body weight. 40

42 1.4 7 h Motility (relative units) Meal size (%BW) 41

43 Figure 5. Spectral analysis (FFT) of gastric motility data from blacktip reef sharks. Data are from sharks # 1-5 (panels a - e respectively). All sharks had the motility sensor pointing towards the mouth, except for shark #5 (e) which had the sensor pointing towards the caudal fin. Different scales were used on the y-axis to clarify data presentation. 42

44 Density Density Density Period (h) a c e Period (h) Period (h) Density Density b Period (h) d Period (h) 43

45 This frequency translates to a time period of 23.4 ± 1.8 h. Shark #5 (the one animal that had the sensor pointing towards the caudal fin) did not show any peaks in the frequency spectra (Fig. 5e). All sharks that fed showed a second peak in motility with a period of 2.0 ± 0.3 h, but this peak was absent from sharks that were fasted (Fig. 6). DISCUSSION The use of intra-lumenal data-loggers to measure digestive variables appears to be a viable technique in medium to large sized sharks. Although we did not quantify the effects of the data-logger on acid secretion in blacktip reef sharks, previous work with leopard sharks (Triakis semifasciata) showed no effect of the data-loggers on gastric acid secretion (Papastamatiou and Lowe 2004). All blacktip reef sharks behaved similarly to non-instrumented sharks, and resumed feeding within one day of deployment. Gastric ph Blacktip reef sharks are capable of secreting highly acidic gastric fluid (minimum measured ph 0.4). Gastric acid secretion appears to be continuous in this species because low ph values were recorded even after long periods of fasting (e.g. shark #3 was fasted for 12 d, see fig.1c). Maximum ph recorded for any blacktip was 5.3, and ph remained at this level for only a short period of time before returning to low levels. It has been proposed that shark species that feed frequently in the wild continuously secrete gastric acid during fasting thereby enabling them to be in a state of physiological readiness for the next meal, whereas sharks which feed less frequently (e.g. nurse sharks, Ginglymostoma cirratum) may periodically cease acid secretion while the stomach is empty as an energy conserving technique (Papastamatiou and Lowe 2004, 2005, 44

46 Figure 6. FFT of gastric motility data from shark #2 (a), #1 (b), #3 (c), and #4 (d). Sharks #2 and #1 (a,b) were fed during deployment of the data-logger, while sharks #3 and #4 (c,d) were fasted. The arrow indicates peaks in the gastric motility spectrum equivalent to 2.0 ± 0.3 h in fed sharks. Different scales were used on the y-axis in panel d to clarify data presentation. 45

47 5e+7 a 5e+7 c 4e+7 4e+7 Power 3e+7 2e+7 Power 3e+7 2e+7 1e+7 1e Log period (min) Log period (min) 5e+7 b 6e+6 d 4e+7 5e+6 Power 3e+7 2e+7 Power 4e+6 3e+6 2e+6 1e+7 1e Log period (min) Log period (min) 46

48 Papastamatiou 2007). Although little is known about the feeding habits of blacktip reef sharks in the wild, they are an active continuously swimming species that lives in semitropical and tropical waters, and spend a considerable amount of time searching over sand flats and along reef ledges (Papastamatiou and Lowe unpublished, Stevens 1984). In combination, these factors suggest that blacktip reef sharks probably have high energy requirements and may have to feed frequently. If this is the case, the present result appears to agree with the hypothesis that feeding frequency influences gastric acid secretion patterns in sharks. Following feeding, a rapid increase in gastric ph occurred with a subsequent gradual decrease back to baseline levels. We interpret the rapid increase in ph as being caused by seawater and the food items themselves (most of which are alkaline) entering the stomach and diluting or buffering the small amounts of gastric fluids that are present in the stomach. After feeding, an increase in gastric acid secretion is presumably triggered by stomach distention and the action of secretagogues such as histamine and gastrin (e.g. Smit 1967, Hogben 1967, Vigna 1983) resulting in re-acidification of the stomach. This interpretation is supported by the fact that the amount of time taken for the stomach to re-acidify appears to be a function of meal size. Gastric pepsin and chitinase enzymes have optimum activity at low ph. For example, pepsin from the lesser spotted dogfish (Scyliorhinus canicula) has an optimum ph of 2.5 (Guerard and Le Gal 1987), while chitinase enzymes from several species of shark and skate show optimal activity at ph of approximately 1.6 (Fange et al., 1979). Although gastric enzymes have not been identified in blacktip reef sharks, it is highly likely that at least one, if not both, of these enzymes are present and the observed gastric 47

49 conditions would be optimal for both enzymes (especially in the h following feeding). Gastric motility In all blacktip reef sharks there appeared to be a delay in heightened stomach activity of 7-12 h following feeding. A delay in active gastric contractions following feeding has also been seen in teleosts such as bluefin tuna, Thunnus thynnus, and rainbow trout, Oncorhynchus mykiss (Carey et al., 1984, Olsson et al., 1999). The delay in active contractions following feeding (also known as gastric accommodation or relaxation), is the initial response following distention of the stomach wall, and is thought to allow for more space in the stomach and for the accumulation of gastric fluids before mixing (Mayer 1994, Holmgren and Holmberg 2005). The delay in motility observed in the current experiment may be related to the post-prandial lag in gastric evacuation of stomach contents observed in other elasmobranchs such as juvenile sandbar sharks, Carcharhinus plumbeus (Medved 1985), and scalloped hammerhead sharks, Sphryna lewini (Bush and Holland 2002). The results from the present study show that gastric motility is a function of abiotic and biotic variables. Meals of squid elicited lower levels of stomach contraction than similar sized meals of mackerel (Scomber spp.), which is contrary to predictions based on data from other vertebrates. Stomach motility in vertebrates is though to be sensitive to lipid levels, with high-lipid prey taking longer to evacuate from the stomach than low-lipid prey (Anderson 2001, Mayer 1994). The mackerel used in the present study have higher lipid levels than those found in squid (e.g. Mackerel is 4.72 % lipids, as opposed to squid which is 1.72 %, O Neal Scientific services inc., MO) and should have 48

50 elicited weaker contractions than squid. These two food items also differ in their physical digestibility however. Squid contains collagen fibers which increase the tissues resistance to digestive action (Jackson et al., 1987). Our results agree with studies of gastric evacuation rates in elasmobranchs. Gastric evacuation rates for little skate (Raja erinacea) fed meals of squid were slower than those fed high lipid sand lance, or lipid poor krill (Nelson and Ross, 1995), while blue sharks (Prionace glauca) took longer to evacuate squid than they did anchovies (Tricas 1977). The reduced motility after squid meals increases the time of exposure of squid tissue to HCl and gastric enzymes, required for the break down of collagen fibers (Jackson et al., 1987). We also found that stomach temperature correlated with increased gastric contractions for meals of mackerel, but not squid. It is well established that gastric evacuation in fish is positively influenced by temperature (e.g. Nelson and Ross 1995, Bush and Holland 2002), but it is unclear why temperature did not influence gastric motility patterns for squid meals in blacktip reef sharks, although the small sample size may have compromised our results. The movement patterns of some elasmobranchs in the field may be for behavioral thermoregulation (e.g. Carey and Scharold 1990, Matern et al., 2000). Moving into warmer water should increase gastric evacuation rates (theoretically lowering digestive efficiency), but our results suggest that this deficit may be countered by improved mixing of stomach contents. The magnitude of gastric contractions during the 7 h following feeding on mackerel was best modeled to meal size using a quadratic equation. Gastric motility increased with meal size until the sharks were consuming % of their body weight (BW), after which there was a decline in motility. It is thought that gastric motility in 49

51 vertebrates increases as a function of distention of the stomach wall (Mayer 1994). Although this has never been explicitly tested in fish, preliminary results from dab (Limnada limnada) suggest that gastric motility is a function of the cube root of the size of stomach contents (Jobling 1974). Previous studies have shown that increased meal size also increases gastric evacuation time in elasmobranchs (Sims et al., 1996, Bush and Holland 2002). In lemon sharks (Negaprion brevirostris), initial processing of prey occurred faster when meal size increased, but total gut transit time also increased, suggesting that the rate of digestion remained constant (Wetherbee and Gruber 1990). Our results agree with those of Wetherbee and Gruber (1990) because motility during the 7 h following ingestion increased with meal size but total motility during the 24 h following feeding did not. Based on our results, we hypothesize that optimum gastric digestive efficiency in blacktip reef sharks occurs when meal size is % of BW. Daily ration has not been measured in blacktip reef sharks, but for other carcharhinid sharks it has been calculated as approximately 1 2 % of BW day -1 (Wetherbee et al., 1990). The observed decrease in gastric motility at high ration levels may be due to stomach fullness reducing stomach contractions and consequently mixing. Gross conversion efficiency (the efficiency by which ingested prey items are converted into predator tissue) in elasmobranchs is thought to decrease at high ration levels (Cortes and Gruber 1994, Duncan 2006). Optimal gross conversion efficiency was achieved at relatively high ration levels (e.g. 5.1 % BW/ per day for scalloped hammerhead sharks, Duncan 2006), but those studies were conducted using juvenile animals with higher mass specific metabolic rates than adults. We found that a large proportion of the variability in gastric motility could be attributed to meal size 50

52 and stomach temperature (at least for meals of mackerel), but we did not measure surface area of prey items or changes in seawater dissolved oxygen concentration, both of which can influence motility (Schurdak and Gruber 1989, Mayer 1995). The results of the FFT suggest that diel changes in gastric motility exist regardless of whether the sharks fed. The 23.4 ± 1.8 h periodicity in motility is most likely a result of the diel fluctuations in stomach temperature, which in turn are related to daily fluctuations in ambient water temperature. Motility was highest in the afternoon when water temperatures were also highest and lowest during the early morning hours when water temperatures were also lowest (see stomach temperature data in figure 2). Based on the data from one fasting shark that showed lower ph values during early morning hours (between 0600 and 0800), we hypothesize that blacktip reef sharks preferably forage during periods (or areas) of lower temperature (in our study such conditions occurred in early morning, 6 8 AM) although they may feed opportunistically at all times of the day. As such, the period of gastric accommodation (low motility) following feeding coincides with periods of low temperature, with increased gastric motility occurring during periods of increased temperature. Our sharks also showed a periodicity in motility with a frequency of 2.0 ± 0.3 h, which, because this cycle was absent from sharks that were fasted, may represent regular periods of stomach contractions involved with mixing stomach contents and passing chyme into the small intestine. In vertebrates, gut motility during the interdigestive state (fasting) is characterized by migrating motor complexes (MMC), which consist of periods of quiescence (phase I), periods of irregular single contractions (phase II), and periods of strong contractions (phase III, Mayer 1995, Holmgren and Holmberg 2005). The interdigestive state in blacktip reef sharks was not 51

53 characterized by long periods of quiescence, nor was there an obvious transition between phase II and III, which are similar to the results found for rainbow trout (Olsson et al., 1999). In conclusion, gastric digestion in blacktip reef sharks appears to be a function of abiotic and biotic variables. While foraging behavior (and subsequent optimal foraging theory) is a function of the tactics used to capture prey, it also may optimize digestive efficiency or energy extraction from prey items (Hume 2005). By quantifying the effects of prey type, meal size, and stomach temperature on gastric digestion in sharks under semi-natural conditions, we can make predictions of foraging strategies in the field. Subsequent studies will aim to quantify gastric processes in free-ranging sharks in the field to test these hypotheses. 52

54 Chapter III A new acoustic ph transmitter for studying the feeding habits of free-ranging sharks ABSTRACT Little is known about the feeding habits of large free ranging fish, due in large part to lack of an appropriate technique for quantifying feeding variables. A previous study demonstrated that changes in gastric ph can be used as a proxy for feeding events in free-ranging sharks. Here we describe the development of a new acoustic ph transmitter to remotely measure gastric ph in sharks in the field. The transmitter consists of a dual sensor (ph and temperature) continuous pinger, and was tested in captive adult blacktip reef sharks (Carcharhinus melanopterus). The transmitter was retained in the shark s stomach from between 5-12 d. The empty stomach had a low ph (1.6 ± 0.2) and feeding induced a rapid increase in gastric ph, which was clearly distinguishable from baseline levels. Meal size showed a significant linear relationship with the magnitude of the ph changes. Measurement accuracy of the ph transmitter ranged from , although resolution of the VR100 receiver was 0.1 units. The ph transmitter can be used to determine when free-ranging sharks in the field are feeding and hence quantify feeding chronology, frequency and daily ration. INTRODUCTION Sharks are widely thought to play an important role in structuring marine communities but to date most evidence for this comes from analyzing stomach contents 53

55 of dead sharks and laboratory experiments to determine gastric evacuation rates of juvenile sharks (Wetherbee et al., 1990, Cortes 1997). Stomach content analyses traditionally involve sacrificing a large number of animals in order to achieve adequate sample sizes and this is increasingly undesirable because shark populations in many areas are already in rapid decline due to over-harvesting (Cortes 1997, Baum et al., 2003). Yet in order to better understand the full impact of shark predation on marine ecosystems, and to predict what may happen if shark populations are depleted, we need to know what sharks eat, how often they feed and how much they typically consume (Wetherbee et al., 1990, Papastamatiou and Lowe 2004). The stomach is the first site of digestion in vertebrates, and is responsible for the breakdown of prey items into semi-liquid chyme, which it achieves by secreting concentrated hydrochloric acid (HCl) and the inactive enzyme pepsinogen. The HCl breaks down prey hard parts and also converts pepsinogen into pepsin, a proteolytic enzyme (see Papastamatiou and Lowe 2004, 2005). Due to these physiological principals, a specific change in gastric ph should occur after a shark consumes a prey item. Studies with captive sharks using autonomous ph data-loggers have shown that by measuring gastric ph continuously, time of feeding can be determined and meal size estimated (Papastamatiou and Lowe 2004, 2005). However, the ph data-loggers used in the captive studies are not practical for field studies because the data-logger must be retrieved for the data to be collected. Here we describe the development of an acoustic ph transmitter, which will enable gastric ph to be measured remotely in free-ranging sharks, hence allowing the technique to be applied to the field. 54

56 MATERIALS AND METHODS Study site We maintained captive adult blacktip reef sharks (Carcharhinus melanopterus) in a pen situated in a lagoon at the Hawaii Institute of Marine Biology (HIMB). The water in the pen is tidally flushed and habitat consists of coral rubble, coral ledges, and sand with a maximum depth of 3 m. Sharks used for experiments were moved through a gate to an adjacent experimental pen (10 x 20m), where they were acclimated for a week or until they resumed normal feeding. No more than two sharks were maintained in the experimental pen at any one time. ph transmitter We measured gastric ph using a ph/temperature dual sensor continuous pinger (Vemco ltd, Nova Scotia). The sensor consists of a glass micro-electrode, and a reference electrode with a free-diffusion liquid junction connected to a piston-driven reservoir of electrolyte (Peters 1997a). The piston ensures a continuous supply of fresh electrolyte is passed over the reference electrode, greatly reducing measurement drift over a 16 d period, or until the reservoir runs out of fluid (Peters 1997a, b). The reference electrode was coupled to a transducer, a 3.6 V lithium battery, and all electronics were encased within a titanium shell (total length of ph transmitter: 17 x 2.5 cm, mass in air = 170 g, nominal battery life 50 d, fig.7). A temperature sensor encased in the electronics also recorded temperature. Both ph and temperature recordings were converted by the transducer into an acoustic signal with a carrier frequency of 60 khz and a power output of 157 db. The transmitter was a two channel 55

57 Figure 7. Acoustic ph transmitter, with transducer and ph electrode labeled. 56

58 10 cm Transducer ph electrode 57

59 continuous pinger, which transmits a continuous sequence of pings with three time intervals; a fixed interval of 1150 ms, followed by a signal for temperature and ph respectively. Changes in either temperature or ph result in a change of the interval of the emitted signal. The sensor measures ph between 0 9 units, and temperature between C. Due to the large amount of drift associated with measuring ph, the sensor had to be calibrated before every use. We calibrated the sensor using NBS standard ph buffers (1.68, 4.01, 6.86) and programmed the calibration parameters into a Vemco VR100 receiver. We deployed an omni directional hydrophone connected to the VR100 receiver, in the experimental pen. The VR100 recorded and stored all data received and were downloaded daily to a laptop computer. To insert the ph transmitter, we restrained sharks on a stretcher, inverted them to induce tonic immobility (a trance like state), and anaesthetized them with a 0.15 g/l solution of MS-222. We then inserted a lubricated PVC pipe through the mouth into the stomach and dropped the transmitter through the pipe with the sensor pointing towards the caudal fin. The shark was then revived, released and fed meals of mackerel (Scomber spp.) over the next 5 to 12 d. After each shark regurgitated the ph transmitter, we recalibrated the electrode in ph buffers, and used the ph drift model described by Peters (1997b) to estimate the uncorrected errors associated with drift of the electrode during deployment. In order to quantify the affect of meal size on gastric ph changes, we determined the area underneath the feeding induced change in the ph-time curve using ArcView (ver. 3.2). We then used multiple regression analysis to determine the influence of stomach temperature and meal size (expressed both in g and percentage body weight 58

60 %BW) on the area under the ph-time curve. Temperature and meal size were also set as an interactive term. RESULTS We deployed the ph transmitter in three blacktip reef sharks (Total Length (TL) = ± 7.6 cm, mass = 23 ± 3.6 kg,) for periods ranging from 5 12 d. During deployment, ph error within the range measured in the stomach was between (Table 3), although resolution of the VR100 receiver was 0.1 units. However, drift of the ph electrode was not linear as error ranged up to 0.92 in the ph 6 range, for one shark. Stomach ph averaged 1.6 ± 0.2 (± 1SD) during periods of fasting but increased rapidly to a peak of 4.1 ± 0.8 following ingestion of food, and then declined more gradually back to baseline levels (Fig.8). Multiple regression analysis revealed that meal size (g) influenced area under the ph-time curve (p = 0.001, t = 3.36), but stomach temperature had no significant affect (p = 0.881, t = -0.15). When analyzed independently, meal size had a significant influence on area under the ph time curve when expressed as % BW (p = 0.01, r 2 = 0.52, F = 9.89), and in g (p = 0.001, r 2 = 0.70, F = 20.9, Fig.9). DISCUSSION The ph transmitter provides viable data on changes in gastric ph for up to 12 d. Measurement drift ( ) was generally low within the typical range of ph values observed in the blacktip stomach (ph 1-4). Custom made software can be used to 59

61 Table 3. Summary information for blacktip reef sharks used in experiments. Error analysis for the ph electrode is also given over three hypothetical ph values using the uncorrected ph drift model described by Peters (1997a). Shark # Total length (cm) Mass (kg) Sex Deployment duration (d) ph drift F M F

62 Figure 8. Continuous measurements of gastric ph and temperature in three freeswimming blacktip reef sharks (Carcharhinus melanopterus, a-c equivalent to shark # 1-3). Data was obtained using a dual sensor ph/temperature transmitter. Arrows point to feeding events and meal size is expressed as a percentage body weight.? indicates a meal of unknown size. 61

63 10 a 30 ph Temperature (C) ph b Time (d) ? Temperature (C) ph c Time (d) Temperature (C) Time (d) 0 62

64 Figure 9. Linear regression analysis between meal size and area under the ph-time curve from blacktip reef sharks, Carcharhinus melanopterus. Meal size is expressed as % BW (a) and in g (b). Data was obtained using a ph/temperature transmitter. 63

65 1.6 a 1.2 Area 0.8 y = 0.79x , R 2 = b Meal size (%BW) 1.2 y = 0.004x , R 2 = 0.70 Area Meal size (g) 64

66 interpolate and correct for drift of the electrode if the transmitter is recovered (see Peters 1997a, b), although such recoveries are unlikely in field studies of sharks. The sharks used in the present study could only dive to a maximum depth of 3 m, and animals diving to greater depths may cause additional drift of the electrode. The flow rate of electrolyte out of the capillary junction of the reference electrode is crucial to the performance of the electrode, and suboptimal flow rates or electrolyte depletion, will increase measurement drift. Therefore, a larger electrolyte reservoir could increase the length of time over which the electrode can accurately measure ph. The current ph transmitter has a depth rating of a 100 m, but embedding all electronic components in epoxy will greatly increase the depth rating (Peters 1997b). Regardless of measurement accuracy, feeding events were clearly distinguishable from background variation in ph. Feeding resulted in a rapid increase in ph followed by a more gradual decrease to baseline levels. The rapid increase in ph is caused by seawater and the prey items buffering the small amounts of gastric acid present in the stomach during fasting (Papastamatiou and Lowe 2004). The presence of prey items in the stomach then causes distention of the mucosa wall and the release of secretagogues, all of which increase gastric acid secretion and the subsequent re-acidification of stomach contents (Papastamatiou and Lowe 2004, 2005). As meal size increases, a greater mass of prey (most of which are weakly acidic or neutral in ph) is in the stomach, and hence the time taken for gastric contents to re-acidify also increases. The diet of blacktip reef sharks in the field consists primarily of reef fish (teleosts), crustaceans, cephalopods, and in some locations reptiles (Lyle and Timms 1987, Cortes 1999). Although daily ration has not been determined for this species, for 65

67 other carcharhinids it has been estimated at approximately 1-2 % BW/day (Wetherbee et al., 1990), within range of the linear regression calculated in this study, relating meal size to the postprandial changes in gastric ph. Extrapolating the regression to zero on the y- axis, predicts that the smallest meal size estimated using the transmitter would be 25 g. However, some caution will be required when applying the technique to the field, as gastric evacuation rates (and presumably post-prandial changes in gastric ph) are thought to be a function of prey lipid content (e.g. Anderson 1999). The prey used in the current experiments (mackerel) most likely have higher lipid contents than prey consumed by blacktip reef sharks in the field, potentially underestimating meal size (high lipid prey will have longer gastric evacuation times). The changes in gastric ph and the linear response of time taken to re-acidify stomach contents, suggests that feeding chronology, frequency and meal size can be determined in free-ranging sharks. Although other similar sized marine organisms may differ in their gastric physiology, it is likely that the transmitter can also be used with large teleosts, marine mammals and birds. The primary factor limiting application of the technique is the size of the species in question, as is it is unlikely than an animal smaller than 1 m will tolerate the transmitter. However, size of the transmitter can be reduced by selecting for a smaller transducer and battery (also reducing life span of the unit and signal strength). 66

68 Chapter IV Scale-dependent effects of habitat on movements and foraging strategies of blacktip reef sharks, Carcharhinus melanopterus, at Palmyra atoll: a predator dominated ecosystem ABSTRACT The effects of habitat on the ecology, movements, and foraging strategies of marine apex predators are largely unknown. We used acoustic telemetry to quantify the movement patterns of blacktip reef sharks (Carcharhinus melanopterus) at Palmyra Atoll National Wildlife Refuge. Sharks had relatively small home ranges over a time-scale of days to weeks (0.55 ± 0.24 km 2 ), and showed strong site fidelity to sand-flat ledges within the west lagoon over a 3 year period. Sharks showed evidence of diel and tidal movements, and utilized certain regions of the west lagoon disproportionately. There were ontogenetic shifts in habitat selection, with smaller sharks showing greater selection for sand-flat habitats, and pups (total length cm) utilizing very shallow waters on flats, potentially as nursery areas. Adult sharks selected ledge habitats and had lower rates of movement when over sand-flats and ledges than they did over lagoon waters. Fractal analysis of movements showed that over periods of days, sharks used patches that were 3 17 % of the scale of their home range. Repeat horizontal movements along ledge habitats consisted of relatively straight movements, which theoretical models consider the most efficient search strategy when forage patches may be spatially and temporally unpredictable. Although sharks moved using a direct walk while in patches, they appeared to move randomly between patches. Micro-habitat quantity and quality has large effects on blacktip reef shark movements and path structure, which has 67

69 consequences for the life-history characteristics of the species and potentially the spatial distribution of behaviorally mediated effects on lower trophic levels throughout the Palmyra ecosystem. INTRODUCTION Apex predators are thought to exert top-down control on marine ecosystems through both density and trait-mediated interactions (Bascompte et al., 2005, Preisser et al., 2005). While ecological processes require an understanding of the diet of the predator, of equal importance is an understanding of activity patterns, habitat utilization and foraging strategies. Foraging theory predicts that animals will select habitats which provide the greatest return in some form of currency such as prey encounter rate (Stephens and Krebs 1986). However, for many animals it is difficult to distinguish habitat selection for foraging purposes from those associated with mating or predator avoidance. Most adult apex predators do not have to invest much energy into antipredatory behavior, and since most reproductive behavior is seasonal, it is possible to separate foraging from these other behaviors. Ontogenetic shifts in habitat selection can further contribute to an understanding of why animals select the habitats they use. Many species of shark are considered apex predators but there may be large geographic differences in their ecological significance (e.g. Stevens et al., 2000, Bascompte et al., 2005). The current worldwide decline in many shark populations further highlights the importance of understanding their ecological significance in multiple habitats and locations (e.g. Stevens et al., 2000). While shark movements and habitat utilization have been studied in several locations (e.g. McKibben and Nelson 68

70 1986, Rechisky and Wetherbee 2003) there are precious few studies that have quantified fine spatial scale habitat selection in relation to foraging strategies (Morrissey and Gruber 1993b, Heithaus et al., 2002, 2006). There are presently no studies that have quantified both high spatial and temporal scale movement patterns for any shark species. The blacktip reef shark (Carcharhinus melanopterus, Quoy and Gaimard 1824) is a common shark species found on coral reefs of the Indo-Pacific (Compagno et al., 2005) and qualitative observations suggest that they occupy shallow reefs and sand-flats at both atolls and high islands (Hobson 1963, Stevens 1984). Blacktip reef sharks are one of the most abundant apex predators at many atolls (Stevens 1984, Compagno et al., 2005), although human impacts have reduced their numbers at many locations (e.g. Sandin et al., 2008). Dietary analysis suggests that C. melanopterus are tertiary predators and feed primarily on teleosts, crustaceans, cephalopods, and in some areas reptiles (Cortes 1999). These factors raise the possibility that blacktip reef sharks could exert top-down control on many coral reef ecosystems. However, currently no detailed analysis of blacktip reef shark movement patterns, space utilization and habitat selection exists. Palmyra Atoll is part of the Line Island chain in the central Pacific Ocean, and has been a U.S. National Wildlife Refuge since 2001, essentially making the atoll a no-take marine reserve. As a consequence of reduced human impacts, a healthy population of fish apex predators exists at the atoll, making up over 65 % of the total fish biomass (Sandin et al. 2008, DeMartini et al., 2008). Diver observations show that blacktip reef sharks are the most abundant predator in the inner lagoons and sand-flats at Palmyra (Hobson 1963, Friedlander et al. 2007, DeMartini et al., 2008). However, the ecological impacts of blacktip reef sharks are partially a function of which habitats the sharks select 69

71 for, and how they behave in such habitats. Knowledge of habitat selection is particularly crucial for understanding predator-dominated coral reef ecosystems, where the behavioral response of prey to predators appears to largely dictate resulting ecosystem trophic structure (Knowlton and Jackson 2008). We used acoustic telemetry to quantify the movement patterns, habitat utilization and foraging strategies of blacktip reef sharks at Palmyra. Our specific objectives were: 1) determine the degree of site fidelity shown by the sharks to sand-flats within a lagoon at Palmyra Atoll over different temporal and spatial scales 2) determine if there are any diel or tidal changes in movement patterns, 3) quantify the selection or avoidance for microhabitats in the lagoon over the scale of hours to days, and 4) use fractal analysis to quantify the movement path structure and subsequent foraging strategies of blacktip reef sharks at Palmyra over short time periods (hours to days). MATERIALS AND METHODS Study Site Palmyra Atoll (N 5 53, W ) is part of the Line Island chain located just north of the equator (Fig.10). Two primary lagoons (west and east) are linked by a small channel, while a larger channel links the west lagoon to the outer reefs (Fig.10). The lagoons have a maximum depth of 50 m, with a mud/sand substratum causing low water visibility, while the outer fore-reefs are characterized by steep slopes with high coral cover and high visibility. Due to Palmyra s location in the Inter-tropical Convergence Zone, the atoll receives up to 500 cm of rainfall per year, and consequently terrestrial habitat is largely rain-forest (Fig. 10). In addition to Palmyra s refuge status, only a small 70

72 Figure 10. a) Map of the Line Islands relative to the central Pacific Ocean. b) Islands of the Line Islands including Palmyra atoll. c) VR2 positions and detection radii within Palmyra Atoll. VR2 s in the west lagoon are Banjos (B), Eddies (E), Nursery (N), Airport (A), and Midchannel (M), and in the east lagoon, Sixes (S), Cookies (C), and Downeast (D). The location of Banjos (circle), Nursery (triangle) and the main Channel (Ch) sand-flats are also shown. Stars in c) show known locations of blacktip reef shark pups. 71

73 72

74 crew of up to 17 refuge staff and scientists inhabit the atoll, hence human influences are maintained at a low level. Active tracking Blacktip reef sharks were attracted to the Banjos, Nursery or Channel sand-flats using squid bait (Fig.10). We then concealed a V13 acoustic transmitter (dimensions 13 x 30 mm, carrier frequencies khz, Vemco ltd., Nova Scotia) in a piece of squid, and allowed one of the sharks to voluntarily consume the transmitter containing bait. To facilitate longer tracks, we also caught some individuals and surgically implanted the transmitters into the body cavity. In those cases, we caught sharks on barbless hooks and brought them alongside the boat where they were restrained, inverted, and placed in tonic immobility; a trance like state. A small incision was then made through the shark s abdominal wall and a transmitter was implanted into the body cavity. The wound was closed with a single suture and the shark was released. We waited a minimum of 48 h before initiating tracks of sharks that were surgically fitted with transmitters, to remove the influence of surgery on movements. A PVC pipe of known length placed on the sand-flat enabled us to estimate shark total length in instances where the shark was fed a transmitter. For sharks fed transmitters, we discarded the first 2 h of data so as to remove the influence of feeding and odors on movement patterns. Continuous tracking was conducted using a kayak technique (Meyer and Holland 2001) and an RJE PRS 275 (RJE International ltd., Irvine, California) handheld underwater receiver, which enabled us to track the sharks in very shallow water. All sharks were tracked continuously during daytime hours with GPS positions taken every 15 min. During tracking we maintained a minimum m distance from the shark, but when GPS positions were taken we 73

75 would move to the location where the shark had previously been, so that we could quantify habitat use. If the shark was over a sand-flat, we were able to determine the exact location of the shark because the shallow water over the flats enabled us to visually track the shark and determine which habitat it was occupying. If the shark was over a ledge or in deeper lagoon waters, we would get an accurate fix by positioning the kayak until the acoustic signal strength was the same in all directions, indicating that we were directly above the animal (ground zero). Notes were taken while tracking to indicate which habitat sharks occupied when positional fixes were obtained, to ground-truth the GPS data. Due to safety regulations, continuous tracking could not be conducted at night. During the night we either; obtained single location checks every two hours, or we tracked continuously for one hour, every two hours. We estimated positional accuracy of shark locations to be ± 7-8 m (based on GPS accuracy). Nursery delineation Neonate and young-of-the-year (YOY) blacktip reef sharks were only observed in very shallow sand-flat habitats, close to shore. As it relates to ontogenetic shifts in habitat use, and because blacktip shark pups were too small to carry acoustic transmitters, we sampled these sharks at locations where they were aggregating. Sharks were caught using a 30 m long seine net that was positioned perpendicular to the shoreline. We then herded the sharks into the net where they were measured, sexed, weighed, and released. Data Analyses Home range 74

76 All spatial analyses of movement data were conducted in ArcView GIS (ver. 3.2) layered over geo-referenced IKONOS images of Palmyra Atoll. We calculated two metrics as a proxy for shark home range. The Kernel Utilization Distribution (KUD, Worton 1989) is a probability distribution that represents the area where there is a 95 % and 50 % chance of finding the individual tracked. The 95 % KUD is considered a measure of the overall home range of the animal, while the 50 % KUD is more representative of the area of core use (e.g. Heupel et al., 2004). A Minimum Convex Polygon (MCP) is the area of a polygon formed by connecting the outer position fixes of an animal s movements. Both estimates were calculated using the Animal Movements extension with ArcView GIS ver. 3.2 (Hooge and Eichenlaub 1997). Home range was only calculated for sharks that were tracked for a minimum of 24 h continuously, so as to include at least one complete diel cycle. Areas of the KUD and MCP which extended onto land were manually removed. Multiple regression analysis was used to determine the influence of shark total length (TL) and water temperature on 95 % KUD area. As a quantitative measure of the shape of the sharks home range, we determined the Index of Eccentricity (ECC), ECC = l / w, where l = maximum length of the animals activity space and w = maximum width of the activity space. A circular activity space will produce ECC = 1, while ECC values greater than 1 indicate an asymmetrical shaped activity space (Morrissey and Gruber 1993a, Rechisky and Wetherbee 2003). We calculated two measurements of site fidelity for blacktip reef sharks. The Linearity Index L i = (F n F 1 ) / D, where F n F 1 is the distance between the first and the last location fixes of the shark, and D is the total distance traveled by the animal. An animal moving in a nomadic fashion should have L i approach 1, while an animal 75

77 exhibiting strong site fidelity should have L i approach 0. We also calculated the Index of Reuse (IOR), IOR = [OV(A1+A2)] / (A1+A2), where [OV(A1+A2)] is the area of overlap between two daily activity spaces (e.g. 12 h) and (A1+A2) is the total area of both daily activity spaces. IOR = 1 indicates 100 % re-use of an area on a daily basis (site fidelity) while an IOR = 0 indicates 0 % overlap in area (nomadic behavior). In order to determine if blacktips showed daily shifts in core areas, we also calculated IOR values between daily 50 % KUD areas. To test for diel behavior, we compared daytime and nighttime activity space size (using the MCP measurement) for each shark using a Student s t-test. We also determined Rate of Movement (ROM) for each shark during day and night periods, with ROM defined as the distance moved by the shark between two points, divided by the time taken to swim between the points. To evaluate the effects of diel and tidal periods on shark ROM, we categorized all ROM values into a) day vs. night and b) tidal period (high slack, low slack, flood, ebb). We tested diel and tidal effects simultaneously by dividing ROM data into 8 diel and tidal groupings and using a one-way ANOVA with a Tukey- Kramer aposteriori test. ROM data were converted to Body Lengths / min (BL min -1 ) to control for shark total length. Data were square root transformed to meet the assumptions of normality (Shapiro Wilk W test). ROM data represents speed over ground, hence an animal swimming in a straight line will have higher ROM than an animal foraging over a small area (see Phillips et al., 2004). Habitat utilization The inner lagoons at Palmyra consist of four microhabitat types: sand-flats, reef ledges, deeper sand-flats, and lagoons. Sand-flats are extensive areas with water depth < 76

78 2 m and benthic habitat is sand/coral rubble, while deeper sand-flats are areas in the central lagoons where water depth is 2 5 m. Lagoons comprised all other regions within the atoll where water depth exceeds 5 m and benthic substratum is primarily fine sand/mud. Ledges are located at the boundary between sand-flats and lagoons. Based on results of fractal analysis (see below) we considered ledge habitats to include the area within 20 m of either side of the drop off. For each shark, we determined the number of position fixes that occurred in each of these habitat types. We then calculated the area of each of these habitats in the west lagoon, as a percentage of the total area, using the georeferenced IKONOS image of Palmyra. We used a Chi-squared test to determine if habitats utilized by the sharks differed significantly from expected based on overall habitat available. We then used the modified Strauss linear index of food selection, L = r i - p i, where L is the habitat selection value, r i is the percentage use of habitat i, and p i is the percent availability of habitat i (Morrissey and Gruber 1993b). Values of L > 0 suggest statistical selection for a particular habitat while values of L < 0 suggest avoidance of a habitat. We determined habitat selection values for all sharks combined, but then performed least-squared linear regression analysis between individual shark length and habitat selection values for sand-flat, ledge, and lagoon habitats. In order to evaluate the effect of habitat type on speed over ground, we also quantified ROM for each shark while they were moving over ledge, sand-flat and lagoon habitats and utilized an ANCOVA test using ROM as the dependent variable, habitat as an independent variable and shark total length as a covariate. A Tukey s test was then used to determine the location of pair-wise differences. For habitat analysis ROM data were log transformed to meet the assumptions of normality. 77

79 Fractal analysis A fractal dimension is a measure of tortuousity of a movement path, and can range from 1 for a straight line to 2 for a path so tortuous that it completely covers a plane (Nams 1996, 2005). Fractal measures of animal movements are generally scaledependent as the tortuousity of a path will vary based on the scale at which it is viewed. Therefore, by examining how the fractal value (D) varies with scale for a movement path, we can quantify the scales at which the animal views its environment and also detect patch use (Nams 2005). For a more detailed description of the use of fractal analysis in animal ecology see Nams (1996, 2005), and Doerr and Doerr (2004). However, a caveat of using fractal analysis to describe animal movements is that if the animal is moving using a correlated random walk (CRW), then the changes in Fractal D with scale may not be caused by a change in tortuousity with scale (Nams 1996, 2005). A CRW occurs when each step of an animal s movements relate to the following rule: θ i = θ i-1 + έ i, where θ i is the direction of step i and έ i is a random angle drawn from a normal distribution. Therefore, turning angles are independent of previous turning angles. To determine if sharks were moving using a CRW, we calculated the CRW Diff test described by Nams (2006a), CRW Diff = 1 k k n = 1 n 2 R 2 n l E ( R 2 2 n ) E ( R 2 n ) where 2 R n represents the observed mean (net distance) 2 for each number of n consecutive moves, E is the expected mean (net distance) 2 according to the CRW equation described by Kareiva and Shigesada (1983), l is the mean step length, and k is the turning angle concentration (Nams 2006a). CRW Diff is therefore a measure of how net displacement of an animal s movements varies from that predicted by a CRW. If CRW Diff > 0 then the animal shows greater net displacement 78

80 than a random walk, while if CRW Diff < 0 the animals movements are more constrained than a CRW. Two fractal measures were used to analyze the movement patterns of blacktip reef sharks. Fractal Mean was used to estimate an overall fractal D value for each blacktip reef shark, by using the traditional divider method (Doerr and Doerr 2004). A range of dividers are used to determine the length of a movement path, with path length decreasing as divider size increases (Mandelbrot 1967, Doerr and Doerr 2004). A log-log plot of divider size versus path length is then generated, which yields a line with slope 1- D, and can be described by L(G) = kg 1-D, where L(G) is path length, k is a constant and G is divider size. If the movement path is more tortuous and has a greater number of turns, then the slope of the line will also increase, and the path has a higher fractal D. Fractal D also incorporates replication by measuring path length twice for each divider size, by running the dividers both forward and backward along the path, which reduces bias associated with previous Fractal D measures (Nams 2006). To measure the change in fractal D with scale, we used the VFractal estimator described by Nams (1996). VFractal calculates the fractal values based on the turning angle between consecutive locations, and its associated error estimator (Nams 1996). We used the VFractal estimator in Fractal ver. 5.0, for divider sizes ranging from m. The 95 % confidence intervals were calculated using a bootstrapping procedure, which randomly selects turning angles from the movement path to calculate VFractal, with 1000 replicates (Nams 1996). To detect patch use, we determined the correlation in tortuousity between adjacent path segments for a range of divider sizes. If the divider size is below the size of a patch 79

81 used by the animal, then it is likely that consecutive path segments will be either inside or outside the patch hence the tortuousity correlation between adjacent path segments should be positive. As divider size approximates patch size then it s likely that one path segment will be inside the patch (with high tortuousity) while the adjacent segment is outside the patch (with low tortuousity), hence the correlation should be negative. Therefore, a positive correlation followed by a negative correlation is indicative of patch use and size (Nams 2005). If there is no patch use, then there should be no correlation between patch segments and the correlation should be zero regardless of whether the animal is moving in a random or directed manner (Nams 2005). All fractal measures and correlation statistics were calculated in Fractal ver Long term movements To quantify longer term site fidelity of blacktip reef sharks to different reef flats, we established an array of 8 omni-directional automated underwater acoustic receivers (model VR2, Vemco, Nova Scotia) throughout the west (five receiver s) and east lagoon (three receiver s, Fig.10). The receivers were moored to the mud/silt lagoon substratum in depths of m, with the receivers suspended m below the surface. We surgically implanted 8 blacktip reef sharks within the west lagoon with Vemco V8SC-2L transmitters (8mm diameter x 20 mm length). Each transmitter produces a unique pulse code that can be detected by the VR2 receivers when a tagged shark is within range (300 m) of the receiver. These transmitters had a nominal battery life of one year. VR2 receivers were retrieved, downloaded and redeployed every 4-6 months. We determined the number of detections at each VR2 for each shark as a percentage of the total number of detections. To determine if sharks were 80

82 disproportionately using certain areas more than others, we compared proportion of detections between the different VR2 receivers. The data did not conform to the assumptions of parametric statistics despite transformation and a non-parametric Kruskal- Wallis Rank Sum Test was utilized. To examine temporal patterns of movement, we used a Fast Fourier Transformation (FFT). An FFT decomposes time series data into component frequencies, and then searches the data-set for cyclical patterns. Sinusoidal patterns with dominant frequencies can be identified as peaks in a power spectrum. As such, FFT analysis can identify diel, tidal or seasonal patterns in animal movements (e.g. Meyer et al., 2007). We binned the number of VR2 detections in every hour for each day of the VR2 deployments and smoothed the data using a Hamming window before applying the FFT. A Hamming window reduces the effects of adjacent spectral components, which can potentially generate biologically meaningless frequency peaks. RESULTS Active tracking Home range We actively tracked 14 blacktip reef sharks (Total Length (TL) 100 ± 17 cm, mean ± 1SD) for periods ranging from 4 72 h, between February 2005 and September 2007 (Table 4). Although this represents continuous tracking times, we would also periodically re-locate sharks up to 14 days following the start of the track. Ten sharks were fed transmitters, while four animals had transmitters surgically implanted. There 81

83 Table 4. Blacktip reef sharks (Carcharhinus melanopterus) actively tracked at Palmyra atoll. TL, Total length. Tagging location indicates reef flat where sharks were tagged. Sharks were either fed transmitters or had them surgically implanted. Shark # TL (cm) Sex Hours tracked Days tracked Month Tagging location Tagging method Feb 05 Banjos Fed July 05 Banjos Fed July 05 Banjos Fed Mar 06 Banjos Fed Mar 06 Banjos Fed Mar 06 Nursery Fed Mar 06 Nursery Fed Nov 06 Banjos Fed M Nov 06 Banjos Implant M 4 1 Nov 06 Nursery Implant Nov 06 Banjos Fed Nov 06 Banjos Fed F 50 4 May 07 Channel Implant F 10 7 Sept 07 Channel Implant 82

84 was no significant difference in overall ROM between sharks that were fed transmitters (11.3 ± 8.7 m/min) and those that had them surgically implanted (11.4 ± 9.4 m/min, t = 1.97, p = 0.77). Sharks moved over a limited area, with repeated use of core locations on a daily to weekly basis (Fig.11). Home range estimates were small, with average 95 % KUD areas of 0.55 ± 0.24 km 2, and MCP areas of 0.33 ± 0.26 km 2, while the maximum linear dimension of the home range was 1.4 ± 0.3 km (Table 5). The 95 % KUD estimates were larger than MCP estimates for five of the six sharks examined, although the difference was not significant (t-test paired for means, t = 1.45, p = 0.76, Table 5). There was no effect of shark TL, water temperature, or the interaction term on 95 % KUD area (F = 0.11, p = 0.77). Blacktip reef sharks tended to have activity spaces which were asymmetrical and oblong in shape, as indicated by the high ECC values (4.8 ± 2.3, Table 5) and there was no influence of shark TL on ECC (F = 1.1, d.f. = 13, p = 0.30). The repeated use of core areas by blacktip reef sharks was also apparent based on the low L i values (0.121 ± 0.096, Table 4). The highest L i value (0.280) was for the shortest tracks (9 and 4 h), but much lower values were obtained from sharks tracked for longer periods. For example shark # 9 was tracked for 72 h and had L i = 0.007, and shark # 4 tracked for 48 h had L i = (Table 5). There was no influence of shark TL on L i (F = 0.002, d.f. = 13, p = 0.89). IOR values (0.19 ± 0.11) were lower than expected based on the low L i values (Table 5). However, the lower IOR values in general were due to low overlap in consecutive 50 % KUD activity spaces. The IOR between 50 % KUD for day 1 and 2 in blacktip # 9 was 0.005, for blacktip # 4 was 0.189, blacktip # 2 was 0.024, and blacktip # 13 was 0.0. When sharks were re-located on subsequent days, they were located within or near the original 50 % KUD, which resulted in low L i values. 83

85 Figure 11. Home range of six blacktip reef sharks at Palmyra atoll. Polygons are 95 % KUD s, and dots are shark locations. The size and sex (where known) of each shark is given in the figure. 84

86 85

87 Table 5. Home range and movement statistics for blacktip reef sharks (Carcharhinus melanopterus) actively tracked at Palmyra atoll. TL Total length, KUD 95 % Kernel Utilization Distribution, MCP Minimum Convex Polygon, Max dim. maximum dimension of home range, ECC Index of Eccentricity, IOR Index of Re-use, D Fractal value. Means and standard deviation (SD) are also given. Sharks with no values in cells, had insufficient data to detect patch size. Patch sizes as a percentage of home range length are also given. Shark # TL (cm) 95% KUD (km 2 ) MCP (km 2 ) Max dim (km) Linearity index ECC IOR D Patch size (m) Patch size/ home range length (%) , , Mean SD

88 Blacktip reef sharks did not exhibit any detectable diel shifts in activity space size or location (day 0.17 ± 0.15 km 2, night 0.14 ± 0.16 km 2, paired t-test for mean, t = 0.42, d.f. = 12, p = 0.68,). However, there were significant differences in ROM values when data was separated by diel tidal periods (ANOVA, F = 2.63, d.f. = 257, p = 0.012). Sharks swam with a greater speed over ground during ebb tides at night (18.1 ± 8.2 BLmin -1 ) compared with flood tides at night (8.5 ± 9.5 BLmin -1 ). Flood and ebb nighttime ROM values did not differ from any of the other categories. However, it should be noted that because not all sharks were continuously tracked during nighttime periods, the smallest ROM samples sizes were for nocturnal flood and ebb tides. Habitat utilization The observed use of habitats by sharks differed significantly from expected based on available area of each habitat type (X 2 = 16.1, p = 0.01). When data from all sharks were combined, high L values (selection) were obtained for ledge habitats (L = 0.59), while lower L values (avoidance) were obtained for sand-flat (L = -0.14) and lagoon habitats (L = -0.38, Fig.12). Sharks showed neither avoidance nor selection for deeper sand-flat habitats. As sharks increased in size, their selection for sand-flat habitats decreased (L decreased, F = 5.52, d.f = 13, p = 0.04, r 2 = 0.36, L = TL , Fig.12). There was no significant relationship between L for ledge or lagoon habitats and shark TL (F = 0.062, p = 0.81). Both shark TL (ANCOVA, F = 7.38, d.f. = 15, p = 0.017) and habitat type (F = 11.83, d.f. = 15, p = 0.001) influenced ROM, although there was no significant interaction effect on ROM (F = 0.82, p = 0.46). As sharks increased in size, ROM also 87

89 Figure 12. Habitat selection by blacktip reef sharks at Palmyra atoll. a) Habitat selection for all sharks combined (n = 14). L > 0 suggests selection for a habitat, while L < 0 suggests avoidance of a habitat. Observed use of habitat differed significantly from expected (p = 0.01). b) Relationship between shark total length and selection coefficient for sand-flat habitats (y = x , r 2 = 0.34, p = 0.04). 88

90 0.8 a L sand flats lagoon ledge deep flat 0.4 b 0.2 L (sandflats) Total length (cm) 89

91 increased, and sharks swam with the greatest speed over ground when over lagoon habitats, and the lowest when over sand-flats. Sharks moved slower over sand-flat habitats (7.6 ± 1.4 m/min) than they did over lagoon (16.8 ± 6.6 m/min, p = ) and ledge (11.7 ± 2.3 m/min, p = 0.022) habitats. Sharks had higher ROM when over lagoon rather than ledge habitats (p = 0.049). Fractal analysis Blacktips showed more constrained movements than predicted by the Correlated Random Walk (CRW) model (CRW diff = , d.f. = 11, p = 0). Therefore, fractal analysis was an appropriate technique for analyzing shark movements. The relatively high D values (1.25 ± 0.08) indicate that sharks had tortuous movement patterns, characterized by repeated back and forth movements along the reef ledges (Fig.11, Table 5). Fractal analysis suggests that sharks view their environment at a minimum of two different scales (Fig.13). When data from all sharks was combined, discontinuities in D existed at 15 67, and > 107 m (Fig.13). At scales between m movements appeared to be scale invariant, as there were no changes in D with scale. D started to slowly increase at scales > 67m and increased more rapidly at scales > 107 m. At scales > 400 m the confidence intervals were too wide for any conclusions to be made with regards to movement structure. However, fractal analysis of individual animals indicates that there is some intra-specific variability in behavior (Fig.14). Both changes in VFractal and correlation coefficients show that shark # 2 used patches at scales of m (Fig. 14a, b). The shark swam in relatively straight paths up to scales of 30 m, after which movements became more tortuous, especially at scales > 150 m. Shark # 9 swam 90

92 Figure 13. Changes in VFractal (D) with scale for movements of all blacktip reef sharks combined (n=13). Solid line is mean, while dashed lines are upper and lower 95 % confidence intervals. The x-axis is on a log scale. The hatched box shows the location of a two discontinuities in D. Numbers are x-axis values at location of the discontinuities. 91

93 Fractal D Scale (m) 92

94 Figure 14. Fractal analysis of blacktip reef shark movement patterns at Palmyra atoll. a) and b) are for a 70 cm TL (shark # 2)individual, while c) and d) are for a 110 cm female shark (Shark # 9). Upper panel shows change in VFractal with scale, while lower panel shows change in correlation in fractal values between adjacent steps. Solid line shows mean values while dotted lines show upper and lower 95 % confidence intervals. Striped rectangular bars show areas of discontinuity in VFractal (upper panel) or scales of patch use (lower panel). Scale values at these locations are given on the figure. The x-axis is on a log scale. 93

95 2.0 a 2.0 c Fractal D Fractal D b Scale (m) d Scale (m) Correlation Correlation Scale (m) Scale (m) 94

96 in a fairly direct manner up until a scale of 30 m after which a discontinuity and increase in D occurred, with progressively more tortuous movement paths at scales from m (Fig. 14c). Correlation coefficients show patch use at scales of m and m (Fig. 14d). In general, blacktip reef sharks in the west lagoon use patches at scales of m, m, and m, which approximate 3 17 % of the scale of their home range (Table 5). Nursery delineation In over 500 h of tracking and fishing for sharks on sand-flats, ledges, and lagoons, neonate and YOY sharks were only seen and captured on sand-flats very close to the shore-line (< 1 m), often in water no more than 10 cm deep. In these areas we caught 43 neonate and YOY blacktip reef sharks (TL 46 ± 5 cm, range cm, 25 females, 18 males, Fig. 10). These potential nursery areas were always located interior from the reef ledge, although it is unknown where the sharks went during extreme low tides (when sand-flats are exposed). Nursery locations only represent areas where we sampled. YOY were observed in areas where we did not sample, and they are always found in the same habitat type (very shallow water over sand-flats, close to the shoreline). Long term movements We deployed long term transmitters in 9 sharks (114 ± 10 cm TL, 5 males and 4 females, Table 6) between February 2004 and February Between February 2004 and October 2007, all 8 (100 %) of our receivers detected 7 of the 9 tagged sharks (78 %) for periods of d (Median 926 d, Table 6). There were significant differences in the percentage detections by each VR2 (Kruskal-Wallis, H = 19.84, d.f. = 7, p = 95

97 Table 6. Summary of acoustic monitoring data for 9 blacktip reef sharks (Carcharhinus melanopterus) tagged in the west lagoon of Palmyra Atoll with long-life acoustic transmitters. All sharks were tagged with V8 transmitters except for 29* which was tagged with a V16 transmitter. Capture location Transmitter code TL (cm) Sex Date deployed Date first detected Date last detected Overall detection period (d) Total detections Banjos F 20 Mar Mar Mar Banjos M 20 Mar Mar Jul Nursery F 22 Mar Mar 04 9 Jun Nursery M 22 Mar Mar 04 6 Oct Nursery F 22 Mar Nursery M 22 Mar Mar May Nursery M 22 Mar Mar Jan Banjos M 26 Feb Nursery 29* 123 F 24 Feb Feb Oct

98 0.006), with a greater proportion of detections of tagged sharks at the Banjos (median 33.5 %) and Airport (median 15.2 %) receivers than any of the other locations (Table 4). Sharks showed site fidelity to a small area as 81 ± 12 % of detections occurred at one core receiver for each shark (Table 7). Detected movements were mostly confined to the west lagoon, with only % of detections occurring in the east lagoon (median 0.1 %). Distinct seasonal changes in movements were only apparent in two individuals (29 %), which showed movements to the east lagoon (Fig. 15). Both sharks (one male, one female), made annual movements to the east lagoon starting in late December, over a three year period. The excursions were brief and occurred periodically over a two month period, with both sharks returned to the west lagoon daily after excursions. Spectral analysis showed evidence of diel and tidal effects (Fig.16). Five of the six (83 %) sharks showed 24 h peaks in the time frequency spectrum, and four (67 %) showed 12, 6 or 8 h peaks associated with tidal movements (Table 7). However, the spectral density of the peaks was low, indicating that diel or tidal behavior did not occur daily, and that there were periods of no detections. DISCUSSION Home range size and site fidelity Blacktip reef sharks at Palmyra Atoll appear to have relatively small home ranges over the scale of days to weeks. Coastal adult sharks in both tropical and temperate waters have significantly larger home ranges than the blacktips in the present study (McKibben and Nelson 1986, Holland et al., 1993, Rechisky and Wetherbee 2003). Blacktip reef sharks tracked at Aldabra Atoll, Indian Ocean, also showed limited 97

99 Table 7. Percentage of detections by VR2 receivers at 8 locations throughout the Palmyra lagoons, for 6 acoustically tagged sharks, and the period of dominant peaks from FFT analysis. Sixes, Cookies, and Downeast are all located in the east lagoon. Median and upper quartile (Q3) are also given. Shark Nursery Banjos Airport Eddies Midchannel Sixes Cookies Downeast Dominant peaks (h) , 12, , , , 6 Median Q

100 Figure 15. Seasonal movements of two blacktip reef sharks from the west to the east lagoon. For clarity, only the detections in the east lagoon have been shown. a) is a 101 cm male, b) a 129 cm F 99

101 00:00:00 a 20:00:00 16:00:00 Time 12:00:00 08:00:00 04:00:00 00:00:00 4/1/04 10/1/04 4/1/05 10/1/05 4/1/06 10/1/06 4/1/07 b Date 00:00:00 Downeast Sixes Cookies 20:00:00 16:00:00 Time 12:00:00 08:00:00 04:00:00 00:00:00 3/1/05 7/1/05 11/1/05 3/1/06 7/1/06 11/1/06 3/1/07 7/1/07 Date 100

102 Figure 16. Examples of long term movements of two acoustically tagged blacktip reef sharks. a) scatter plot showing movements for shark 62 and b) associated spectral analysis (FFT). The periods with dominant peaks in the FFT have been labeled. c) scatter plot for shark 56 and d) associated spectral analysis. Note the use of different scales on the y-axis for b) and d). 101

103 Time 00:00:00 20:00:00 16:00:00 12:00:00 08:00:00 04:00:00 a Spectral density b :00:00 4/1/04 10/1/04 4/1/05 10/1/05 4/1/06 10/1/06 Date Period (h) 00:00:00 c 20 d Time 20:00:00 16:00:00 12:00:00 08:00:00 04:00:00 Spectral density :00:00 4/1/04 8/1/04 12/1/04 4/1/05 Mid channel Cookies Banjos Airport Nursery Date Period (h) 102

104 movements, but moved up to 2.5 km in 7 h (Stevens 1984). Aldabra Atoll (34 km) is a much larger atoll than Palmyra (12 km) and sharks were tracked by attaching monofilament and surface floats to the dorsal fins, which could have affected their behavior (Stevens 1984). We found no effect of shark size and water temperature on home range size in blacktip reef sharks. Theory predicts that as an animal increases in size, energetic requirements and consequently area over which resources are obtained (home range) also increase (see review in Lowe and Bray 2006). While a number of studies have shown an ontogenetic expansion in home range with shark length (e.g. Heupel et al., 2004, Garla et al., 2006), only juvenile lemon sharks (Negaprion brevirostris) in the Bahamas have been shown to display a linear increase in home range size with shark length over the smaller size ranges ( cm PCL, Morrissey and Gruber 1993a) Blacktip reef sharks showed a high degree of site fidelity to the west lagoon, in particular, to core areas within the lagoon for periods over several years. While the active tracking indicated strong fidelity to the Banjos and Channel sand-flat ledges, acoustic monitoring data also showed that blacktip reef sharks consistently utilized these areas of the west lagoon for many years, although they occasionally make brief excursions to other locations within the west lagoon or to the east lagoon. Further evidence for strong site fidelity is the fact that the majority of detections (mean 81 %) for each shark were on one core receiver. Tag and recapture data of blacktips at Aldabra Atoll also indicated high site fidelity as 81 % of recaptures occurred within 1 km of the tagging location (Stevens 1984). Similar levels of site attachment have been seen in both juvenile and adult species of sharks from tropical islands and atolls (McKibben and Nelson 1986, Chapman et al., 2005, Garla et al., 2006) although those species tended to 103

105 move over a larger area than the blacktips in the present study. Our data provides the longest time-frame over which site fidelity to a small area has been quantified for any species of shark. Similarly sized coastal sharks from sub-tropical and temperate bays, show less site attachment and perform extensive seasonal migrations, which are most likely attributed to the much greater seasonal variation in environmental conditions in those areas (e.g. Rechisky and Wetherbee 2003, Heupel et al., 2004). The repeated use of ledge habitats suggests that blacktip reef sharks at Palmyra are able to meet most of their energetic needs in relatively small areas (at least on certain ledges). The sharks also showed daily shifts in 50 % KUD s within their home range, which has also been seen in juvenile lemon sharks and may be related to behaviorally mediated resource depletion (Morrisey and Gruber 1993a, Brown et al., 1999). Diel and tidal effects on behavior Several shark species show increased rates of movement and size of activity space at night suggesting nocturnal foraging (Nelson and Johnson 1980, McKibben and Nelson 1986, Garla et al., 2006). Blacktips at Palmyra show some degree of diel behavior, although there are intra-specific differences in the magnitude and consistency of the behavior between sharks. Although there are numerous explanations for diel behavior including feeding, predator avoidance, reproduction, and energetic advantages (Lowe and Bray 2006), it is unclear as to what factors regulate this behavior for blacktips at Palmyra. However, it is possible that the behavioral variation among individuals is a form of an Evolutionary Stable Strategy (ESS) reducing intra-specific competition in predator dominated ecosystems, where competition for resources may be strong. 104

106 Tidal stage has been shown to effect shark behavior in several locations, with individuals moving onto previously exposed sand or mud flats at high tide to forage (e.g. Nelson and Johnson 1980, Wetherbee et al., 2007). Both passive and active tracking suggest a tidal component to blacktip movements at Palmyra, although there were intraspecific differences in the magnitude of the response. Sharks had significantly lower rates of movement during the nocturnal flood tide than the ebb tide. The reduced rates of movements for the sharks at Palmyra with the flood tides, corresponds with the influx of cooler water and subsequent decrease in water temperature (up to 3 ºC, NOAA Coral Reef Ecosystem Division). There are three potential explanations for reduced swimming speeds during this time period: 1) reduced metabolic rates caused by lower temperature, 2) reduced swim speeds due to foraging in small patches, 3) decrease in swim speeds from reduced search behavior and foraging. Explanation 1) is unlikely due to the fact that average ROM halved when the tidal cycle switched from the nocturnal ebb to the nocturnal flood period, yet based on the Q 10 (as determined for other tropical species) ROM should have decreased by % (Carlson et al., 2004). Based on endogenous rhythms in gastric motility and ph in captive blacktip reef sharks, it was previously hypothesized that individuals would preferably forage during times of low water temperature, as the natural delay in gastric motility following feeding (gastric accommodation) would coincide with times of increased water temperature (Papastamatiou et al., 2007). The hypothesis would fit well with explanation 2, but presently either explanation 2) or 3) are plausible. Changes in rate of movement have also been shown to effect detection frequency by acoustic monitors (Topping et al., 2006), which may also explain the tidal peaks in the VR2 detections. Previous studies of 105

107 sand flat associated fishes at Palmyra (e.g., bonefish, jacks) have shown that these fishes invade sand flats during flooding tides, but leave during falling tides via discrete corridors (Friedlander et al. 2007). It is likely that, at times, blacktip movements may be linked with these tidally driven prey migrations. Finally, qualitative data also suggests that blacktip reef shark pups and YOY show tidally-mediated movements. These sharks occupy areas (shallow sand-flats) that are inaccessible during low tide, hence some movements correlated with tidal flow must exist (similar behavior is seen in juvenile lemon sharks, Wetherbee et al., 2007). Habitat utilization Although there was no influence of shark length on home range area, there were ontogenetic shifts in habitat selection, with smaller sharks showing stronger selection for sand-flat habitats. Sand-flat habitats are characterized by shallow waters and would provide small sharks with protection from larger predators. Neonatal and YOY blacktip pups were only found in very shallow water, very close to shore. These smaller sharks could potentially have a number of predators including adult blacktip reef sharks, gray reef sharks (Carcharhinus amblyrhynchos), tiger sharks (Galeocerdo cuvier), and large teleosts. While the use of coastal bays as nursery areas is well documented in elasmobranchs (e.g. Heupel et al. 2007), far less is known about the use of small scale nursery zones (on the scale of meters) at atolls and islands (Garla et al., 2006, Wetherbee et al., 2007). There should be strong selection for the utilization of shallow (safe) habitats by shark pups in predator dominated ecosystems. Similarly, juvenile lemon sharks select for shallow inshore mangrove habitats or tidal pools to obtain protection from predation by larger sharks (Morrissey and Gruber 1993b, Wetherbee et al., 2007). 106

108 Larger sharks showed a clear habitat preference for reef ledges, often spending their time patrolling back and forth along the reef ledge (also indicated by the relatively high fractal values and the oblong shaped activity spaces used by the sharks). Use of ledges as foraging sites appears to be a common feature of top level predators in both terrestrial and marine systems (e.g. Heithaus et al., 2006, Phillips et al., 2004). Ledge or edge habitat use has been seen in several elasmobranchs (e.g. Morrissey and Gruber 1993a, Rechisky and Wetherbee 2003, Heithaus et al., 2006), but only one other study has quantified ledge use (Heithaus et al., 2006). We have conducted dives on the steep ledges in the lagoons at Palmyra, and they appear to support a high abundance of potential prey items, so we propose that blacktip reef sharks either obtain a higher forage base over the ledges, or obtain greater encounter rates with prey (possibly prey moving off the flats). The reduced swim speeds over ledge and sand-flat habitats are most likely a consequence of the sharks foraging in patches (see below) in these locations. Although adult sharks spent less time than expected (based on available area) over sand-flats, all adult sharks made brief excursions onto the flats. The increased rates of movement over lagoon waters, is a consequence of straight-line swimming, suggesting that these habitats are mainly used to transit between ledges and sand-flats. Large tiger sharks are occasionally seen in the west lagoon, and it is also possible that adult blacktip reef sharks reduce predation risk by avoiding deeper lagoon habitats. Both active and passive tracking suggest that sand-flats within the west lagoon may differ in quality, as sharks showed strong fidelity to the Banjos, Channel, and Airport receiver areas, but much lower fidelity to the Nursery receiver even though these locations are only a few hundred meters apart. All sharks tagged and actively tracked at 107

109 the Nursery s ledge had left the area after 24 h, and could not be re-located there over several days. Between 0 4 % of detections occurred at the Nursery s receiver for acoustically tagged sharks, even though four of the six sharks were tagged on the Nursery s sand-flats. The Nursery s sand-flat does not have a coral ledge, unlike the other flats, and therefore supports a lower biomass of potential prey items (per. observations, unpublished data). Therefore the ledge at Nursery s, most likely represents an area of low habitat quality which may be driving the low level of site attachment shown by the sharks. Clearly, habitat quality is important for regulating the levels of site attachment, even over small spatial scales. The lack of movement of sharks from the west to the east lagoon is also striking and may further be a function of habitat quality. However, the east lagoon serves some purpose for the life history of at least some individuals as indicated by the highly seasonal and synchronous migrations by two individuals. The reproductive cycle of blacktip reef sharks varies by location and can occur either annually (Porcher 2005) or every other year (Stevens 1984). The reproductive cycle of blacktips at Palmyra is unknown, but it is possible that the seasonal movements observed for some individuals may be related to mating behavior. We were not able to quantify habitat use on the outer reefs at Palmyra, where conditions are very different from the inner lagoons. Although we observe blacktip reef sharks when diving on the outer reefs, the dominant predator is the grey reef shark, and therefore the ecological importance of blacktips may be reduced. However, there can be differences in behavior of sharks on ocean ledges versus lagoons (e.g. McKibben and Nelson 1986). 108

110 Movement path structure and hunting strategy Fractal analysis is a powerful tool in the study of animal movement paths although the majority of its application to date has focused on terrestrial animals (e.g. Doerr and Doerr 2004, Nams 2005). However, the technique is gaining popularity for use with marine animals and has been used to look at seasonal changes in movement path structure and to identify Area Restricted Search zones (Laidre et al., 2004, Tremblay et al., 2007). The overall movement paths of blacktip reef sharks at Palmyra could not be modeled with a correlated random walk, but instead they showed directed movements within patches, while moving more randomly between patches. Fractal analysis was able to detect patch use by blacktip reef sharks, with sharks using small sized patches (most of a scale between m, or approximately 3 17 % of the scale of the shark s home range length) on ledges and sand-flats. Movements within the m scale range are scale-invariant suggesting that sharks move using a directed walk while in patches most likely by orienting to ledge habitats (Nams 2006). The directed walk within patches appears to be a common behavior amongst all sharks, as indicated by the narrow confidence intervals at those scales. Shark movements became more tortuous at scales between m, which is most likely a function of more tortuous movements in larger patches. The final domain occurs at scales > 107 m, with D rapidly increasing at larger scales indicating that the sharks may use a highly correlated random walk to move between patches (e.g. Doerr and Doerr 2004). The small home ranges utilized suggests that blacktip reef sharks should have good information about the spatial distribution of patches within their home range. However, patches are still likely to be spatially and temporally dynamic, and theoretically, highly correlated random walks, leading to almost 109

111 straight movements, are thought to be the most efficient search strategy within a heterogenous environment (e.g. Zollner and Lima 1999, Philips et al., 2004). The repeated back and forth movements with a relatively straight movement path, should enable blacktip reef sharks to maximize search efficiency. Excursions and patch use on to the sand-flats are also made by some sharks, although at present we can only speculate that these are for foraging purposes. Short track time (maximum 3 d) is a limitation of active telemetry tracking, and it would certainly be desirable to analyze movement path structure over the scale of months to years. However, presently there is no other technique for obtaining high spatial resolution movement data from fish predators, especially those that confine their movements to small areas. Clearly, our results can not be directly extrapolated to blacktip reef shark populations world-wide, but they do show how micro-habitat quality and quantity can effect movements, behavior and life history of top-level predators. The design of efficient marine reserves to conserve shark populations is particularly difficult due to the wide ranging movements of these animals. However, by using the analytical framework presented here, and by quantifying the scales at which sharks view and respond to their environment, we will be able to improve models used to predict population level dispersal at other locations. Furthermore, the mounting evidence for shifting baselines at predator depleted atolls (Sandin et al., 2008), is making it increasingly important to quantify predator behavior in the few remaining pristine atolls like Palmyra. 110

112 Chapter V Distribution, size frequency, and sex ratios of blacktip reef sharks, Carcharhinus melanopterus, at Palmyra atoll: a predator-dominated ecosystem ABSTRACT Understanding the dynamics of marine apex predators in unperturbed predatordominated ecosystems is important for obtaining baseline information on predator/prey relationships and understanding ecosystem function. We conducted a study of blacktip reef shark Carcharhinus melanopterus, populations at Palmyra Atoll, a US National Wildlife Refuge where we captured 254 individuals between March 2004 and November Blacktip reef sharks were the most abundant apex predator in the Palmyra lagoons and were evenly distributed throughout the lagoons, except for lower Catch Per Unit Effort (CPUE) at Banjos sand-flats in the west lagoon. Sharks ranged in size from cm Total Length (TL, 94.9 ± 23.5 cm, mean ± 1SD), with females being caught at significantly larger sizes than males. Sex ratios did not differ from unity except for the Channel sand-flats where the ratio was skewed towards females. Male sharks possessed calcified claspers when they reached cm TL, suggesting that they were sexually mature, with 50 % of sharks caught having calcified claspers at 97 cm TL. Analysis of stomach contents (n = 14) revealed the first account of blacktip reef sharks foraging or scavenging on seabirds, with bird remains present in 29 % of stomachs. We recaptured 3.1 % of 193 sharks dart tagged after d at liberty, with no evidence of movements between lagoons, suggesting a large population size and site fidelity to specific lagoons. Due to the potentially high numbers of sharks at the atoll, blacktip reef sharks may have a 111

113 strong regulatory function on the marine ecosystem at Palmyra. The smaller sizes of blacktip reef sharks at Palmyra compared to other locations, suggests that this population may be growth limited, possibly due to intra-specific competition associated with high population densities. INTRODUCTION Marine apex predators are generally the first trophic group to be impacted by fishing and other anthropogenic influences (Jennings and Kaiser 1998, Jackson et al. 2001). Many species of shark are apex predators and due to their slow growth and other life history characteristics, population declines worldwide from over fishing are becoming apparent (e.g. Stevens et al., 2000, Meyer and Worm 2003). Consequently, areas that are closed to fishing or in remote locations are often characterized by a higher biomass of apex predators, in particular sharks (Friedlander and DeMartini 2002, Sandin et al., 2008, DeMartini et al., 2008). It is becoming increasingly important to quantify the ecological impacts of sharks in un-fished, pristine ecosystems, as this will provide baseline data which can help management decisions in areas subjected to fishing pressure. Understanding the ecological impacts of a population of predators requires knowledge of the predator s diet and feeding habits, movements and habitat utilization, distribution, population size and age and sex structure. Palmyra is an uninhabited atoll located in the central Pacific Ocean and has been a US National Wildlife Refuge since Due to Palmyra s protected status and remote location, it supports a healthy population of marine apex predators which makes up to 35% of the fish biomass (DeMartini et al., 2008, Sandin et al., 2008). A large shark population exists at Palmyra, and the dominant species inside the lagoons and over the 112

114 sand-flats is the blacktip reef shark, Carcharhinus melanopterus (Hobson 1963, Friedlander et al., 2007, Papastamatiou Chapter 4). Blacktip reef sharks are a common species on shallow coral reefs of the Indo-Pacific, where they are often the most abundant species of shark (Stevens 1984, Compagno et al., 2005). Dietary analysis in other locations suggests that blacktip reef sharks are tertiary predators and may therefore exert top-down control on some coral reef ecosystems (Lyle and Timms 1987, Cortes 1999). In order to describe the population of blacktip reef sharks at Palmyra, we conducted a population study of sharks over a three year period. Our specific goals were to: 1) describe the distribution of sharks throughout the Palmyra lagoons, 2) quantify sex ratios for sharks and how these may vary spatially, 3) quantify the size frequency distribution of sharks in the lagoons and 4) determine the size at which male sharks possess calcified claspers (as an indicator of sexual maturity). Although we had limited data due to collecting restrictions, we also obtained some information on shark diet. MATERIALS AND METHODS Study location Palmyra Atoll (N 5 53, W ) is located in the central Pacific and is part of the Line Island chain (Fig.17). The atoll is approximately 12.5 km in length, has an area of 27.6 km 2, and consists of two large lagoons (west and east) that are connected by a small channel. The lagoons are separated by a deteriorating road which was constructed by the US military during their occupation of Palmyra during WW2. A large channel (2 km length x 5 m deep) connects the west lagoon to the outer reefs (Fig.17). Water in the lagoons reaches a maximum depth of 50 m with a mud/sand bottom causing low 113

115 Figure 17. Location of the Line Islands in the Pacific Ocean (box), and Palmyra s location within the Line Island chain. Aerial image shows the location of the west (W) and east (E) lagoon, as well as fishing locations (circles). Sand-flats are B (Banjos), C (Channel), N (Nursery), and S (Sixes). 114

116 C B W Ba E Ba W A B Ed. In. N Ed.N J M S E S DE Co C 115