ECOLOGICAL AND LIFE HISTORY FACTORS INFLUENCE HABITAT AND TOOL USE IN WILD BOTTLENOSE DOLPHINS (TURSIOPS SP.)
A Dissertation submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Biology
By
Eric Michael Patterson, B.S.
Washington, DC December 7, 2012
Copyright 2011 by Eric Michael Patterson All Rights Reserved
ii ECOLOGICAL AND LIFE HISTORY FACTORS INFLUENCE HABITAT AND TOOL USE IN WILD BOTTLENOSE DOLPHINS (TURSIOPS SP.)
Eric Michael Patterson, B.S.
Thesis Advisor: Janet Mann, Ph.D.
ABSTRACT
While it has long been known that individual animals behave quite distinctively from other conspecifics, only recently has this intraspecific behavioral variation itself been the subject of investigation rather than a nuisance in statistical analyses. As research amounts, it is becoming ever more apparent that the ecology of individuals can have real and measurable biological consequences. Previous studies have documented pronounced inter-individual variation in foraging behaviors in Shark Bay wild bottlenose dolphins, which often coincide with individual specialization. In this dissertation I examine how individuals’ phenotypic characteristics relate to their foraging behavior, habitat use, and ranging.
In Chapter 1 I explore habitat and ranging for adult male and female bottlenose dolphins.
Dolphins show marked intraspecific variation in both habitat use and ranging, which partially relates to foraging tactic, sex, and season. However, some variation cannot be explained by any obvious phenotypic or ecological factors, highlighting the importance of individual level analyses, especially for conservation and management.
In Chapter 2 I examine the underlying basis of one foraging tactic involving tool use: sponge foraging. By mimicking dolphin sponging behavior, I found that dolphins that sponge forage (spongers) target prey that are difficult to detect with their echolocation, and dolphins have to use sponge tools to access this resource. Spongers appear to hunt an easy to catch,
iii reliable food source that non-spongers cannot access, thereby minimizing feeding competition and providing them access to an empty niche.
Finally, Chapter 3 explores life-long learning of the sponge foraging. Like humans, and many other animals, dolphins improve upon their skills with age, reaching a peak in sponging efficiency in midlife and declining thereafter. Dolphins also improved in their ability to locate sponges with age, and both this and efficiency likely help females meet the energetic demands of lactation.
In all, this dissertation increases our knowledge concerning how a species ecology and life history relates to intraspecific behavioral variation and has implications for conservation, tool-use, innovation, social learning, culture, and ultimately fitness and evolution.
iv To my friends, family, and to all the animals (both dolphin and non-dolphin alike) that made this work possible…
My sincere gratitude, ERIC
v ACKNOWLEDGEMENTS
I would first like to thank my thesis committee. Dr. Lisa Singh, Dr. Gina Wimp, and Dr.
Shweta Bansal were incredibly helpful throughout my entire PhD and my work would not have been possible without them. I would especially like to thank Dr. Wimp for being on all of my committees throughout my time at Georgetown, and for always asking the toughest questions and pushing me to be my best. Our long-term database, which all of my work greatly depends on, would not be in the shape it is today without the unending dedication of Dr. Lisa Singh.
There are so many friends that not only supported me in my work, but also kept me sane during the last 5 years. While in DC, the biology graduate students were an extended family.
Aaron Howard was always willing to help me with anything, from moving to various apartments to discussing the intricacies of our statistics (his being way more complicated than mine) and helping me learn R. I cannot thank Jenn Urbanksi enough for always being a friend to me and helping me when I needed it most. To all my other DC friends, both from Georgetown and beyond, I sincerely thank you for making this journey fun! I am also forever grateful to my cohort of associates while in Australia. Paul Hollick and family (Kate, Freya, Tim, and Ned) have always taken me in as one of their own, allowing me to stay at their place in Perth, buy their truck, and even park our boat in their only bay for weeks on end. Finally, to everyone that makes
Shark Bay not only a great place to do research, but also an enjoyable place to be stuck, thank you.
I would like to directly thank all of the researchers and organizations that supported this work, everyone in the Shark Bay Research Project (past and present), all members of the Shark
Bay Ecosystem Research Project, and all of our funding sources throughout the years (noted specifically for each chapter below). In particular, I would like to thank the members of the
vi Mann lab. Maggie Stanton has always been an incredible role model and is always willing to discuss methodological details. Ewa Krzyszczyk has taught me much of what I know about cetaceans and is a wonderful friend and colleague. Vivienne Foroughirad has been a oustanding assistant and research associate, and was always willing to look things over for me and help me get the data I needed. Meg Wallen brought in much needed fresh blood and a new perspective that was invaluable in helping me make my work more accessible. Finally, while I did not directly work with her, I am grateful to Dr. Brooke Sargeant for lighting the path with her work on dolphin foraging and individual specialization. The foundation that she set inspired me and made much of my work possible and to that I am indebt.
To my family, what can I say, without them I wouldn’t even exist. I can never repay them for the support they have given me, not only during graduate school, but throughout my entire life. I could not ask for a more supportive mother and father. To my siblings Bryce, Andrea, and
Elise, I am forever thankful for their willingness to entertain my interests, to listen to me talk about animals even when I have no idea what I am talking about, and for allowing us to watch
Discovery Channel and Animal Planet, even when there may have been something better on TV.
Of course I have to thank my dogs, Belle and Bella, for keeping my feet warm while I write, distracting me and pulling me out of my computer struck daze, and for always listening to me and never talking back. They are the best friends a man could ever ask for. Finally, to my amazing fiancé Katie: I am the luckiest man alive to have found her and I owe these last few years of happiness completely to Katie. Words cannot express my gratitude towards her.
Of course I must thank the dolphins, for being there, for exhibiting interesting behaviors, and for allowing us strange land mammals to watch from our noisy boats. Before graduate
vii school, I am not sure I had ever seen a wild dolphin, certainly never one up close, but after the first day out on the boat in Shark Bay I was hooked. Thank you to all the dolphins of Shark Bay.
Finally, I have to thank Dr. Janet Mann, my advisor. Dr. Mann took me under her wing the second we met. She gave me an opportunity to prove myself as I had no previous work with cetaceans and actually only had two years of undergrad in Biology due to switching my major. I only hope I have made her proud. She has taught me virtually everything I know about dolphins, animal behavior, and then some. Throughout my studies, I have tried some new techniques, new methods, some of which worked, some of which failed miserably, but she has supported them all, to the fullest. I hope that we continue to work together so I can soak up her incredible dedication to the project and to studying dolphin behavioral ecology.
Thank you everyone!
viii TABLE OF CONTENTS
Dissertation Introduction ...... 1
References ...... 6
Chapter I. Individual, Sex, and Seasonal Differences in Bottlenose Dolphin Habitat Use and Home Ranges in Shark Bay, Australia: Implications for Conservation and Management ...... 10
Abstract ...... 10
Introduction ...... 11
Materials and Methods ...... 15
Study site and design ...... 15
Habitat classification and availability ...... 16
Habitat use ...... 20
Home ranges ...... 24
Statistical Considerations ...... 25
Results ...... 26
Cluster and sex habitat use differences ...... 26
Cluster and sex habitat use diversity differences ...... 27
Seasonal habitat use ...... 28
Seasonal habitat use diversity ...... 30
Home ranges ...... 30
Discussion ...... 32
Acknowledgements ...... 40
References ...... 41
ix Figures and Tables ...... 55
Chapter II. The Ecological Conditions that Favor Tool Use and Innovation in Wild Bottlenose Dolphins (Tursiops sp.)...... 73
Abstract ...... 73
Introduction ...... 74
Results ...... 77
Discussion ...... 78
Materials and Methods ...... 81
Study Site ...... 81
Data collection...... 82
Data processing and statistical analysis ...... 84
Acknowledgements ...... 85
References ...... 87
Figures and Tables ...... 95
Chapter III. Becoming an Expert: Lifelong Learning in Tool-using Wild Bottlenose Dolphins (Tursiops sp.) ...... 101
Summary ...... 101
Results and Discussion ...... 102
Sponging experience ...... 103
Tool use efficiency ...... 103
Traveling with a tool ...... 107
Locating and detaching sponges ...... 108
x Energetic demands ...... 109
Implications ...... 111
Experimental Procedures ...... 115
Study site and data collection ...... 115
Genetic analysis...... 118
Statistical analysis ...... 119
Acknowledgements ...... 122
References ...... 124
Figures and Tables ...... 136
Dissertation Conclusion ...... 154
Appendix A ...... 156
xi LIST OF FIGURES AND TABLES
Chapter I
Figure 1. Map of study area ...... 55
Figure 2. Zonal histogram of depths for habitat types ...... 56
Figure 3. Zonal histogram of slopes for habitat types ...... 57
Figure 4. Habitat depths...... 58
Figure 5. Habitat slopes ...... 59
Figure 6. Habitat current ...... 60
Figure 7. Duality diagram of the eigenanalysis of habitat use selection ratios of
bottlenose dolphins in Shark Bay, Western Australia ...... 61
Figure 8. Cluster habitat selection ratios from type II design ...... 62
Figure 9. Population level selection ratios from type I design ...... 63
Figure 10. Male and female habitat use diversity by cluster ...... 64
Figure 11. Seasonal habitat selection ratios...... 65
Figure 12. Seasonal habitat diversity ...... 66
Figure 13. Provisioned and non-provisioned female home range sizes in the Mixed
Deep open cluster ...... 67
Figure 14. Male and female home range sizes by cluster ...... 68
Table 1. Habitat classifications ...... 69
Table 2. Habitat use clusters ...... 70
Table 3. Habitat comparison results for type II design...... 71
Table 4. Habitat comparison results for type I design ...... 72
xii Chapter II
Figure 1. Sponging in Shark Bay...... 95
Figure 2. Sponging map...... 96
Figure 3. Ratio of prey without swimbladders (SB) to prey with swimbladders
during both Sponging and Non-Sponging on transects ...... 97
Figure 4. Abundance of prey species extracted during Sponging and abundance of
these same families during Non-Sponging on transects ...... 98
Table 1. Prey abundance from Sponging and Non-Sponging, pooled from both
transects and verification dives ...... 99
Table 2. Sponging and historical prey families ...... 100
Chapter III
Figure 1. Female bottlenose dolphin (Tursiops sp.) with marine sponge tool
(Ircinia sp.) in Shark Bay, Australia ...... 136
Figure 2. Percent time sponging as a function of age ...... 137
Figure 3. Partial effect of age on the amount of foraging with a single sponge ...... 138
Figure 4. Partial effect of age on the amount of time females traveled with a
single sponge ...... 139
Figure 5. Ratio of time spent searching for and detaching sponges, to time spent
sponging as a function of age ...... 140
Figure 6. Partial effect of age on the ratio of time spent searching for and
detaching sponges, to time spent sponging both before and after the peak in tool
use efficiency separately ...... 141
xiii Figure 7. Sponge density ...... 142
Figure 8. Partial effect of age on probability of lactating ...... 143
Table 1. Primers used in genetic analysis of dolphin fecal sample ...... 144
Table 2. Results from genetic analysis of sponger fecal sample and Parapercis nebulosa ...... 145
Table 3. MCMC glmm for tool bout duration ...... 146
Table 4. MCMC glmm for time traveled with a single sponge tool ...... 147
Table 5. MCMC glmm for the ratio of locating and detaching to sponging ...... 148
Table 6. MCMC glmm for the probability of lactating ...... 149
Table 7. Cox-proportional hazards model for tool bout duration ...... 150
Table 8. Cox-proportional hazards model for the time traveled with a single sponge ...... 151
Table 9. Maximum likelihood glmm for the ratio of locating and detaching to sponging...... 152
Table 10. Maximum likelihood glmm for the probability of lactating ...... 153
xiv DISSERTATION INTRODUCTION
Successful foraging is essential to an animal’s survival and reproduction. Consequently, most animals spend a substantial portion of their time dedicated to foraging, and natural selection should favor those behaviors that are most effective in meeting their energetic demands. In fact, the competition for food likely drives speciation and has certainly contributed to the great diversity of life inhabiting our planet (Dieckmann & Doebeli 1999). In general, animals’ foraging ecology is usually described at the species or population level, but mounting evidence suggests that intrapopulation variation in foraging behaviors is widespread and has important consequences for our understanding of species’ evolution, ecology, and conservation (see
Bolnick et al., 2003 and Araújo et al. 2011 for review).
Numerous factors can drive this variation including an individual’s life history, development, cognitive abilities, sociality, ecology, and the structure of the population in which it lives (Hughes 1979; Bolnick et al. 2003; Svanbäck & Bolnick 2005; Tinker et al. 2008, 2009;
Araújo et al. 2011). In Shark Bay, Australia, wild bottlenose dolphins (Tursiops sp.) are known to vary widely in their use of 18 distinct foraging tactics (Mann & Sargeant 2003, unpublished data), with some females demonstrating marked individual specialization (Mann & Sargeant
2003; Sargeant 2005; Sargeant et al. 2005; Sargeant & Mann 2009). This extensive variability in foraging behavior appears to at least partially relate to the different ecological and social contexts an individual experiences during its life (Sargeant et al. 2007; Sargeant & Mann 2009), but many aspects of this intraspecific variation in foraging remain unexplored.
For example, previous data show that many of these foraging tactics are fairly habitat specific (Sargeant et al. 2007). Coupling this with the inter-individual variation in the use of
1 these tactics suggests that individuals should vary greatly in their use of available habitats and likely their ranging. However, most previous studies of Shark Bay bottlenose dolphin habitat use were performed at the population level assuming identical habitat use across animals (Heithaus
& Dill 2002, 2006; Sargeant et al. 2007), and as such are unable to explore how animal’s foraging behavior, along with other phenotypic traits, affects an animal’s space use. In fact, despite the vast amount of detailed information the 29 year long-term Shark Bay Dolphin
Research Project has provided on wild bottlenose dolphin behavioral ecology, no single study has simultaneously explored population-wide patterns of both habitat use and ranging for individual dolphins (although see Mann & Watson-Capps 2005 for some habitat use analysis and
Tsai & Mann 2012 for some home range analysis).
In Chapter 1, I examine habitat use and ranging among adult bottlenose dolphins in an effort to fill this gap in our knowledge. In doing so, I explore space use at the population and individual level, both across and during three distinct seasons. I discuss the important role an individual’s phenotype plays in shaping its spatial distribution and ranging patterns, and how an individual’s habitat use and ranging relates to our previous knowledge concerning Shark Bay dolphins’ sociality, reproduction, foraging, and mating system. Such information is vital for understanding a species’ behavioral ecology, both at the population and individual level, and is also critical for successful management and conservation.
The variation among individuals in their foraging tactics and the restricted use of these tactics in particular habitats, suggests that individuals adjust their behavior to meet the demands of their local ecology. While the specifics regarding how some tactics are tailored to the local environment and prey are fairly evident (e.g. beach hunting, Sargeant et al., 2005), for other tactics the ecological underpinnings are less clear. In what remains as the only well documented
2 case of wild cetacean tool use, Smolker et al. (1997) first described their observations of a subset of dolphins in Shark Bay, primarily females, wearing basket shaped marine sponges
(Echinodictyum mesenterinum) over their beaks. They hypothesized that the sponge is used during the search phase of foraging and provides dolphins with protection from sharp rocks, shells, and possibly noxious organisms as they probe the substrate. Since then we have learned a great deal about sponge foraging (hereafter sponging, e.g. Mann and Sargeant, 2003; Krützen et al., 2005; Sargeant et al., 2007; Mann et al., 2008; Sargeant and Mann, 2009; Bacher et al.,
2010), but the underlying basis remains unknown. Why dolphins need to probe the substrate, which in turn promotes the use of protective sponges, is unclear given their exquisite echolocation abilities. Although dolphins’ echolocation skills are quite remarkable (Au 2004), they rely heavily on a difference in medium density between the target (prey) and the surrounding environment (water).
In Chapter 2, I test the hypothesis that the primary prey targeted during sponging is inaudiable to dolphins’ sonar and thus requires physical extraction from the substrate. I predicted that these prey, like several benthic fish (McCune & Carlson 2004), lack swimbladders giving them a density similar to water (Jones & Marshall 1953) and little acoustic impedance (Foote
1980). Such a result would be in line with observations of other tool using animals in that individuals resort to tools primarily to overcome their anatomical limitations (Gibson 1986).
Individual foraging specializations exist when at least some individuals have a narrower niche than the population as a whole (Bolnick et al., 2003; although see Sargeant, 2007 for a discussion of definitions). Specialists have either a restricted diet, use a single or few foraging tactics, or both. If the specialization is temporally consistent, specialists may gain frequency dependent advantages such as increased detection rates, foraging success, decreased handling
3 time, or improved digestive efficiency (Bolnick et al. 2003; Tinker et al. 2009). Of all the individuals in Shark Bay, dolphins that sponge forage (spongers) are the most specialized, spending ~ 96% of their foraging time using sponge tools. Sponging may also be the most difficult foraging tactic to learn as it requires complex object manipulation, occurs in deep channels with strong tidal currents, and onset occurs a year later than most other foraging tactics
(Mann & Sargeant 2003; Mann et al. 2008; Sargeant & Mann 2009). Despite these proximate costs, previous research indicates that sponging provides females with equal calving success compared to their non-tool-using counterparts (Mann et al. 2008). It may be that by specializing, spongers reduce the costs associated with their foraging tactic by refining their skills with experience, but beyond the initial onset of sponging around 2 years of age (Sargeant & Mann
2009), we know very little about how spongers change their behavior throughout their lives.
In the final chapter, I explore life-long learning in spongers and examine whether females continuously improve their sponging efficiency with experience. Specifically, I test whether spongers develop what, in the human literature, is referred to as expertise, while controlling for foraging effort and possible ecological factors. I also examine how sponging efficiency relates to female life history including the costs associated with maternal care and discuss the potential fitness consequences, role in behavioral specialization, and influence on social learning and culture.
From examining some of the most basic aspects of dolphin behavior, like habitat use and ranging, to investigating the details concerning the ecological characteristics, persistence, and efficiency of particular foraging tactics, this study aims to improve our understanding of how individual foraging specialization evolves and shapes an animal’s behavior. Such inter-individual
4 variation in behaviors directly tied to fitness has profound consequences for a population’s ecological and evolutionary dynamics.
5 REFERENCES
Araújo, M. S., Bolnick, D. I. & Layman, C. A. 2011. The ecological causes of individual
specialisation. Ecology Letters, 14, 948–958.
Au, W. W. L. 2004. The dolphin sonar: excellent capabilities in spite of some mediocre
properties. In: High Frequency Ocean Acoustics Conference: AIP Conference Proceedings,
Vol 728 (Ed. by M. B. Porter, M. Siderius, & W. A. Kuperman), pp. 247–259. Melville:
American Institute of Physics.
Bacher, K., Allen, S. J., Lindholm, A. K., Bejder, L. & Krützen, M. 2010. Genes or culture:
are mitochondrial genes associated with tool use in bottlenose dolphins (Tursiops sp.)?
Behavior Genetics, 40, 706–714.
Bolnick, D. I., Svanbäck, R., Fordyce, J. A., Yang, L. H., Davis, J. M., Hulsey, C. D. &
Forister, M. L. 2003. The ecology of individuals: incidence and implications of individual
specialization. The American Naturalist, 161, 1–28.
Dieckmann, U. & Doebeli, M. 1999. On the origin of species by sympatric speciation. Nature,
400, 354–357.
Foote, K. G. 1980. Importance of the swimbladder in acoustic scattering by fish: a comparison
of gadoid and mackerel target strengths. Journal of the Acoustical Society of America, 67,
2084–2089.
Gibson, K. R. 1986. Cognition, brain size and the extraction of embedded food resources. In:
Primate Ontogeny, Cognition, and Social Behaviour, (Ed. by J. G. Else & P. C. Lee), pp.
93–103. Cambridge: Cambridge University Press.
Heithaus, M. R. & Dill, L. M. 2002. Food availability and tiger shark predation risk influence
bottlenose dolphin habitat use. Ecology, 83, 480–491.
6 Heithaus, M. R. & Dill, L. M. 2006. Does tiger shark predation risk influence foraging habitat
use by bottlenose dolphins at multiple spatial scales? Oikos, 2, 257–264.
Hughes, R. 1979. Optimal diets under the energy maximization premise: the effects of
recognition time and learning. The American Naturalist, 113, 209–221.
Jones, F. R. H. & Marshall, N. B. 1953. The structure and functions of the theleostean
swimbladder. Biological Reviews of the Cambridge Philosophical Society, 28, 16–82.
Krützen, M., Mann, J., Heithaus, M. R., Connor, R. C., Bejder, L. & Sherwin, W. B. 2005.
Cultural transmission of tool use in bottlenose dolphins. Proceedings of the National
Academy of Sciences of the United States of America, 102, 8939–8943.
Mann, J. & Sargeant, B. L. 2003. Like mother, like calf: the ontogeny of foraging traditions in
wild Indian ocean bottlenose dolphins (Tursiops sp.). In: The Biology of Traditions: Models
and Evidence, (Ed. by D. Fragaszy & S. Perry), pp. 236–266. Cambridge: Cambridge
University Press.
Mann, J. & Watson-Capps, J. J. 2005. Surviving at sea: ecological and behavioural predictors
of calf mortality in Indian Ocean bottlenose dolphins, Tursiops sp. Animal Behaviour, 69,
899–909.
Mann, J., Sargeant, B. L., Watson-Capps, J. J., Gibson, Q. A., Heithaus, M. R., Connor, R.
C. & Patterson, E. M. 2008. Why do dolphins carry sponges? PloS ONE, 3, e3868.
McCune, A. R. & Carlson, R. L. 2004. Twenty ways to lose your bladder: common natural
mutants in zebrafish and widespread convergence of swim bladder loss among teleost
fishes. Evolution & Development, 6, 246–259.
7 Sargeant, B. L. 2005. Foraging development and individual specialization in wild bottlenose
dolphins (Tursiops sp.). Ph.D. thesis. Washington, D.C. U.S.A.: Georgetown University,
Department of Biology.
Sargeant, B. L. 2007. Individual foraging specialization: niche width versus niche overlap.
Oikos, 116, 1431–1437.
Sargeant, B. L. & Mann, J. 2009. Developmental evidence for foraging traditions in wild
bottlenose dolphins. Animal Behaviour, 78, 715–721.
Sargeant, B. L., Mann, J., Berggren, P. & Krützen, M. 2005. Specialization and development
of beach hunting, a rare foraging behavior, by wild bottlenose dolphins (Tursiops sp.).
Canadian Journal of Zoology, 83, 1400–1410.
Sargeant, B. L., Wirsing, A. J., Heithaus, M. R. & Mann, J. 2007. Can environmental
heterogeneity explain individual foraging variation in wild bottlenose dolphins (Tursiops
sp.)? Behavioral Ecology and Sociobiology, 61, 679–688.
Smolker, R. A., Richards, A. F., Connor, R. C., Mann, J. & Berggren, P. 1997. Sponge
carrying by dolphins (Delphinidae, Tursiops sp.): a foraging specialization involving tool
use? Ethology, 103, 454–465.
Svanbäck, R. & Bolnick, D. I. 2005. Intraspecific competition affects the strength of individual
specialization: an optimal diet theory method. Evolutionary Ecology Research, 7, 993–
1012.
Tinker, M. T., Bentall, G. & Estes, J. A. 2008. Food limitation leads to behavioral
diversification and dietary specialization in sea otters. Proceedings of the National Academy
of Sciences of the United States of America, 105, 560–565.
8 Tinker, M. T., Mangel, M. & Estes, J. A. 2009. Learning to be different: acquired skills, social
learning, frequency dependence, and environmental variation can cause behaviourally
mediated foraging specializations. Evolutionary Ecology Research, 11, 841–869.
Tsai, Y.-J. J. & Mann, J. 2012. Dispersal, philopatry, and the role of fission-fusion dynamics in
bottlenose dolphins. Marine Mammal Science, 28, doi: 10.1111/j.1748–7692.2011.00559.x.
9 CHAPTER I
Individual, Sex, and Seasonal Differences in Bottlenose Dolphin Habitat Use and Home
Ranges in Shark Bay, Australia: Implications for Conservation and Management*
ABSTRACT
Habitat use and home range studies provide some of the most basic and essential data for understanding an animal’s behavioral ecology. Such studies are also critical for effective management and wildlife protection, but difficult to conduct in marine environments. Here we investigate habitat use, home ranges, and their relationship to sex and seasonality for one of the best-studied populations of wild bottlenose dolphins (Tursiops sp.). Consistent with previous research, we find marked intraspecific variation in habitat use and home ranges. Dolphins fell into one of four habitat use clusters. Across clusters, males had greater habitat use diversity and larger home ranges, which is consistent with sex-specific mating strategies: males roam further in search of females, while females tend to specialize in habitat specific foraging tactics. Other sex and seasonal differences depended greatly on cluster. Thus, for this behaviorally plastic species habitat use and ranging depend on both sex and season, but moreover, on the individual animal itself. Our study suggests that an understanding of individual variation is essential for managers to make informed decisions concerning marine protected areas (MPAs) and wildlife conservation.
*Authorship for paper: Eric M. Patterson & Janet Mann 10 INTRODUCTION
Understanding the habitat use and spatial distributions of top marine predators is not only a priority of marine ecologists, but vital to marine conservation. Wide-ranging, top predators such as cetaceans serve as sentinel species for their respective ecosystems (Hooker 2004; Wells et al. 2004; Bossart 2006; Twiner et al. 2011). Thus, knowledge concerning their habitat use, and in particular critical habitats, aids in conserving many trophic levels (Hoyt 2005) and can be used to establish marine protected (MPAs, e.g. Cañadas et al., 2005). Moreover, the immense public interest in cetacean welfare increases the likelihood that proposed protected areas are implemented and ultimately successful (Hoyt 2005). Bottlenose dolphins (Tursiops spp.) are a cosmopolitan, coastal genus (Connor et al. 2000) and face numerous anthropogenic threats worldwide including boat traffic and ecotourism (Allen & Read 2000; Constantine 2001; Bejder et al. 2006), recreational and commercial fishing, (Chilvers et al. 2003; Bearzi et al. 2006; Díaz
López 2006; Allen & Loneragan 2010), aquaculture (Watson-Capps & Mann 2005; Díaz López
& Shirai 2008), pollution (Wilson et al. 1999; Wells et al. 2005), and habitat degradation
(Harzen & Brunnick 1997; Bearzi et al. 2004). At sites where human impact is minimal, studies on bottlenose dolphin spatial movements have been useful for examining the effects of subtle anthropogenic changes (e.g. Watson-Capps and Mann, 2005; Bejder et al., 2006), and provide a baseline for areas where historical data prior to substantial impact are incomplete or non-existent
(e.g. Bearzi et al., 2004; Wells et al., 2005).
Bottlenose dolphins demonstrate remarkable behavioral plasticity enabling them to occupy an extraordinary array of habitats throughout the world (Shane et al. 1986). Prey availability presumably drives their habitat use and movements (Shane et al. 1986; Wilson et al.
11 1997; Hastie et al. 2004; Bailey & Thompson 2010), although a number of other factors may be involved. These include, season (defined most often by changes in temperature, but may include a breeding season; Shane et al. 1986; Heithaus & Dill 2002; Chilvers et al. 2003), time of day
(Shane et al. 1986; Allen et al. 2001), tidal state (Scott et al. 1990; Sargeant et al. 2005), the distribution of predators (Scott et al. 1990; Heithaus & Dill 2002, 2006), and the demographic, reproductive, and ontogenetic traits of individuals (Scott et al. 1990; Wilson et al. 1997;
Heithaus & Dill 2002; McHugh et al. 2011).
In Shark Bay, Australia, previous studies found that bottlenose dolphins vary habitat use with behavior, season, the presence of tiger sharks (Galeocerdo cuvier), age, and sex (Heithaus
& Dill 2002, 2006; Mann & Watson-Capps 2005; Sargeant et al. 2007). Specifically, in a population level study, Heithaus and Dill (2002) found that dolphins foraged in deep water habitats (7–13 m) during warm months when relative tiger shark abundance is high in shallow water habitats, indicating strong predator avoidance. Dolphins also appeared to rest more in deep water habitats, regardless of season (Heithaus & Dill 2002). In contrast, in an individual level study Mann and Watson-Capps (2005) found that mothers and calves (presumably most vulnerable to predation) did not appear to shift habitat use during warm months. Mothers, however, did adjust resting behavior seasonally, resting less in deep water during warm months.
They suggest that mothers increase vigilance in deep water during warm months due to the higher total predator abundance here, but maintain that predation is not the primary cause of calf mortality. In a follow up study, Heithaus and Dill (2006) found that seasonal increases in shark abundance also correlated with dolphins foraging more in shallow edge habitat, a pattern that surprisingly has dolphins matching the distribution of their tiger shark predators (Heithaus & Dill
2006; Heithaus et al. 2006). Thus, it remains unclear how seasonal and demographic differences
12 in habitat use fit into the more complete picture of what we currently know about Shark Bay dolphins’ sociality (Gibson & Mann 2008; Frère et al. 2010; Stanton et al. 2011; Randic et al.
2012; Mann et al. 2012), reproduction (Mann et al. 2000), foraging (Mann & Sargeant 2003;
Sargeant et al. 2005, 2007; Mann et al. 2008; Sargeant & Mann 2009; Patterson & Mann 2011), and mating system (Connor et al. 2006; Connor 2010).
For example, many female bottlenose dolphins (and some males) in Shark Bay exhibit marked individual specialization in habitat specific foraging behaviors (Mann & Sargeant 2003;
Sargeant et al. 2005, 2007; Patterson & Mann 2011). While some individuals use more than one tactic (Mann & Sargeant 2003), others almost exclusively use a single tactic (~ 96% of foraging) bound to a specific habitat (Mann et al. 2008) making them likely less responsive to seasonal changes in predator abundance. In fact, because of these sex differences in degree and type of foraging specialization, males in general may respond more readily to changes in ecological conditions compared to females, be it changes in predator abundance (Heithaus 2001; Wirsing et al. 2006) or prey communities (Heithaus 2004). For males, specializing in a particular habitat or foraging tactic may restrict their ranging (Randic et al. 2012) and ability to form long-term social bonds (alliances) with other males necessary for gaining access to females (Connor et al. 2006;
Mann et al. 2008).
Furthermore, even though previous studies on habitat use in Shark Bay identified individual dolphins, animals were not used as the sampling unit and data were analyzed at the population level (Heithaus and Dill, 2002, 2006; Sargeant et al., 2007; type I design of Thomas and Taylor, 2006). Such population level inferences assume that all animals select habitat similarly (Thomas & Taylor 2006) aside from differences captured by factors in the analysis itself (e.g. age, sex, etc.), which is rarely the case (Alldredge et al. 1998). When intraspecific
13 variation in habitat use within a population is expected, individual animals should be used as the sampling unit (Aebischer et al. 1993; Manly et al. 2002; McDonald et al. 2005; Alldredge &
Griswold 2006; Thomas & Taylor 2006). Pooling data across individuals biases towards the more frequently encountered animals and can lead to erroneous results (Garshelis 2000) and poor management policies (Conde et al. 2010; Buij et al. 2012). While one previous study examined individual level habitat use (Mann & Watson-Capps 2005), it is limited in terms of management applications because only mothers and calves were observed.
Here we examine fine-scale habitat use patterns for individual adult male and female bottlenose dolphins (Tursiops sp.) over a 24-year period in Shark Bay, Australia. By investigating habitat use at this level, we aim to better understand individual habitat use strategies and how they relate to findings from previous research. Wildlife management often needs such critical habitat data since rarely are all individuals managed identically and/or can all habitats be completely safeguarded. We focus on five main questions 1.) Do all dolphins use habitat in a similar way? 2.) Do dolphins use habitat randomly with respect to availability? 3.)
Which habitats are preferred and by which dolphins? 4.) Are there seasonal and/or sex differences in habitat use or habitat use diversity? 5.) How does habitat use and sex relate to home range size? In doing so we re-characterize habitat for our main study in much greater detail and use more biologically relevant seasons than previous research.
14 MATERIALS AND METHODS
Study site and design
The Shark Bay Dolphin Research Project has an extensive 29-year database that includes demographic, genetic, association, life history, ecological, and behavioral data on > 1,400 dolphins (past and present, including over 600 births) residential to a ~ 500 km2 area of the eastern gulf of Shark Bay, Western Australia (25° 47’ S, 113° 43’ E, Fig. 1). Data for this study were collected during boat-based surveys aboard 3–6 m research vessels from 1988–2011.
Surveys consisted of short 5–60 minute behavioral observations in which group composition
(dolphin identity determined using Photo ID, Würsig & Würsig 1977), location, behavioral, and ecological data were collected (Mann 1999; Gibson & Mann 2009). Primarily, we utilized a type
II design where individual animals are recognized and considered replicates, and habitat availability is assessed throughout the study area (Johnson 1980; Thomas & Taylor 1990, 2006;
Manly et al. 2002). However, to investigate the effect of study design on our results, we also performed some analyses using a type I design where animal identities are not used, data from all sightings are pooled and used as replicates, but habitat availability is still assessed throughout the study area (Johnson 1980; Thomas & Taylor 1990, 2006; Manly et al. 2002). For this study, only data from adult animals (≥ 10 years of age) were used because calves’ habitat use is likely dependent on their mothers (Tsai & Mann 2012) and the juvenile period may reflect distinct socio-ecological challenges compared to adults (McHugh et al. 2011). Additionally, if animals were sighted more than once in the same day, only the last location point per day was used to
15 reduce spatial autocorrelation and biases associated with the location of our boat-launching site
(Monkey Mia, Fig. 1).
Habitat classification and availability
In past studies, habitat in our main study area (Fig. 1) has been described as consisting of embayment plains (5–13 m), shallow sand flats (0.5–4 m), seagrass beds (0.5–4 m), and bisecting deep channels (7–13 m). However, classifying habitats is not trivial and depends both on the scale of the study and the specific environmental conditions of the study site. In general the spatial scale should match the study species and the question being investigated
(Meentemeyer & Box 1987). For coastal resident bottlenose dolphins with small home ranges relative to other cetaceans (40–100 km2; McHugh et al. 2011; Randic et al. 2012; Tsai & Mann
2012) often containing extreme habitat heterogeneity (e.g. Sargeant et al. 2007; Torres et al.
2008), fine scale habitat measures are necessary for uncovering detailed habitat preferences
(Allen et al. 2001).
As such, we aimed to improve on previous habitat classifications by incorporating as many environmental variables as possible to characterize habitat for a very well sampled area. We also wanted to maintain habitat definitions that would be relevant and applicable to management practices. Depth, substrate, slope, and current flow were considered when defining habitat categories. We utilized both historical environmental data from our long-term database and newly collected data from 84 transects and other opportunistic habitat sampling.
In order to determine a reasonable study area to restrict our habitat assessment and dolphin-sighting data to, we created 90% isopleths of fixed kernel densities (Gaussian
16 distribution) with a least-squares cross validation (LSCV) smoothing parameter (Worton 1989;
Seaman & Powell 1996; adehabitatHR package in R, Calenge 2006) around both our historical survey sighting locations (1984–2011), as well as available vessel track data (2002–2011). This provided a guide to construct a study area of approximately 130 km2 (Fig. 1) that was extremely well sampled in terms of search effort and dolphin sightings. Bathymetry within this study area was determined by compiling historical sounding data from boat surveys and tracks corrected for tides using predictions by JTides (Lutus 2011), a simple java program that predicts tides using publicly available data. All tide corrected depth values and corresponding GPS locations
(111,294 points in total) were projected (WGS 1984 UTM Zone 49S) in ArcGIS Map 10 (ESRI
2011) and an ordinary kriging interpolation technique was used to generate a digital elevation model (DEM). From this DEM, slope was derived using the slope function in the Spatial Analyst
Toolbox in ArcGIS 10. Substrate for shallow areas was determined from an enhanced high- resolution orthophotograph (Fig. 1). For deeper areas where substrate is difficult to determine from aerial photographs, 84 randomly located dives (40 in the western portion, 44 in the eastern portion of the study area, Fig. 1) were performed in which divers filmed benthic substrate and habitat using a belt transect method (see Patterson and Mann, 2011 and Chapter 2 for full details). Approximately five sediment samples from 80 transects were also collected and analyzed for particle diameters using a Beckman Coulter LS200 to classify substrate using the
Wentworth grade scale (Wentworth 1922). While the extreme currents in the channels in the eastern half of the study area were evident to divers and have been previously modeled (Burling et al. 2003), surface current velocity measurements were opportunistically taken at several locations in deep and shallow areas (N=38) using a Speedtech FL-K1 Redesigned Swiss-made
Flowmeter Kit.
17 From these data, six habitats types were defined. Channel habitat occurs in the eastern portion of our study area, has a fragmented rock, shell, and coral debris substrate with intermittent small patches of seagrass and sand, relatively deep depths, strong current, and fairly gradual slopes (Patterson & Mann 2011). Deep open habitat occurs in the western portion of the study area, consists of a soft substrate (sand/silt/clay mix), has similar depths to Channel habitat, but weaker currents, and also fairly gradual slopes. Sand flats have a continuously sand substrate, extremely shallow depths, very gradual slopes, and little current. Seagrass beds are continuously covered by seagrass (predominantly Amphibolis antarctica and occasionally Posidonia australis,
Walker et al. 1988) and occur mainly in moderate to shallow depths, have gradual slopes, and little current. Two ecotone zones were also defined in order to 1.) capture areas of increased slope, which have been demonstrated to be preferred by bottlenose dolphins elsewhere (Bailey &
Thompson 2010) and 2.) investigate the importance of edge habitat, which is sometimes neglected in habitat classifications (Conner et al. 2003). Deep ecotone is the 100 m wide zone centered on the border between shallower habitats (Seagrass beds and Sand flats) and deeper habitats (Deep open and Channel), has moderate depths, and relatively steep slopes. Surface current data were not collected for Deep ecotone as this habitat is mixed and has complex topography and hydrology that would not be captured very well by surface current velocity measurements. Shallow ecotone is the 100 m zone centered on the border between two shallower habitats (Seagrass and Sand flats), has shallow depths, gradual slopes, and low current. Because these two ecotone zones represent habitats transitions, both are made up of a mixture of habitats and therefore also capture the patchy and discontinuous nature of edge habitat.
To further explore how the six selected habitat types capture the continuous environmental variables depth and slope, we generated 100 random survey points in each habitat
18 using built in toolboxes in ArcGIS 10. We then extracted depth and slope values from the DEM model and compared the average depth and slope for these random surveys between all habitats using multiple permutation t-tests with a Bonferroni adjustment. The results, along with zonal histograms of depth and slope are presented on Figures 2–6. This simulation study demonstrated that if random surveys were taken in each habitat type, all habitats would be different in terms of depth except Deep open and Channel, the two deepest habitats. However, these two habitats are readily distinguished by the differences in substrate, as well as current (Fig. 6). Less distinction was clear in terms of slope values for the various habitat types. However, as intended the increase in slope occurring in edge habitat between shallow and deep areas was well captured by the Deep ecotone habitat type. These 6 habitats adequately represented the variation in depth, slope, substrate, and current found in our study site (Table 1). However, it is important to note that these habitat types differ from those in previous studies investigating bottlenose dolphin habitat use where depth (Heithaus & Dill 2002) or distance to edge (Heithaus & Dill 2006) were the main factors used to define habitats.
Finally, to determine habitat availability, we calculated the percent area covered by each habitat in the study area (Table 1). Because some habitats have datum depths (depth at lowest recorded tide) below 1 m, which is assumed to be minimum depth for dolphins in this study, we corrected for the reduced availability caused by the tidal cycle. After determining the average depth of the portion of each habitat below 1 m, we calculated the tide required to make this area available (i.e. > 1 m depth). We then weighted this habitat area by the percent of time tides were greater than or equal to this required tide using 1-minute interval tide predictions from JTides.
The area below 1 m datum depth for each habitat was then added back into the habitat area for that particular habitat (Table 1). For seasonal analyses, this same process was used to calculate
19 habitat availability for each season separately since the average tidal height of each season differs. All portions of the study area were considered to be equally accessible to all dolphins given that only one location point per day, per dolphin, was used and the total area for this study was ~130 km2 with a perimeter of ~ 50 km, an area easily traversed by bottlenose dolphins in a single day (reviewed in Connor et al. 2000).
Habitat use
To determine habitat use, survey location points were projected in ArcGIS 10 (WGS
1984 UTM Zone 49S) and overlaid on habitat classifications. For each survey, habitat type was scored for all animals present, but because bottlenose dolphins live in a fission-fusion society
(Smolker et al. 1992) and do not cooperatively forage in Shark Bay, an animal’s habitat use is not inherently dependent on its associates and this does not constitute pseudo-replication
(Millspaugh et al. 1998, although see later discussion on male alliances). A minimum number of
22 sightings were needed for each dolphin to be sighted in all habitats at least once if habitats were used in proportion to availability. However, after initial analyses, this 22-point cut off proved too restrictive as the scarcest of habitats (Sand flats) was avoided by almost all dolphins and was no animal’s primary habitat. As such, we lowered our cut off to 14 points in accordance with the next scarcest of habitats (Shallow and Deep ecotone). Nevertheless, all analyses (except seasonal) were performed using both the 22-point and 14-point cut offs and results were identical so only those from the 14-point analyses with the larger data set (N= 221, 107 females, 113 males, one unknown sex) are reported.
20 For initial investigation into habitat use we first calculated selection ratios both at the population level (type I design) and at the individual animal level (type II design) as described by
Manly et al. (2002). Selection ratios are simply the ratio of proportional use of a habitat to the proportional availability of that habitat; a value of one indicates use equal to availability while values above and below one indicate positive and negative selection respectively. To investigate similarity in habitat use among animals, which can only be done with the type II design, we performed eigenanalysis (adehabitatHS package in R, Calenge 2006; Calenge et al. 2006). This graphically displays how animals use habitats in accordance with availability (as denoted by distance from the origin), which habitats may be preferred (by comparing to a habitat loading graph), and if animals have similar habitat use (Calenge & Dufour 2006; Palatitz et al. 2011).
Permutation chi-square tests were then used to statistically determine if all animals use habitat in a similar way (type II design), if they use it in accordance with availability assuming similar habitat use (type I and II designs), and if they use it in accordance with availability, even when habitat use is dissimilar (type II; Manly et al. 2002).
From these analyses it was apparent that Shark Bay dolphins employed a variety of habitat use strategies and habitat use should be examined at the individual level. As such, a cluster analysis (ClusterSim package in R, Walesiak 2008) was performed on selection ratios in order to group animals based on habitat use similarity. The clustering method that maximized the the variance ratio criterion proposed by Caliński and Harabasz (1974), resulted in four biologically meaningful groups using PAM clustering (Partition Around Medoids, Kaufman &
Rousseeuw 2005; fcp package in R, Hennig 2010) with the general distance measure (ClusterSim package in R, Walesiak 2006, 2008). PAM is a more robust version of the common K-means clustering where clusters are centered around medoids rather than means (Kaufman &
21 Rousseeuw 2005), which is more appropriate for non-normally distributed data such as selection ratios. The general distance measure (GDM1 from ClusterSim package in R) ranges from 0–1 and is based on the idea of a generalized correlation coefficient (Jajuga et al. 2003). We also explored several fuzzy clustering algorithms (fuzzy c-means, noise clustering: vegclust package in R, Caceres 2012; fanny: cluster package in R, Maechler et al. 2012), but on average ~ 94% of dolphins could be placed in one of four clusters using an α threshold of 0.5, further confirming that partitioning-based clustering is appropriate for our data. These four crisp clusters are essential to the rest of our analyses and our goal of providing management with practical habitat use information. Eigenanalysis and permutation chi-square tests were re-run on these clusters individually (results not presented) to confirm that clustering had improved the within group habitat use similarity.
For each cluster, the selection of habitats was then compared to availability and to each other using permutation Z tests. For illustration, we also performed the same comparisons using the data from the type I design that ignores intraspecific variation and clustering, but all analyses beyond this were performed only with the data from the type II design. Diversity of habitat use was determined by calculating the Shannon’s diversity index (H’) on selection ratios for each animal (Shannon & Weaver 1949; du Toit & Owen-Smith 1989). Sex and cluster differences in habitat use and habitat use diversity were examined using multiple permutation tests on selection ratios and H’.
To explore seasonal changes in habitat use and habitat use diversity, seasonal selection ratios and diversity indices (H’) were calculated for a subset of animals that had at least 14 points per season (N=72, 39 females, 33 males). Studies of seasonal patterns rarely consider that the factor season can only be applied at the population level (i.e. winter for one animal is winter for
22 all others as well). Thus, preferably one should consider years as replicates since using anything else (individual animals, transects, etc.) results in pseudoreplication (Hurlbert 1984). Without longitudinal data, few studies examining seasonality address this issue (although see Kratz et al.
1987; and see Hatch 2003 for how this issue effects statistical power). While sample size in this study did not allow us to use years as replicates (14 sightings per dolphin within each season of the same year, for multiple years is an extremely tall order), our long-term database allows us to substantially improve on previous studies by exploring seasonal shifts over a 24-year period.
Although here we use animals as replicates, all animals were sighted across multiple but different years (mean 14.64 ± SD 4.99, range 2 – 24), meaning that each animal represents a unique composite average of seasonal patterns, which along with permutation tests, drastically reduces pseudoreplication.
Three seasons were defined: breeding season when 85% of births and conceptions occur
(due to a 12 month gestation; September 1–January 15th), shark season when 86% of shark attacks occur (January 16th–May 15th), and the cold season when average temperatures drop below 21°C (May 16th–August 31st), few births occur, and large tiger sharks are virtually absent from the eastern gulf of Shark Bay (unpublished data; Mann et al. 2000; Heithaus 2001; Wirsing et al. 2006). Because the average tidal height differs between seasons (shark > cold > breeding), availability of habitat was recalculated for each season separately. Requiring animals to have at least 14 points per season reduced our sample such that one cluster had only 3 individuals.
Consequently, these three individuals were assigned to their closest cluster based on their GDM1 distance to the medoids of the remaining clusters, which did not increase any cluster’s width.
Within each cluster, a permutation repeated measures ANOVA was used to explore how season
23 and a sex-season interaction relate to habitat use and habitat use diversity, followed by multiple permutation tests for comparisons.
Home ranges
To examine how habitat use relates to home range size, we estimated home ranges using fixed kernel densities (Gaussian distribution) with a least-squares cross validation (LSCV) smoothing parameter (Worton 1989; Seaman & Powell 1996; adehabitatHR package in R,
Calenge 2006). We also employed a new adaptive local convex hull method (a-LoCoH; Getz &
Wilmers 2004; adehabitatHR package in R, Calenge 2006; Getz et al. 2007) that is known to outperform kernel density (KD) estimators, especially in areas with hard boundaries such as shoreline. A minimum of 50 location points were needed for each animal to be considered in the analyses because in a preliminary analysis similar to that of Urian et al. (2009), visual plots revealed that beyond 50 points home range sizes for KD estimates did not predictably change.
However, because a-LoCoH home ranges were still affected by sample size beyond 50 points, we selected a random subset of 50 points for each animal to construct home ranges for both methods (a-LoCoH and KD). Because we use the exact same number of location points to calculate each animal’s home range, any differences between sexes or habitat clusters in home range size cannot be due to differences in sample size. In order to determine an initial value for the adaptive a parameter used in estimating a-LoCoH home ranges, the average maximum distance between any two location points for each animal was used (Getz et al. 2007). Plots of the area of home range size vs. a range of a values indicated our above a value was appropriate for the majority of animals. Instead of the usual 95% isopleths routinely used when estimating
24 the boundary of home ranges, we choose to use 90% isopleths for both KD and a-LoCoH as these are known to be less biased to sample size (Börger et al. 2006; Getz et al. 2007). Land was removed from all final home range estimates. Our home range sample included 130 individuals,
59 females and 71 males. However, 7 of these females, all in the same habitat use cluster, were part of the regular provisioning program located at Monkey Mia (Mann & Kemps 2003) and likely have reduced home range sizes (Watson-Capps 2005). Comparing these females to other females within the same cluster using a permutation test revealed that provisioned females had significantly smaller home ranges. While these results are presented, provisioned females were removed from overall home range sex and habitat use cluster comparisons. For the remaining animals, sex, habitat use cluster, and habitat use cluster by sex interaction terms were tested using a permutation ANOVA, followed by multiple permutation tests for comparisons.
Statistical considerations
Since all data in this study deviated substantially from normality, permutation tests (1,000 randomizations) were used to determine significance in all analyses (Z tests, t-tests, Chi-square tests, ANOVA, etc.,) and Bonferroni adjustments were applied to p-values where appropriate to maintain an experiment-wise type I error rate of α = 0.05. All p-values greater than 0.80 after
Bonferroni correction are reported as p > 0.80. Permutations were performed either using the coin package in R (Hothorn et al. 2006) or, for more complicated randomizations and restrictions, custom permutations were written in R (Anderson 2001; Manly 2007; Bates et al.
2010; lme4 package in R for ANOVA tables, R Development Core Team 2011). We do not assume the data follow any particular distribution since test statistics were compared to
25 distributions created from the original data through randomizations. However, homoscedasticity, which is still a necessary assumption of permutation tests (Anderson 2001), was explored graphically using the HH package in R (Heiberger 2012) and no violations were observed.
RESULTS
Cluster and sex habitat use differences
The eigenanalysis of selection ratios from the type II design (Fig. 7) clearly demonstrated that Shark Bay bottlenose dolphins used habitats in a variety of different ways. Furthermore, the
2 chi-square test revealed individual differences in overall habitat use (χL = 15315.00, p = 0.0001,
1,000 permutations). PAM cluster analysis resulted in four clusters (based on the Caliński &
Harabasz (1974) index = 1042.44, Table 2) representing animals that heavily used Deep open habitat (the Deep open cluster), Channel habitat (the Channel cluster), or one of these deeper habitats and a variety of shallow habitats (Mixed Deep open and Mixed Channel clusters, Fig. 7).
Dolphins in all clusters selected habitat differently than expected based on availability,
2 not assuming identical habitat use (Deep open cluster: χL = 4911.63, p = 0.0001, Channel
2 2 cluster: χL = 3984.93, p = 0.0001, Mixed Deep open cluster: χL = 1389.10, p = 0.0001, Mixed
2 2 Channel cluster: χL = 1043.92, p = 0.0001, 1,000 permutations), although the much lower χL values for the two mixed clusters indicate less overall selection. Figure 8 displays how dolphins in each cluster used habitats in accordance with availability and compared to each other (see
Table 3 for statistics). In the Deep open cluster females selected Deep open habitat more than males (Fig. 8a, Z = 3.72, p = 0.0006), and males selected Channel habitat more than females
26 (Fig. 8a, Z = -5.51, p = 0.0001), while in the Channel cluster males selected Deep ecotone habitat more than females (Fig. 8b, Z = -3.61, p = 0.0006). In the Mixed Deep open cluster, males selected Channel habitat more than females (Fig. 8c, Z = -3.55, p = 0.0024) and there were no sex differences in habitat use for the Mixed Channel cluster.
For comparison, Fig. 9 shows habitat use at the population level when animal identities were not used, data were pooled from all observations, and clusters were not separated (type I design). Dolphins used habitats differently than expected based on availability, but here we must
2 assume individuals use habitat in a similar way (χL = 2561.34, p = 0.0001, 1,000 permutations).
Because animals were not treated as the sampling unit in this analysis, we could not test for sex differences in habitat use. From Fig. 9, one can see how, at the population level, habitats were selected in accordance with availability and compared to each other (see Table 4 for statistics).
Cluster and sex habitat use diversity differences
Both clusters and sexes varied in their diversity of habitat use. First, there was no significant interaction between sex and cluster so only the main effects of sex and cluster on habitat use diversity were examined. The Mixed Deep open cluster had the highest diversity (Fig.
10; Mixed Deep open–Mixed Channel cluster, Z = 4.94, p = 0.0001; Mixed Deep open–Deep open cluster, Z = 9.15, p = 0.0001; Mixed Deep open–Channel cluster, Z = 9.20, p = 0.0001;
1,000 permutations), followed by the Mixed Channel cluster (Fig. 10; Mixed Channel–Deep open cluster: Z = 5.70, p = 0.0001; Mixed Channel–Channel cluster: Z = 6.61, p = 0.0001; 1,000 permutations). The Deep open and Channel clusters had the lowest diversity indices and did not differ from each other (Fig. 10; Z = 1.35, p > 0.80). Finally, using cluster as a blocking factor
27 revealed that males had higher habitat use diversity than females (Males = 3.40 ± 0.11 (SE);
Females = 3.21 ± 0.12 (SE); Z = -3.00, p = 0.0026, 1,000 permutations). For comparison, without the cluster blocking factor no significant sex difference would have been detected (Z = -
1.12, p = 0.2549, 1,000 permutations), which highlights the importance of examining intraspecific variation in habitat use before attempting to explore population wide patterns.
Seasonal habitat use
In some habitats seasonal shifts occurred, but results depended greatly on cluster (Fig.
11). Too few animals had enough data to explore seasonal patterns in the Mixed Channel cluster
(see methods). No sex by season interactions were significant unless noted. In Channel habitat, all clusters examined showed seasonal changes in habitat use (Fig. 11a). Dolphins in both the
Deep open and the Mixed Deep open clusters selected Channel habitat more during the breeding and shark season compared to the cold season (Deep open cluster: breeding–cold, Z = 4.06, p =
0.0001; shark–cold, Z = -3.21, p = 0.0001; Mixed Deep open cluster: breeding–cold, Z = 3.90, p
= 0.0001; shark–cold, Z = -2.58, p = 0.0216; 1,000 permutations). However, dolphins in the
Channel cluster selected Channel habitat more in the shark season compared to both the breeding and cold seasons (shark–breeding, Z = -2.93, p = 0.0468; shark–cold, Z = -2.49, p = 0.0468;
1,000 permutations), so there appeared to be an overall tendency to use Channel habitat more during shark season
In Deep ecotone habitat there was a significant sex by season interaction for the Deep open cluster (Fig. 11b; F = 6.54, p = 0.036, 1,000 permutations). Females did not change their
Deep ecotone habitat use seasonally, while males selected Deep ecotone habitat more during the
28 breeding season than during either the cold or shark seasons (breeding–cold, Z = 2.77, p =
0.0144; breeding–shark, Z = 2.64, p = 0.0396; 1,000 permutations). However, in the Mixed Deep open cluster, both sexes selected Deep ecotone habitat the most during the breeding season
(breeding–cold, Z = 3.61, p = 0.0018; breeding–shark, Z = 3.13, p = 0.0162; 1,000 permutations). In contrast, for the Channel cluster, Deep ecotone habitat was selected more during the breeding season compared to only the shark season (Z = 2.85, p = 0.0306, 1,000 permutations).
Sand flats, which in general were selected at very low levels, were only selected seasonally by the Mixed Deep open cluster: Sand flats were selected more during the cold season compared to the breeding season (Fig. 11c; cold–breeding, Z = -2.86, p = 0.0486, 1,000 permutations). Likewise, this was the only cluster that changed their use of Seagrass beds and
Shallow ecotone habitat seasonally, which were both selected the most during the cold season
(Fig. 11d, Seagrass beds: cold–breeding, Z = -4.74, p = 0.0001; cold–shark, Z = 3.50, p =
0.0018; Fig. 11e, Shallow ecotone: cold–breeding, Z = -4.25, p = 0.0001; cold–shark, Z = 3.20, p
= 0.0144; 1,000 permutations). Finally, in Deep open habitat, two clusters showed seasonal shifts
(Fig. 11f). The Deep open cluster selected Deep open habitat more during the cold season compared to the breeding season (Z = -3.85, p = 0.00001, 1,000 permutations), while the Mixed
Deep open cluster selected Deep open habitat more during both the breeding and shark seasons compared to the cold season (breeding–cold: Z = 3.52, p = 0.0072; shark–cold: Z = -3.50, p =
0.0018; 1,000 permutations).
Seasonal habitat use diversity
29
There was no significant interaction between sex and season on habitat use diversity, but dolphins in all three clusters analyzed changed their diversity of habitat use seasonally (Fig. 12).
Both the Deep open and Channel clusters had higher diversity during the breeding season compared to the shark season (Deep open: Z = 2.38, p = 0.0489, 1,000 permutations; Channel cluster: Z = 2.71, p = 0.0123, 1,000 permutations). However, the Channel cluster also had higher diversity during the cold season compared to shark season (Z = 2.80, p = 0.0072, 1,000 permutations). In contrast, the Mixed Deep open cluster dolphins had the highest diversity of habitat use during the cold season (cold–breeding: Z = -2.37, p = 0.0468; cold–shark: Z = 2.71, p
= 0.0147; 1,000 permutations).
Home ranges
In general, the KD method estimated larger home sizes compared to a-LoCoH. The KD method also regularly included land in the home range estimates for individuals that come close to shore, where as a-LoCoH home ranges almost never included land, highlighting the advantage of using the a-LoCoH method in areas with hard habitat boundaries and/or true habitat holes.
The seven provisioned females at Monkey Mia (Mann & Kemps 2003) had smaller home ranges compared to non-provisioned females in the same cluster (Mixed Deep open), regardless of the home range estimation method (Fig. 13; a-LoCoH, Z = 2.36, p = 0.0123; KD, Z = 2.63, p =
0.0067; 1,000 permutations). For both home range methods there was a significant interaction between sex and cluster, suggesting habitat use affects male and female home ranges differently
(a-LoCoH: F = 6.15, p = 0.002, KD: F = 4.96, p = 0.005, 1,000 permutations). For females,
30 home range size did not depend on cluster membership (Fig. 14). However, males in the cluster with the highest diversity (Mixed Deep open cluster) had the smallest home ranges using the a-
LoCoH method (Fig. 14a; Mixed Deep open–Mixed Channel cluster, Z = -3.19, p = 0.0084;
Mixed Deep open–Deep open cluster, Z = -3.97, p = 0.0006; Mixed Deep open–Channel cluster:
Z = -2.80, p = 0.0264; 1,000 permutations), while males in all other clusters had similar home range sizes. In contrast, when using the KD method, males in the Mixed Deep open cluster no longer had smaller home ranges than males in the Channel cluster (Fig. 14b; Mixed Deep open–
Mixed Channel: Z = -3.22, p = 0.003, Mixed Deep open–Deep open: Z = -4.27, p = 0.0001;
Mixed Deep open–Channel cluster Z = -2.04, p = 0.2238, 1,000 permutations), and males in the
Deep open cluster had larger home ranges than males in the Channel cluster (Fig. 14b; Z = 2.78, p = 0.024, 1,000 permutations). Additionally, within the Deep open cluster, males had larger home ranges than females using both methods (a-LoCoH: Z = -3.99, p = 0.0001; KD: Z = -4.16, p = 0.0001,1,000 permutations). While no other clusters revealed significant sex differences,
Figure 14 demonstrates that the significant interaction between sex and cluster membership was driven by males differing their home range size between clusters, while females maintained a constant home range size regardless of their habitat use cluster. Furthermore, the overall pattern of males having larger home ranges than females is visible in all clusters except perhaps the
Mixed Deep open cluster. As such, sex differences in home range size were re-examined using cluster as a blocking factor, which revealed that males have larger home ranges than females, regardless of method (a-LoCoH: Z = -4.54, p = 0.0001; KD: Z = -4.93, p = 0.0001, 1,000 permutations)
DISCUSSION
31
A major aim of habitat use studies is not only to provide data on which habitats are used, but also which habitats are critical and should be a priority for conservation. Here we illustrate why examining intraspecific variation in habitat use and home ranges is vital for proper wildlife management and protection. As a population, the bottlenose dolphins in Shark Bay use a variety of habitats (Fig. 7), but this diversity is driven both by intra- and inter-individual variation in habitat use. Some dolphins specialize in a habitat and show little seasonal change, while others use a variety of habitats and adjust throughout the year. Failing to recognize these individual differences leads to patterns driven by the animals, or “type” of animals (cluster), most frequently encountered.
For example, if data are pooled from all animals, habitat use patterns are simply driven by the two clusters containing the most animals (Deep open and Mixed Deep open, Table 2, Fig.
8–9). Here one might conclude that Deep open habitat should be a priority for management and conservation (Fig. 8), while in reality, this habitat is important for many animals, but others
(Channel and Mixed Channel clusters) rarely use Deep open habitat and instead utilize the
Channels (Fig. 8). Shark Bay is most well known for two groups of dolphins that do not primarily use Deep open habitat: the provisioned Monkey Mia beach dolphins (Mann & Kemps
2003) and the sponge tool-using dolphins (spongers; Smolker et al. 1997; Krützen et al. 2005;
Mann et al. 2008, 2012; Patterson & Mann 2011). The Monkey Mia dolphins attract over 95% of the ~ 100,000 annual visitors to Shark Bay (a World Heritage Area, UNESCO) and support its multi-million dollar tourism industry (Stoeckl et al. 2005). The spongers have attracted the attention of both researchers and the public worldwide with their exquisite tool-use behavior
(e.g. Hotz 2008; Stringer 2011; Bhanoo 2011; Vergano 2012). Clearly, management requires
32 individual level knowledge on habitat use to properly safeguard these dolphins and, in turn, the
Shark Bay community’s livelihood and uniqueness.
Examining home ranges, which is only possible at the individual level, is also critical for exploring the effects human activities have on dolphin spatial patterns. Provisioned females at
Monkey Mia have drastically smaller home ranges compared to non-provisioned females in the same habitat use cluster (Fig. 13). This result, which has been noted before (Mann & Kemps
2003; Watson-Capps 2005), is not surprising given that these females visit the same beach every morning to receive fish handouts from tourists. While this may cause concern, especially since previous work demonstrated higher calf mortality in the provisioned group (Mann et al. 2000), recent data show that offspring of provisioned females are surviving at high rates once effective management was put in place (Foroughirad & Mann, in press). However, small home ranges may have detrimental effects that deserve consideration; home range size may affect a female’s sociality and social network, which could have long-term negative impacts (Foroughirad,
Stanton, Bejder & Mann, unpublished data).
The underlying cause of the observed individual differences in habitat use and home ranges likely relates to female foraging tactic and sex specific mating strategies. For example, male dolphins in the Deep open cluster use Channel habitat more, while females use Deep open
(Fig. 8a). Such sex specific habitat selection may relate to sexual conflict (e.g. female avoidance of aggressive male alliances, Scott et al. 2005) or intrasexual competition (e.g., avoidance between male groups). A similar pattern was not observed in the Channel cluster, perhaps because most of these females specialize in sponge foraging (61%) and are not able to shift their habitat use away from males (Fig. 8b). In fact, 40% of the dolphins in this cluster specialize in sponge foraging (14 females, 6 males), and while the relationship between habitat use and
33 foraging tactics in other clusters is less clear, previous research strongly indicates that many foraging tactics are tied to specific habitats (Sargeant et al. 2007).
Females are often characterized as “the ecological sex” because their fitness is directly tied to resources (Wrangham 1980), possibly making them more susceptible to anthropogenic disturbances (e.g., overfishing, habitat loss). In Shark Bay, females maintain smaller, constant home ranges regardless of their habitat use, and have lower habitat use diversity, which we believe is driven by an individual’s foraging and energetic demands. We suggest that females know their home ranges well and depend on this knowledge for survival, making them less capable of shifting their habitat use compared to males. While the smaller size of female home ranges may seem easier to conserve and manage, they differ greatly in terms of habitat composition due to intraspecific variation, which again highlights the need for investigation of habitat use at the individual level.
However, while habitat specialization may be possible and even beneficial for females, such behavior may be too costly for males. Male fitness is tied to the distribution of, or access to, females more than resources per se. In Shark Bay, male dolphins use a greater diversity of habitats and have larger home ranges than females (Fig. 10 and 14), which may be the result of social factors. Randic et al. (2012) found that individual male home ranges were almost as large as combined alliance home ranges (76.2 km2 for an individual, 92.3 km2 for alliances, individual range size differs from that reported here due to different methodologies). Furthermore, a male’s range size was positively correlated with his alliance size (number of partners), and an alliance’s range size was positively correlated with the number of males it contained (Randic et al. 2012).
So while females may only need to attend to their own (and possibly their offspring’s) demands in terms of habitat use and home range size, males appear to face trade offs between individual
34 level home range and habitat use needs and the needs of their alliance in order to successfully consort females (Connor et al. 2006; Randic et al. 2012). These same social factors may drive the increased habitat use diversity seen in males. Even so, habitat availability may place a limit on a male’s (and possibly his alliance’s) home range size, since males that primarily use habitats with low availability (e.g. Mixed Deep open cluster, Seagrass beds, Fig 8c, Fig. 14, Table 1) have smaller home ranges than those that utilize more abundant habitats (Deep open cluster,
Deep open; Fig. 8a, Fig. 14, Table 1).
Such sex differences are consistent with socio-ecological theory in that female habitat use and spatial patterns are driven primarily by ecological factors, while male habitat use is driven more by social factors and access to females (Wrangham 1980; Sterck et al. 1997). Although, because both sexes grouped together into clusters with similar sex ratios (Table 2), there doesn’t appear to be strong sex segregation in terms of habitat use, which is consistent with the open fission-fusion nature of dolphin societies (Connor et al. 2000).
Seasonal habitat use shifts may relate to mating and reproductive behavior (Mann et al.
2000), predation risk (Wirsing et al. 2006), temperature changes and thermoregulation (Ross &
Cockcroft 1990), and changes in prey communities and abundance (Heithaus 2004). Specialists show little seasonal shift compared to animals with diverse habitat use (Fig. 11). However, where seasonal shifts occur, they are subtle and would not cause animals to change cluster membership. During the breeding season, Deep ecotone habitat appears to be important, especially for males (Fig. 11b). One possible explanation is that males patrol this edge habitat in search of females. In Shark Bay, alliances might maximize access to females by shifting their habitat use to encounter females in both shallow and deep habitats. Such mate searching and patrolling by males is well documented in a variety of species (Parker 1978; Davies & Halliday
35 1979; Barnes 1982; Sherman 1989; Schwagmeyer 1994). During consortships, males may cause a shift in females’ use of Deep ecotone habitat, which would explain why females in all clusters, except one, exhibit the same seasonal pattern as males (Fig. 11b). Males are known to consort females for weeks to months (Connor et al. 1996) and may take females into habitats outside of their normal habitat use, which could impose additional direct costs of sexual conflict and warrants further investigation (see also Watson-Capps 2005).
Alternatively, females could be driving this shift. Females from deeper waters may move closer to this edge habitat when nursing newborn offspring, as here they may be able to forage in productive shallow habitats and reduce the amount of time newborn calves are unaccompanied at the surface. Males may then simply follow suit in their search for cycling females. Deep ecotone habitat may be crucial to early calf development, so while it is not the primary habitat of most females, it may be vital for successful reproduction. This highlights the need for information on not only how much habitats are used, but also why particular habitats are used since even those used infrequently may have unexpected fitness consequences. While previous investigations suggest that shallower depths correlate with an increase in reproductive success (Mann et al.
2000), future studies will provide essential data on the relationship between habitat use and female calving success and the underlying cause(s) of this seasonal shift.
In a previous study, Deep ecotone habitat (as defined here) was thought to be used during the shark season because it may provide dolphins with an easy escape from predators (Heithaus
& Dill 2006). However, Heithaus and Dill (2006) combined the breeding and shark seasons into one warm season. Our results suggest that these seasons are distinct, and that breeding, not predation risk, may drive increases in Deep ecotone habitat use during warm seasons. When analyzed separately, Deep ecotone habitat, which is preferred by tiger sharks (Heithaus et al.
36 2006), appears to be selected most during the breeding season and least during the shark season, suggesting possible predator avoidance (Fig. 11b). Accordingly, and similar to Heithaus and Dill
(2002), we found that during both the shark and breeding seasons, dolphins in the Mixed Deep open cluster decrease their use of shallow habitats, which are also preferred by tiger sharks (Fig.
11c-e, Heithaus et al. 2002). Dolphins in the Deep open and Channel clusters did not shift their use of shallow habitats, which suggests they are less responsive to changes in predator abundance because either, they prefer relatively safe habitats, or, due to their habitat specialization, they are less able to respond to a predator influx. Still, habitat use diversity for all clusters decreased in shark season, which may indicate dolphins are less willing to explore unfamiliar habitats when predation risk is high (Fig. 12). Future studies relating shark attack rates to individual differences in habitat use may reveal habitat specific predation risks.
The seasonal shifts in the Mixed Deep open cluster could also be explained by dolphins’ thermoregulation needs. Previous research assumed equal temperature across habitats based on surface temperature data (Heithaus 2001), but deeper in the water column temperatures decline significantly. So while the surface temperatures across habitats at mid-day may be similar, temperatures below the surface are certain to vary. This may be important for bottlenose dolphins in colder months where temperatures in Shark Bay fall below their thermo-neutral zone
(20–25 °C) and increased energy intake and heat conservation become important (Ross &
Cockcroft 1990). Shallow habitats might be more energetically efficient in the colder months, at least during the day when researchers are able to collect data. However, this proposed temperature response may only be possible for dolphins that already use shallow habitats to some degree, as dolphins in the two deeper clusters (Deep open and Channel) do not shift during cold
37 months. Thus, the seasonal change in temperature itself may explain habitat shifts in this and previous studies (Heithaus & Dill 2002, 2006, Mann & Watson-Capps 2005).
Finally, Heithaus (2004) found seasonal changes in prey communities in Shark Bay, which may explain some of the habitat shifts observed here. As previously mentioned, many individuals (particularly females) specialize in habitat specific foraging behaviors (Sargeant et al.
2007), some of which appear to be prey specific (e.g. Sargeant et al. 2005; Patterson & Mann
2011, Chapter 3) and show seasonal variation (Sargeant et al. 2007). Furthermore, some foraging tactics occur exclusively in either cold months or warm months (unpublished data). Thus, dolphins may shift habitat use seasonally to align foraging behavior with changes in prey communities, but more data on dolphin diets and the seasonal abundance of particular prey species across habitats are needed.
Regardless of the factor(s) driving these seasonal shifts, climate changes, such as an increase in water temperature, could potentially influence long-term shifts in habitat use.
However, because not all animals shift habitat use seasonally, the effects of climate change are probably not ubiquitous across the population. Thus, managers must consider how warming trends may differentially affect a dolphin’s mating behavior, predation risk, thermoregulation requirements, and food resources.
We hope this study will elicit future exploration of how habitat use relates to social, demographic, developmental, and ecological factors that influence delphinid behavioral ecology.
For example, while here it was assumed that an animal’s individual habitat use is not dependent on its associates, males may alter their habitat use in order to meet the demands of their alliance.
Even for females, it is currently unknown if an animal’s home range and habitat use affects its sociality, or alternatively, if its social preferences drive its home range and habitat use. We know
38 that foraging tactics have an impact on female sociality and relationships (Mann et al. 2012), so it would not be surprising if ranging and habitat use also played an important role in social dynamics. Resource competition, especially between females, is likely to be important and influence habitat use, even within clusters. Future studies will provide insight on how females minimize foraging competition and whether or not they partition habitat use, which would require fine scale temporal locational data. Additionally, previous studies suggest that daughters adopt their mothers foraging tactics more so than sons (Mann et al. 2008; Sargeant & Mann
2009). Although both sexes are philopatric (Tsai & Mann 2012), we do not know whether maternal habitat use is inherited and/or if any sex biases exist. However, Tsai and Mann (2012) did find that daughters moved further from their natal (pre-weaning) area than sons, possibly indicating resource competition with their mothers. Finally, although small cetaceans’ spatial movements are expected to be primarily driven by prey, bottlenose dolphins live in a dynamic fission-fusion society where behavioral context plays a critical role (e.g., dolphins can join up to socialize and separate to forage in different habitats). To understand whether dolphins are choosing habitats because of temperature, prey, predators, social factors, or some other reason(s), we need to examine the role of behavioral context in relation to habitat use based on an individual’s reproductive, social, foraging profile, or life history status. Together, these factors will help us understand why such strong individual differences are characteristic of bottlenose dolphins in Shark Bay, Australia.
39 ACKNOWLEDGEMENTS
We thank our field assistants Kippy Searles, Jenny Smith, and Katie Gill, all members of the
Mann lab past and present, and all members of the Shark Bay Dolphin Research Project
(SBDRP) who contributed to the long-term data. The DEM was created in collaboration with Dr.
Jordan A. Thomson of the Shark Bay Ecosystem Research Project (SBERP), using data compiled from SBDRP, SBERP, and the Department of Environment and Conservation (DEC).
We are grateful Dr. Wayne Getz and Dr. Brian Manly for their statistical and analytical expertise, which greatly improved the methodology of this study. We also thank SBERP for help in the field, and the Monkey Mia Dolphin Resort and DEC for field and logistical support.
Finally, we are indebt to Thomas Jorstad and the Smithsonian Natural History Museum for their assistance with sediment analysis. All animal work was approved by the Georgetown University
Animal Care and Use Committee (GUACUC) under permits 07–041 and 10–023. Observational and field studies were approved by DEC under permits SF007975, SF006897, and SF007457.
Support for this study was provided by Georgetown University, the National Geographic Society
Young Explorers Grant, the Explorers Club Exploration Fund Grant, the Achievement Rewards for College Scientist, the Animal Behavior Society Cetacean Behavior and Conservation Award, and the American Society of Mammologists Grant in Aid of Research to EMP. Additional funding was provided by the National Science Foundation (NSF) Grants 0847922, 0316800,
0918303, 0918308, Georgetown University, and Office of Naval Research Grant 10230702 to
JM. The orthophoto in Figure 1 and 2 was supported by NSF grant OCE0526065 to Dr. Mike
Heithaus and DEC.
40 REFERENCES
Aebischer, N. J., Robertson, P. A. & Kenward, R. E. 1993. Compositional analysis of babitat
use from animal radio-tracking data. Ecology, 74, 1313–1325.
Alldredge, J. R. & Griswold, J. 2006. Design and analysis of resource selection studies for
categorical resource variables. Journal of Wildlife Management, 70, 337–346.
Alldredge, J. R., Thomas, D. L. & McDonald, L. L. 1998. Survey and comparison of methods
for study of resource selection. Journal of Agricultural, Biological, and Environmental
Statistics, 3, 237–253.
Allen, S. J. & Loneragan, N. 2010. Reducing dolphin bycatch in the pilbara finfish trawl
fishery. Report to the Fisheries Research & Development Corporation, Canberra, ACT, 66.
Allen, M. C. & Read, A. J. 2000. Habitat selection of foraging bottlenose dolphins in relation to
boat density near Clearwater, Florida. Marine Mammal Science, 16, 815–824.
Allen, M. C., Read, A. J., Gaudet, J. & Sayigh, L. S. 2001. Fine-scale habitat selection of
foraging bottlenose dolphins Tursiops truncatus near Clearwater, Florida. Marine Ecology
Progress Series, 222, 253–264.
Anderson, M. J. 2001. Permutation tests for univariate or multivariate analysis of variance and
regression. Canadian Journal of Fisheries and Aquatic Sciences, 58, 626–639.
Bailey, H. & Thompson, P. M. 2010. Effect of oceanographic features on fine-scale foraging
movements of bottlenose dolphins. Marine Ecology Progress Series, 418, 223–233.
Barnes, R. F. W. 1982. Mate searching behaviour of elephant bulls in a semi-arid environment.
Animal Behaviour, 30, 1217–1223.
Bates, D., Maechler, M. & Bolker, B. 2011. lme4: Linear mixed-effects models using S4
classes. R package version 0.999999-0, http://cran.r-project.org/package=lme4.
41 Bearzi, G., Holcer, D. & Notarbartolo di Sciara, G. 2004. The role of historical dolphin takes
and habitat degradation in shaping the present status of northern Adriatic cetaceans. Aquatic
Conservation: Marine and Freshwater Ecosystems, 14, 363–379.
Bearzi, G., Politi, E., Agazzi, S. & Azzellino, A. 2006. Prey depletion caused by overfishing
and the decline of marine megafauna in eastern Ionian Sea coastal waters (central
Mediterranean). Biological Conservation, 127, 373–382.
Bejder, L., Samuels, A., Whitehead, H., Gales, N., Mann, J., Connor, R., Heithaus, M.,
Watson-Capps, J., Flaherty, C. & Krützen, M. 2006. Decline in relative abundance of
bottlenose dolphins exposed to long-term disturbance. Conservation Biology, 20, 1791–
1798.
Bhanoo, S. N. 2011 July 25. Dolphin tool helps to find fare on seafloor. New York Times, D3.
Bossart, G. D. 2006. Marine mammals as sentinel species for oceans and human health.
Oceanography, 19, 134–137.
Buij, R., Van Der Goes, D., De Iongh, H. H., Gagare, S., Haccou, P., Komdeur, J. & De
Snoo, G. 2012. Interspecific and intraspecific differences in habitat use and their
conservation implications for Palaearctic harriers on Sahelian wintering grounds. Ibis, 154,
96–110.
Burling, M. C., Pattiaratchi, C. B. & Ivey, G. N. 2003. The tidal regime of Shark Bay,
Western Australia. Estuarine, Coastal and Shelf Science, 57, 725–735.
Börger, L., Franconi, N., De Michele, G., Gantz, A., Meschi, F., Manica, A., Lovari, S. &
Coulson, T. 2006. Effects of sampling regime on the mean and variance of home range size
estimates. The Journal of Animal Ecology, 75, 1393–1405.
42 Caceres, M. De. 2011. vegclust: Fuzzy clustering of vegetation data. R package version 1.5.1,
http://cran.r-project.org/package=vegclust.
Calenge, C. 2006. The package “adehabitat” for the R software: a tool for the analysis of space
and habitat use by animals. Ecological Modelling, 197, 516–519.
Calenge, C. & Dufour, A. B. 2006. Eigenanalysis of selection ratios from animal radio-tracking
data. Ecology, 87, 2349–2355.
Caliński, T. & Harabasz, J. 1974. A dendrite method for cluster analysis. Communications in
Statistics, 3, 37–41.
Cañadas, A., Sagarminaga, R., De Stephanis, R., Urquiola, E. & Hammond, P. S. 2005.
Habitat preference modelling as a conservation tool: proposals for marine protected areas
for cetaceans in southern Spanish waters. Aquatic Conservation: Marine and Freshwater
Ecosystems, 15, 495–521.
Chilvers, B. L., Corkeron, P. J. & Puotinen, M. L. 2003. Influence of trawling on the
behaviour and spatial distribution of Indo-Pacific bottlenose dolphins (Tursiops aduncus) in
Moreton Bay, Australia. Canadian Journal of Zoology, 81, 1947–1955.
Conde, D. A., Colchero, F., Zarza, H., Christensen, N. L., Sexton, J. O., Manterola, C.,
Chávez, C., Rivera, A., Azuara, D. & Ceballos, G. 2010. Sex matters: modeling male and
female habitat differences for jaguar conservation. Biological Conservation, 143, 1980–
1988.
Conner, L. M., Smith, M. M. D. & Burger, L. W. 2003. A comparison of distance-based and
classification-based analyses of habitat use. Ecology, 84, 526–531.
Connor, R. C. 2010. Cooperation beyond the dyad: on simple models and a complex society.
Philosophical Transactions of the Royal Society B: Biological Sciences, 365, 2687–2697.
43 Connor, R. C., Richards, A. F., Smolker, R. A. & Mann, J. 1996. Patterns of female
attractiveness in Indian Ocean bottlenose dolphins. Behaviour, 133, 37–69.
Connor, R. C., Wells, R. S., Mann, J. & Read, A. J. 2000. The bottlenose dolphin: social
relationships in a fission-fusion society. In: Cetacean Societies, (Ed. by J. Mann, R. C.
Connor, P. L. Tyack, & H. Whitehead), pp. 91–125. Chicago: University of Chicago Press.
Connor, R. C., Smolker, R. & Bejder, L. 2006. Synchrony, social behaviour and alliance
affiliation in Indian Ocean bottlenose dolphins, Tursiops aduncus. Animal Behaviour, 72,
1371–1378.
Constantine, R. 2001. Increased avoidance of swimmers by wild bottlenose dolphins (Tursiops
truncatus) due to long-term exposure to wwim-with-dolphin tourism. Marine Mammal
Science, 17, 689–702.
Davies, N. B. & Halliday, T. R. 1979. Competitive mate searching in male common toads, Bufo
bufo. Animal Behaviour, 27, 1253–1267. du Toit, J. T. & Owen-Smith, N. 1989. Body size, population metabolism, and habitat
specialization among large African herbivores. The American Naturalist, 133, 736–740.
Díaz López, B. 2006. Interactions between Mediterranean bottlenose dolphins (Tursiops
truncatus) and gillnets off Sardinia, Italy. ICES Journal of Marine Science, 63, 946–951.
Díaz López, B. & Shirai, J. A. B. 2008. Marine aquaculture and bottlenose dolphins’ (Tursiops
truncatus) social structure. Behavioral Ecology and Sociobiology, 62, 887–894.
ESRI. 2011. ArcGIS Desktop. Version 10. Redlands, CA: Environmental Systems Research
Institute.
Foroughirad, V. & Mann, J. in press. Long-term impacts of provisioning on the behavior and
survival of bottlenose dolphins (Tursiops sp.). Biological Conservation
44 Frère, C. H., Krützen, M., Mann, J., Connor, R. C., Bejder, L. & Sherwin, W. B. 2010.
Social and genetic interactions drive fitness variation in a free-living dolphin population.
Proceedings of the National Academy of Sciences of the United States of America, 107,
19949–19954.
Garshelis, D. L. 2000. Delusions in habitat evaluation: measuring use, selection, and
importance. In: Research Techniques in Animal Ecology: Controversies and Consequences,
(Ed. by L. Boitani & T. K. Fuller), pp. 111–164. New York: Columbia University Press.
Getz, W. M. & Wilmers, C. C. 2004. A local nearest-neighbor convex-hull construction of
home ranges and utilization distributions. Ecography, 4, 489–505.
Getz, W. M., Fortmann-Roe, S., Cross, P. C., Lyons, A. J., Ryan, S. J. & Wilmers, C. C.
2007. LoCoH: nonparameteric kernel methods for constructing home ranges and utilization
distributions. PloS ONE, 2, e207.
Gibson, Q. A. & Mann, J. 2008. The size, composition and function of wild bottlenose dolphin
(Tursiops sp.) mother–calf groups in Shark Bay, Australia. Animal Behaviour, 76, 375–387.
Gibson, Q. A. & Mann, J. 2009. Do sampling method and sample size affect basic measures of
dolphin sociality? Marine Mammal Science, 25, 187–198.
Harzen, S. & Brunnick, B. J. 1997. Skin disorders in bottlenose dolphins (Tursiops truncatus),
resident in the Sado estuary, Portugal. Aquatic Mammals, 23, 59–68.
Hastie, G. D., Wilson, B., Wilson, L. J., Parsons, K. M. & Thompson, P. M. 2004. Functional
mechanisms underlying cetacean distribution patterns: hotspots for bottlenose dolphins are
linked to foraging. Marine Biology, 144, 397–403.
Hatch, S. A. 2003. Statistical power for detecting trends with applications to seabird monitoring.
Biological Conservation, 111, 317–329.
45 Heiberger, R. M. 2012. HH: statistical analysis and data display: Heiberger and Holland. R
package version 2.3-17, http://cran.r-project.org/package=HH.
Heithaus, M. R. 2001. The biology of tiger sharks, Galeocerdo cuvier, in Shark Bay, Western
Australia: sex ratio, size distribution, diet, and seasonal changes in catch rates.
Environmental Biology of Fishes, 61, 25–36.
Heithaus, M. R. 2004. Fish communities of subtropical seagrass meadows and associated
habitats in Shark Bay, Western Australia. Bulletin of Marine Science, 75, 79–99.
Heithaus, M. R. & Dill, L. M. 2002. Food availability and tiger shark predation risk influence
bottlenose dolphin habitat use. Ecology, 83, 480–491.
Heithaus, M. R. & Dill, L. M. 2006. Does tiger shark predation risk influence foraging habitat
use by bottlenose dolphins at multiple spatial scales? Oikos, 2, 257–264.
Heithaus, M. R., Dill, L. M., Marshall, G. & Buhleier, B. 2002. Habitat use and foraging
behavior of tiger sharks (Galeocerdo cuvier) in a seagrass ecosystem. Marine Biology, 140,
237–248.
Heithaus, M. R., Hamilton, I. M., Wirsing, A. J. & Dill, L. M. 2006. Validation of a
randomization procedure to assess animal habitat preferences: microhabitat use of tiger
sharks in a seagrass ecosystem. Journal of Animal Ecology, 75, 666–676.
Hennig, C. 2010. fpc: Flexible procedures for clustering. R package version 2.1-4, http://cran.r-
project.org/package=fpc.
Hooker, S. 2004. Marine reserves as a tool for ecosystem-based management: the potential
importance of megafauna. BioScience, 54, 27–40.
Hothorn, T., Hornik, K., van de Wiel, M. A. & Zeileis, A. 2006. A lego system for conditional
inference. American Statistician, 60, 257–263.
46 Hotz, R. L. 2008 December 25. Among Dolphins, Tool-Using Handymen Are Women. Wall
Street Journal, A9.
Hoyt, E. 2005. Marine Protected Areas for Whales, Dolphins and Porpoises. A World Handbook
for Cetacean Habitat Conservation. London: Earthscan.
Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments.
Ecological Monographs, 54, 187–211.
Johnson, D. H. 1980. The comparison of usage and availability measurements for evaluating
resource preference. Ecology, 61, 65–71.
Jajuga, K., Walesiak, M. & Bak, A. 2003. On the general distance measure. In: Exploratory
data analysis in empirical research, (Ed. by M. Schwaiger & O. Opitz), pp. 104–109.
Berlin: Springer-Verlag.
Kaufman, L. & Rousseeuw, P. J. 2005. Finding Groups in Data: An Introduction to Cluster
Analysis. Hoboken: John Wiley & Sons, Inc.
Kratz, T. K., Frost, T. M. & Magnuson, J. J. 1987. Inferences from spatial and temporal
variability in ecosystems: long-term zooplankton data from lakes. The American Naturalist,
129, 830–846.
Krützen, M., Mann, J., Heithaus, M. R., Connor, R. C., Bejder, L. & Sherwin, W. B. 2005.
Cultural transmission of tool use in bottlenose dolphins. Proceedings of the National
Academy of Sciences of the United States of America, 102, 8939–8943.
Lutus, P. 2011. JTides. http://www.arachnoid.com/JTides/.
Maechler, M., Rousseeuw, P. J., Struyf, A., Hubert, M. & Hornik, K. 2012. cluster: Cluster
analysis basics and extensions. R package version 1.14.3, http://cran.r-
project.org/package=cluster.
47 Manly, B. F. J. 2007. Randomization, Bootstrap, and Monte Carlo Methods in Biology. 3rd edn.
Boca Raton: CRC Press.
Manly, B. F. J., Mcdonald, L. L., Thomas, D. L., Mcdonald, T. L. & Erickson, W. P. 2002.
Resource Selection by Animals Statistical Design and Analysis for Field Second Edition.
2nd edn. Dordrecht: Kluwer Academic Press.
Mann, J. 1999. Behavioral sampling methods for cetaceans: a review and critique. Marine
Mammal Science, 15, 102–122.
Mann, J. & Kemps, C. 2003. The effects of provisioning on maternal care in wild bottlenose
dolphins, Shark Bay, Australia. In: Marine Mammals and Humans: Towards a Sustainable
Balance, (Ed. by N. J. Gales, M. A. Hindell, & R. Kirkwood), pp. 304–320. Collingwood:
CSIRO.
Mann, J. & Sargeant, B. L. 2003. Like mother, like calf: the ontogeny of foraging traditions in
wild Indian ocean bottlenose dolphins (Tursiops sp.). In: The Biology of Traditions: Models
and Evidence, (Ed. by D. Fragaszy & S. Perry), pp. 236–266. Cambridge: Cambridge
University Press.
Mann, J. & Watson-Capps, J. J. 2005. Surviving at sea: ecological and behavioural predictors
of calf mortality in Indian Ocean bottlenose dolphins, Tursiops sp. Animal Behaviour, 69,
899–909.
Mann, J., Connor, R. C., Barre, L. M. & Heithaus, M. R. 2000. Female reproductive success
in bottlenose dolphins (Tursiops sp.): life history, habitat, provisioning, and group-size
effects. Behavioral Ecology, 11, 210–219.
Mann, J., Sargeant, B. L., Watson-Capps, J. J., Gibson, Q. A., Heithaus, M. R., Connor, R.
C. & Patterson, E. M. 2008. Why do dolphins carry sponges? PloS ONE, 3, e3868.
48 Mann, J., Stanton, M. A., Patterson, E. M., Bienenstock, E. J. & Singh, L. O. 2012. Social
networks reveal cultural behaviour in tool-using using dolphins. Nature Communications, 3,
980.
McDonald, L. L., Alldredge, J. R. & Boyce, M. S. 2005. Measuring availability and vertebrate
use of terrestrial habitats and foods. In: Techniques for Wildlife Investigations and
Management, 6th edn. (Ed. by C. E. Braun), pp. 465–488. Bethesda: The Wildlife Society.
McHugh, K. A., Allen, J. B., Barleycorn, A. A. & Wells, R. S. 2011. Natal philopatry, ranging
behavior, and habitat selection of juvenile bottlenose dolphins in Sarasota Bay, Florida.
Journal of Mammalogy, 92, 1298–1313.
Meentemeyer, V. & Box, E. O. 1987. Scale effects in landscape studies. In: Landscape
Heterogeneity and Disturbance, (Ed. by M. G. Tuner), pp. 15–34. New York: Springer-
Verlag.
Millspaugh, J. J., Skalski, J. R., Kernohan, B. J., Raedeke, K. J., Brundige, G. C. &
Cooper, A. B. 1998. Some comments on spatial independence in studies of resource
selection. Wildlife Society Bulletin, 26, 232–236.
Palatitz, P., Fehérvári, P., Solt, S., Kotymán, L., Neidert, D. & Harnos, A. 2011. Exploratory
analyses of foraging habitat selection of the red-footed falcon. Acta Zoologica Academiae
Scientiarum Hungaricae, 57, 255–268.
Parker, G. 1978. Evolution of competitive mate searching. Annual Review of Entomology, 23,
173–196.
Patterson, E. M. & Mann, J. 2011. The ecological conditions that favor tool use and innovation
in wild bottlenose dolphins (Tursiops sp.). PloS ONE, 6, e22243.
49 R Development Core Team. 2011. R: A Language and Environment for Statistical Computing,
Version 2.12.1. Vienna, Austria: R Foundation for Statistical Computing.
Randic, S., Connor, R. C., Sherwin, W. B. & Krützen, M. 2012. A novel mammalian social
structure in Indo-Pacific bottlenose dolphins (Tursiops sp.): complex male alliances in an
open social network. Proceedings of the Royal Society B: Biological Sciences, 279, 3083–
3090.
Ross, G. J. B. & Cockcroft, V. G. 1990. Comments on Australian bottlenose dolphins and the
taxonomic status of Tursiops aduncus (Ehrenberg, 1832). In: The Bottlenose Dolphinolphin,
(Ed. by S. Leatherwood & R. R. Reeves), pp. 101–128. San Diego: Academic Press.
Sargeant, B. L. & Mann, J. 2009. Developmental evidence for foraging traditions in wild
bottlenose dolphins. Animal Behaviour, 78, 715–721.
Sargeant, B. L., Mann, J., Berggren, P. & Krützen, M. 2005. Specialization and development
of beach hunting, a rare foraging behavior, by wild bottlenose dolphins (Tursiops sp.).
Canadian Journal of Zoology, 83, 1400–1410.
Sargeant, B. L., Wirsing, A. J., Heithaus, M. R. & Mann, J. 2007. Can environmental
heterogeneity explain individual foraging variation in wild bottlenose dolphins (Tursiops
sp.)? Behavioral Ecology and Sociobiology, 61, 679–688.
Schwagmeyer, P. L. 1994. Competitive mate searching in thirteen-lined ground squirrels
(Mammalia, Sciuridae): potential roles of spatial memory. Ethology, 98, 265–276.
Scott, M. D., Wells, R. S. & Irivine, A. B. 1990. A long-term study of bottlenose dolphins on
the west coast of florida. In: The Bottlenose dolphin, (Ed. by S. Leatherwood & R. Reeves),
pp. 235–244. San Diego: Academic Press.
50 Scott, E. M., Mann, J., Watson-Capps, J. J., Sargeant, B. L. & Connor, R. C. 2005.
Aggression in bottlenose dolphins: evidence for sexual coercion, male-male competition,
and female tolerance through analysis of tooth-rake marks and behaviour. Behaviour, 142,
21–44.
Seaman, D. E. & Powell, R. A. 1996. An evaluation of the accuracy of kernel density estimators
for home range analysis. Ecology, 77, 2075–2085.
Shane, S. H., Wells, R. S. & Wursig, B. 1986. Ecology, behavior and social organization of the
bottlenose dolphin: a review. Marine Mammal Science, 2, 34–63.
Shannon, C. E. & Weaver, W. 1949. The Mathematical Theory of Communication. Urbana:
University of Illinois Press.
Sherman, P. W. 1989. Mate guarding as paternity insurance in Idaho ground squirrels. Nature,
338, 418–420.
Smolker, R. A., Richards, A. F., Connor, R. C. & Pepper, J. W. 1992. Sex differences in
patterns of association among Indian Ocean bottlenose dolphins. Behaviour, 123, 38–69.
Smolker, R. A., Richards, A. F., Connor, R. C., Mann, J. & Berggren, P. 1997. Sponge
carrying by dolphins (Delphinidae, Tursiops sp.): a foraging specialization involving tool
use? Ethology, 103, 454–465.
Stanton, M. A., Gibson, Q. A. & Mann, J. 2011. When mum’s away: a study of mother and
calf ego networks during separations in wild bottlenose dolphins (Tursiops sp.). Animal
Behaviour, 82, 405–412.
Sterck, E. H. M., Watts, D. P. & van Schaik, C. P. 1997. The evolution of female social
relationships in nonhuman primates. Behavioral Ecology and Sociobiology, 41, 291–309.
51 Stoeckl, N., Smith, A., Newsome, D., Lee, D. & Newsom, D. 2005. Regional economic
dependence on iconic wildlife tourism: case studies of Monkey Mia and Hervey Bay. The
Journal of Tourism Studies, 16, 69–81.
Stringer, N. 2011. The Dolphins of Shark Bay. BBC Natural World/Mutual of Omaha’s Wild
Kingdom.
Thomas, D. L. & Taylor, E. J. 1990. Study designs and tests for comparing resource use and
availability. The Journal of Wildlife Management, 54, 322–330.
Thomas, D. L. & Taylor, E. J. 2006. Study designs and tests for comparing resource use and
availability II. Journal of Wildlife Management, 70, 324–336.
Torres, L. G., Read, A. J. & Halpin, P. 2008. Fine-scale habitat modeling of a top marine
predator: do prey data improve predictive capacity? Ecological Applications, 18, 1702–17.
Tsai, Y.-J. J. & Mann, J. 2012. Dispersal, philopatry, and the role of fission-fusion dynamics in
bottlenose dolphins. Marine Mammal Science, 28, doi: 10.1111/j.1748–7692.2011.00559.x.
Twiner, M. J., Fire, S., Schwacke, L., Davidson, L., Wang, Z., Morton, S., Roth, S., Balmer,
B., Rowles, T. K. & Wells, R. S. 2011. Concurrent exposure of bottlenose dolphins
(Tursiops truncatus) to multiple algal toxins in Sarasota Bay, Florida, USA. PloS ONE, 6,
e17394.
Urian, K. W., Hofmann, S., Wells, R. S. & Read, A. J. 2009. Fine-scale population structure
of bottlenose dolphins (Tursiops truncatus) in Tampa Bay, Florida. Marine Mammal
Science, 25, 619–638.
Vergano, D. 2012 July 31. “Sponger” dolphins like to keep each other company. USA Today,
52 Walesiak, M. 2006. Uogólniona Miara Odległości w Statystycznej Analizie Wielowymiarowej
[The Generalized Distance Measure in Multivariate Statistical Analysis]. Wroclaw:
Wydawnictwo Akademii Ekonomicznej.
Walesiak, M. 2008. Cluster analysis with ClusterSim computer program and R environment.
Acta Universitatis Lodziniensis, Folia Oeconomica, 216, 303–311.
Walker, D. L., Kendrick, G. A. & McComb, A. J. 1988. The distribution of seagrass species in
shark bay, Western Australia, with notes on their ecology. Aquatic Botany, 30, 305–317.
Watson-Capps, J. J. 2005. Female mating behavior in the context of sexual coercion and female
ranging behavior of bottlenose dolphins (Tursiops sp.) in Shark Bay, Western Australia.
Ph.D. thesis. Washington, D.C. U.S.A.: Georgetown University, Department of Biology.
Watson-Capps, J. J. & Mann, J. 2005. The effects of aquaculture on bottlenose dolphin
(Tursiops sp.) ranging in Shark Bay, Western Australia. Biological Conservation, 124, 519–
526.
Wells, R. S., Rhinehart, H. L., Hansen, L. J., Sweeney, J. C., Townsend, F. I., Stone, R.,
Casper, D. R., Scott, M. D., Hohn, A. A. & Rowles, T. K. 2004. Bottlenose dolphins as
marine ecosystem sentinels: developing a health monitoring system. EcoHealth, 1, 246–
254.
Wells, R. S., Tornero, V., Borrell, A., Aguilar, A., Rowles, T. K., Rhinehart, H. L.,
Hofmann, S., Jarman, W. M., Hohn, A. A. & Sweeney, J. C. 2005. Integrating life-
history and reproductive success data to examine potential relationships with
organochlorine compounds for bottlenose dolphins (Tursiops truncatus) in Sarasota Bay,
Florida. The Science of the Total Environmentotal, 349, 106–119.
53 Wentworth, C. 1922. A scale of grade and class terms for clastic sediments. The Journal of
Geology, 30, 377–392.
Wilson, B., Thompson, P. M. & Hammond, P. S. 1997. Habitat use by bottlenose dolphins:
seasonal distribution and stratified movement patterns in the Moray Firth, Scotland. The
Journal of Applied Ecology, 34, 1365–1374.
Wilson, B., Arnold, H., Bearzi, G., Fortuna, C. M., Gaspar, R., Ingram, S., Liret, C.,
Pribanić, S., Read, A. J., Ridoux, V., Schneider, K., Urian, K. W., Wells, R. S., Wood,
C., Thompson, P. M. & Hammond, P. S. 1999. Epidermal diseases in bottlenose dolphins:
impacts of natural and anthropogenic factors. Proceedings of the Royal Society B:
Biological Sciences, 266, 1077–1083.
Wirsing, A. J., Heithaus, M. R. & Dill, L. M. 2006. Tiger shark (Galeocerdo cuvier)
abundance and growth in a subtropical embayment: evidence from 7 years of standardized
fishing effort. Marine Biology, 149, 961–968.
Worton, B. J. J. 1989. Kernel methods for estimating the utilization distribution in home-range
studies. Ecology, 70, 164–168.
Wrangham, R. W. 1980. An Ecological Model of Female-Bonded Primate Groups. Behaviour,
75, 262–300.
Würsig, B. & Würsig, M. 1977. The photographic determination of group size, composition,
and stability of coastal porpoises (Tursiops truncatus). Science, 198, 755–756.
54 FIGURES AND TABLES
Figure 1. Map of study area. This figure shows our boat-launching site (Monkey Mia), transect locations for habitat assessment, and habitat classifications.
55
Figure 2. Zonal histogram of depths for habitat types. This figure represents a histogram of depths by habitat type for the entire ~130 km2 study area (using 10 m2 cells). As one would expect, Deep open and Channel habitats appear to be similar in depth (also see Fig. 4). Seagrass beds tend to be fairly shallow, while Sand flats and Shallow ecotone habitats are on the shallow extreme. Deep ecotone appears to fall somewhere in between deep habitats and shallow habitats as one would expect by design.
56
Figure 3. Zonal histogram of slopes for habitat types. This figure represents a histogram of slopes by habitat type for the entire ~130 km2 study area. Habitat types appear to be less distinct in terms of their slopes (also see Fig. 5). However, one goal of utilizing ecotone habitat types was to capture areas of increased slope, as can be seen with the distribution of Deep ecotone slopes being skewed the to right.
57
Figure 4. Habitat depths. This figure displays the average zonal depth (over the entire ~130 km2 study area) along with standard deviations. Bars with different letters indicate significant differences were found between habitat types in the simulation performed, p < 0.05.
58
Figure 5. Habitat slopes. This figure displays the average zonal slope (over the entire ~130 km2 study area) along with standard deviations. Bars with different letters indicate significant differences were found between habitat types in the simulation performed, p < 0.05.
59
Figure 6. Habitat current. This figure displays the average surface current for habitat types.
Because surface measurements were only taken opportunistically, and because shallow habitats are unlikely to drastically differ in terms of current velocity since topology is relatively constant, all shallow habitat measurements were grouped together. Bars with different letters indicate significant differences in surface current velocity as tested by permutation t-tests with a
Bonferroni adjustment, p < 0.05.
60
Figure 7. Duality diagram of the eigenanalysis of habitat use selection ratios of bottlenose dolphins in Shark Bay, Western Australia. The top figure shows habitat loadings on the first two factorial axes (Deep ecotone abbreviated as D.e.), while below animals scores are displayed on the first two factorial axes. The origin of the habitat-loading figure corresponds to a habitat being used in accordance with availability by all animals, while the origin of the animal-scores figure corresponds to an animal using all habitats randomly. Bottlenose dolphins fall into one of four groups based on the PAM clustering of selection ratios: the Deep open, Channel, Mixed Deep open, or Mixed Channel cluster.
61
Figure 8. Cluster habitat selection ratios from type II design. Mean selection ratios (± SE) for male and female bottlenose dolphins in the Deep open cluster (a), the Channel cluster (b), the
Mixed Deep open cluster (c), and the Mixed Channel cluster (d). Arrows before habitat classifications indicate whether a habitat is selected significantly more or less than expected based on availability (a selection ratio of 1 indicates random use in accordance with availability).
Lettering indicates significant differences between habitat types, while a * within habitat types indicates a significant difference between sexes, p < 0.05. The follow abbreviations were used:
D. ecotone–Deep ecotone, S. flats–Sand flats, S. beds–Seagrass beds, S. ecotone–Shallow ecotone, and D. open–Deep open. Statistical tests are displayed in Table 3.
62
Figure 9. Population level selection ratios from type I design. Mean selection ratios (± SE).
Arrows before habitat classifications indicate whether a habitat is selected significantly more or less than expected based on availability (a selection ratio of 1 indicates random use in accordance with availability). Lettering indicates significant differences between habitat types, p
< 0.05. The follow abbreviations were used: D. ecotone–Deep ecotone, S. flats–Sand flats, S. beds–Seagrass beds, S. ecotone–Shallow ecotone, and D. open–Deep open. Statistical tests are displayed in Table 4.
63
Figure 10. Male and female habitat use diversity by cluster. Mean Shannon’s diversity index (±
SE) of selection ratios for male and female bottlenose dolphins in the Deep open, Channel,
Mixed Deep open, and Mixed Channel clusters. Lettering indicates significant differences between clusters, p < 0.05. Males had higher habitat use diversity than females when blocking by cluster: males = 3.40 ± 0.11, females = 3.21 ± 0.12.
64
Figure 11. Seasonal habitat selection ratios. Mean seasonal selection ratios (± SE) in Channel
(a), Deep ecotone (b), Sand flats (c), Seagrass beds (d), Shallow ecotone (e), and Deep open (f) habitats for bottlenose dolphins in the Deep open, Channel, and Mixed Channel clusters.
Dolphins in the Deep open cluster were broken down by sex (Deep open m. – Deep open males,
Deep open f. – Deep open females) for Deep ecotone habitat because of a significant sex by season interactions. A * within cluster indicates a significant difference between seasons, p <
0.05. 65
Figure 12. Seasonal habitat use diversity. Mean Shannon’s diversity index (± SE) of selection ratios for bottlenose dolphins in the Deep open, Channel, and Mixed Deep open clusters. A * within cluster indicates a significant difference between seasons, p < 0.05.
66
Figure 13. Provisioned and non-provisioned female home range sizes in the Mixed Deep open cluster. Mean home range size (± SE) for both a-LoCoH and Kernel Density home range estimates. A * within home range estimation method indicates a significant difference, p < 0.05.
67
Figure 14. Male and female home range sizes by cluster. Mean home range size (± SE) for both a-LoCoH (a) and Kernel Density (b) home range estimates for male and female bottlenose dolphins in the Deep open, Channel, Mixed Deep open, and Mixed Channel. Lowercase letters indicate significant differences between females in different clusters, while upper case letters indicate significant differences between males in different clusters. A * within a cluster indicates a significant difference between sexes. 68 Table 1. Habitat classifications. Details of the six different habitat classification used in this study.
Habitat Area Tide corrected area Depth ± SD Current ± SD Slope ± SD Substrate (km2) (km2) (m) (cm/s) (%) Channel 37.72 37.72 7.13 ± 1.33 37.00 ± 11.00 0.63 ± 0.70 Rock, shell, coral debris, w/sand and seagrass and sand patches Deep ecotone 8.58 8.58 4.13 ± 1.17 NA 1.58 ± 1.38 Mix of all substrates Sand flats 10.70 4.46 -0.11 ± 0.68 8.78 ± 2.51 0.22 ± 0.31 Continuous sand Seagrass beds 25.13 24.14 2.00 ± 1.48 8.78 ± 2.51 0.87 ± 0.85 Continuous seagrass Shallow ecotone 9.99 8.33 0.74 ±1.26 8.78 ± 2.51 0.60 ± 0.73 Seagrass and sand Deep open 35.44 35.44 6.56 ± 1.66 11.26 ± 10.77 0.44 ± 0.46 Sand/silt/clay sediments
69
Table 2. Habitat use clusters. Demographic break down of the four habitat use clusters.
Cluster Females Males Sex unknown Sex ratio (F/M) Total dolphins Deep open 35 40 1 0.88 76 Channel 23 27 0 0.85 50 Mixed Deep open 35 28 0 1.25 63 Mixed Channel 14 18 0 0.78 32
70 Table 3. Habitat comparison results for type II design. Statistical results from analyses of habitat use compared to availability and to each other. Avail–Availability, eco–ecotone.
Deep open cluster Channel cluster Mixed Deep open cluster Mixed Channel cluster Habitat comparison Z p Habitat comparison Z p Habitat comparison Z p Habitat comparison Z p Channel–Avail* -7.45 0.0001 Channel–Avail* 9.22 0.0001 Channel–Avail* -8.47 0.0001 Channel–Avail* 5.89 0.0001 Deep eco–Avail* -5.72 0.0001 Deep eco–Avail -0.80 > 0.80 Deep eco–Avail* 4.30 0.0001 Deep eco–Avail* 3.50 0.0012 Sand flats–Avail* -11.75 0.0001 Sand flats–Avail* -9.54 0.0001 Sand flats–Avail* -8.71 0.0001 Sand flats–Avail* -6.45 0.0001 Seagrass beds–Avail* -10.44 0.0001 Seagrass beds–Avail* -8.95 0.0001 Seagrass beds–Avail* 5.98 0.0001 Seagrass beds–Avail 1.30 > 0.80 Shallow eco–Avail* -11.50 0.0001 Shallow eco–Avail* -9.37 0.0001 Shallow eco–Avail -0.14 > 0.80 Shallow eco–Avail* -4.54 0.0001 Deep open–Avail* 10.50 0.0001 Deep open–Avail* -8.23 0.0001 Deep open–Avail* 2.73 0.0486 Deep open–Avail* -7.10 0.0001 Channel–Deep eco -1.44 > 0.80 Channel–Deep eco* 6.52 0.0001 Channel–Deep eco* -7.12 0.0001 Channel–Deep eco -0.28 > 0.80 Channel–Sand flats* 6.17 0.0001 Channel–Sand flats* 6.99 0.0001 Channel–Sand flats 0.48 > 0.80 Channel–Sand flats* 5.47 0.0001 Channel–Seagrass beds* 3.89 0.0150 Channel–Seagrass beds* 6.95 0.0001 Channel–Seagrass beds* -7.22 0.0001 Channel–Seagrass beds* 3.23 0.0030 Channel–Shallow eco* 5.43 0.0001 Channel–Shallow eco* 7.00 0.0001 Channel–Shallow eco* -5.67 0.0001 Channel–Shallow eco* 5.17 0.0001 Channel–Deep open* -8.15 0.0001 Channel–Deep open* 6.88 0.0001 Channel–Deep open* -6.28 0.0001 Channel–Deep open* 5.45 0.0001
Deep eco–Sand flats* 6.19 0.0001 Deep eco–Sand flats* 5.88 0.0001 Deep eco–Sand flats* 6.98 0.0001 Deep eco–Sand flats* 4.97 0.0001
Deep eco–Seagrass beds* 5.46 0.0001 Deep eco–Seagrass beds* 5.64 0.0001 Deep eco–Seagrass beds* -3.53 0.0015 Deep eco–Seagrass beds* 2.82 0.0285 71 Deep eco–Shallow eco* 6.44 0.0001 Deep eco–Shallow eco* 5.80 0.0001 Deep eco–Shallow eco* 3.86 0.0015 Deep eco–Shallow eco* 4.56 0.0001 Deep eco–Deep open* -8.11 0.0001 Deep eco–Deep open* 5.42 0.0001 Deep eco–Deep open 1.23 > 0.80 Deep eco–Deep open* 4.95 0.0001 Sand flats–Seagrass beds* -4.41 0.0001 Sand flats–Seagrass beds* -4.28 0.0001 Sand flats–Seagrass beds* -7.09 0.0001 Sand flats–Seagrass beds* -4.72 0.0001 Sand flats–Shallow eco -1.09 > 0.80 Sand flats–Shallow eco -2.07 0.7845 Sand flats–Shallow eco* -6.39 0.0001 Sand flats–Shallow eco* -3.21 0.0090 Sand flats–Deep open* -8.52 0.0001 Sand flats–Deep open* -2.91 0.0045 Sand flats–Deep open* -6.85 0.0001 Sand flats–Deep open -0.48 > 0.80 Seagrass beds–Shallow eco* 4.09 0.0001 Seagrass beds–Shallow eco* 2.80 0.0180 Seagrass beds–Shallow eco* 5.04 0.0001 Seagrass beds–Shallow eco* 4.41 0.0001 Seagrass beds–Deep open* -8.41 0.0001 Seagrass beds–Deep open 1.46 > 0.80 Seagrass beds–Deep open* 3.32 0.0090 Seagrass beds–Deep open* 4.89 0.0001 Shallow eco–Deep open* -8.52 0.0001 Shallow eco–Deep open -1.12 > 0.80 Shallow eco–Deep open -2.58 > 0.80 Shallow eco–Deep open 2.86 0.0765
Table 4. Habitat comparison results for type I design. Statistical results from analyses of habitat use compared to availability and to each other.
Habitat comparison Z p Channel–Availability* -14.46 0.0001 Deep ecotone–Availability* 4.45 0.0001 Sand flats–Availability* -25.44 0.0001 Seagrass beds–Availability 0.06 > 0.80 Shallow ecotone–Availability* -20.87 0.0001 Deep open–Availability* 34.26 0.0001 Channel–Deep ecotone* -9.05 0.0001 Channel–Sand flats* 19.27 0.0001 Channel–Seagrass beds* -7.71 0.0001 Channel–Shallow ecotone* 13.11 0.0001 Channel–Deep open* -28.93 0.0001 Deep ecotone–Sand flats* 22.17 0.0001 Deep ecotone–Seagrass beds* 3.65 0.0060 Deep ecotone–Shallow ecotone* 17.31 0.0001 Deep ecotone–Deep open* -9.60 0.0001 Sand flats–Seagrass beds* -22.61 0.0001 Sand flats–Shallow ecotone* -7.41 0.0001 Sand flats–Deep open* -33.24 0.0001 Seagrass beds–Shallow ecotone* 17.45 0.0001 Seagrass beds–Deep open* -19.46 0.0001 Shallow ecotone–Deep open* -30.36 0.0001
72 CHAPTER II
The Ecological Conditions that Favor Tool Use and Innovation in Wild Bottlenose
Dolphins (Tursiops sp.)*
ABSTRACT
Dolphins are well known for their exquisite echolocation abilities, which enable them to detect and discriminate prey species and even locate buried prey. While these skills are widely used during foraging, some dolphins use tools to locate and extract prey. In the only known case of tool use in free-ranging cetaceans, a subset of bottlenose dolphins (Tursiops sp.) in Shark Bay,
Western Australia habitually employs marine basket sponge tools to locate and ferret prey from the seafloor. While it is clear that sponges protect dolphins’ rostra while searching for prey, it is still not known why dolphins probe the substrate at all instead of merely echolocating for buried prey as documented at other sites. By ‘sponge foraging’ ourselves, we show that these dolphins target prey that both lack swimbladders and burrow in a rubble-littered substrate. Delphinid echolocation and vision are critical for hunting but less effective on such prey. Consequently, if dolphins are to access this burrowing, swimbladderless prey, they must probe the seafloor and in turn benefit from using protective sponges. We suggest that these tools have allowed sponge foraging dolphins to exploit an empty niche inaccessible to their non-tool-using counterparts.
Our study identifies the underlying ecological basis of dolphin tool use and strengthens our understanding of the conditions that favor tool use and innovation in the wild.
*A version of this chapter appears as: Patterson, E. M. & Mann, J. 2011. PLoS ONE, 6, e22243 73 INTRODUCTION
Tool use (Bentley-Condit & Smith 2009; Seed & Byrne 2010) has long been of interest to biologists, anthropologists, and psychologists because of its role in cognition, culture, and hominid evolution (Lancaster 1968; Gibson & Ingold 1995; Matsuzawa 2001). Studying tool use in animals provides insight into the social, ecological, and evolutionary contexts in which it arises (van Schaik et al. 1999). In mammals and birds, tool use positively correlates with brain size, social transmission, and innovation (Lefebvre et al. 2004) and is considered to be a sign of cognitive capacity, i.e., problem solving (Inoue-Nakamura & Matsuzawa 1997; Tebbich &
Bshary 2004 but see Hansell & Ruxton 2008; Bird & Emery 2009). Most animal tools are used during foraging, especially extractive foraging (Gibson 1986; Bentley-Condit & Smith 2009). In
Shark Bay, Western Australia some bottlenose dolphins (Tursiops sp.) use marine basket sponge tools to protect their rostra during foraging (Smolker et al. 1997; Krützen et al. 2005; Mann et al.
2008; Bacher et al. 2010). Thus far, we know that sponge foraging, hereafter sponging, is primarily a female behavior, appears to be vertically socially transmitted (Mann et al. 2008), and is limited to 54 animals (hereafter the spongers) in the eastern gulf of Shark Bay. This solitary behavior occurs in deep channels, requires long dives, and consumes the majority of spongers’ activity budgets, yet does not appear to have any fitness costs (Mann et al. 2008).
On several occasions, during exceptional visibility, researchers have directly observed dolphins wearing marine sponges they have removed from the substrate (Figure 1A) over their rostra (Figure 1B) as they probe the rough seafloor (Figure 1C) while searching for hidden prey
(Figure 1D). Once prey have been extracted, dolphins drop their sponges, occasionally surface for a quick breath, chase and consume their prey, and finally, return to pick up their sponges and
74 continue foraging (Smolker et al. 1997; Mann et al. 2008). Spongers are suspected to target one or few benthic prey species, including the barred sandperch, Parapercis nebulosa, previously mis-identified as Parapercis clathrata (Mann et al. 2008), whose confamilials are consumed by bottlenose dolphins elsewhere (dos Santos et al. 2007). The sponge is thought to function as a shield by providing protection from the sharp and rough seafloor, and possibly venomous or spiny benthic marine organisms, while dolphins search for and extract prey (Smolker et al. 1997;
Mann et al. 2008). Dolphins use a single sponge for an average of 59 ± 43 (SD) minutes (Max =
3.5 hrs, Min = 1 minutes, N = 266 sponging bouts) before dropping it to search for a replacement presumably because the sponge has lost its protective value. However, why dolphins continuously probe the substrate when searching for prey is unclear given that at other locations
(e.g. crater feeding in the Bahamas, Rossbach & Herzing 1997) dolphins detect buried prey indirectly via echolocation and minimize contact with the seafloor until prey are located. In fact, delphinids’ target detection ability using echolocation is quite impressive (Au 2004; Au et al.
2009) and has long been used by the U.S. Navy to locate buried mines (Moore 1997). So in contrast to other extractive tool users (Gibson 1986), dolphins appear to have the anatomical machinery necessary for the task at hand leaving one to consider: why do Shark Bay dolphins probe or skim the substrate with their rostra and risk injury instead of simply echolocating for prey?
It is well known that the major acoustic backscatter of fishes (over 90%) comes from the gas-filled swimbladder (Haslett 1962; McCartney & Stubbs 1971; Foote 1980, 1986; Edwards &
Armstrong 1983a, 1983b; Foote & Ona 1985; Goddard & Welsby 1986; Ona 1990; Szczucka
2002). This is not surprising as sound waves most readily echo when encountering a difference in medium density (i.e. liquid to gas, Sen 1990). Fishes without swimbladders have relatively
75 weak acoustic signals, as fish flesh has an acoustic impedance only 10% greater than water
(Haslett 1962). Many fish species have lost their swimbladders, presumably as an adaptation to their benthic or deep-sea lifestyle (McCune & Carlson 2004). While some odontocetes are capable of echolocating swimbladderless prey (cephalopod detection by Globicephala,
Ziphiidae, and Physeteridae, Clarke 1996; Whitehead et al. 2003), the majority of these cetaceans’ prey are free swimming, not buried (Roper & Young 1975), and echolocation in these cephalopod specialists appears to be modified by longer click intervals and higher source levels when compared to bottlenose dolphins (Madsen et al. 2007). We hypothesized that spongers probe the substrate because they target prey that lack swimbladders and thus are difficult to detect with echolocation. Moreover, when these prey are at least partially buried beneath a debris-laden substrate, which causes interfering reverberation and echo clutter (echoes from objects other than the targeted prey, Turl et al. 1991), the effectiveness of echolocation is reduced even further. In contrast, dolphins that crater-feed in the Bahamas (Rossbach & Herzing
1997) appear to target prey with swimbladders (Denton et al. 1972) buried in an uncluttered, soft sand substrate that is less likely to injure dolphins’ rostra or dramatically interfere with echolocation (Roitblat et al. 1995). While some echolocation has been documented during sponging, it may only be useful once prey have been extracted and dolphins have dropped their sponges since the sponge itself is likely to interfere with echolocation by obstructing the sound receiving lower jaw and the sound emitting melon (Thomas et al. 2004). Thus, we predicted that the majority of prey that spongers encounter lack swimbladders.
76 RESULTS
Of the 134 prey extracted during 13.3 hours of human Sponging on both transect and verification dives (Figure 2), 78% lacked swimbladders (Table 1). In contrast only 19% of prey encountered during Non-Sponging lacked swimbladders (Table 1). Barred sandperch (Figure
1D), which lack swimbladders, were by far the most common prey extracted during Sponging, constituting 65% of the total count, but only made up 18% of Non-Sponging prey. Purple tuskfish (Choerodon cephalotes), which possess swimbladders, were second in abundance during both Sponging and Non-Sponging at 17% and 27% respectively; however, 74% of purple tuskfish extracted during Sponging were from locations where less sponging has been documented (verification dives, Figure 2), and this species was only extracted when divers probed small seagrass tufts which are both uncharacteristic of channel habitat and less likely to harm dolphins’ rostra. Striped whiptail (Pentapodus vitta), which possess swimbladders, were the predominant prey during Non-Sponging at 34%, but not extracted at all during Sponging. No additional families were extracted during Sponging on verification dives, although verification dives did yield 3 additional Non-Sponging families (Table 1), indicating that our transects are representative of spongers’ prey, but not of all non-burrowing prey in the eastern gulf of Shark
Bay, which is not surprising.
The ratio of prey without swimbladders to those with swimbladders was significantly higher during Sponging compared to Non-Sponging on video transects (Figure 3, Wilcoxon signed-rank test W = 28, p = 0.016), demonstrating that dolphins encounter a greater proportion of swimbladderless prey when sponging than is available to them without disturbing the substrate. Furthermore, the abundance of prey extracted during Sponging on transects was
77 significantly greater than that for the same prey families during Non-Sponging on transects
(Figure 4, Wilcoxon signed-rank test W = 28, p = 0.016), indicating that these prey are primarily concealed in the substrate and that sponging is an effective method of extraction. Together these results show that sponging dolphins extract concealed swimbladderless prey, and do so with greater efficiency than could be done without a sponge tool. Finally, a permutation test revealed that the number of prey families extracted during Sponging that lack swimbladders was significantly greater than expected when compared to 27 years of data from the Shark Bay
Dolphin Research Project’s long-term study (p = 0.0132, Table 2), suggesting that sponging is the primary way dolphins access swimbladderless prey.
DISCUSSION
Our results demonstrate that sponging dolphins regularly encounter swimbladderless prey that are concealed beneath a rubble-littered substrate. Fish are comprised primarily of water (fish flesh: 82% water, 17% protein, and 0.35% fat by weight, Norman & Greenwood 1975) so the density of fish (less the swimbladder) is only slightly greater than seawater (1.076 compared to
1.026 g/cm3, Jones & Marshall 1953). As such, swimbladderless prey have little acoustic impedance and are difficult to detect with echolocation (Haslett 1962; McCartney & Stubbs
1971; Foote 1980, 1986; Edwards & Armstrong 1983a, 1983b; Foote & Ona 1985; Goddard &
Welsby 1986; Ona 1990; Szczucka 2002). To make matters more difficult, Shark Bay channels are strewn with fragmented rock, shell, and coral that are not only likely to injure dolphins’ rostra, but also create an echo-cluttering environment leaving dolphins ‘acoustically blind’ to swimbladderless prey. Similarly, echolocating bats seem to have trouble detecting prey on the
78 surface of ponds that are covered with duckweed (Boonman et al. 1998). Our data demonstrate that spongers have developed a way of effectively extracting hidden prey by probing the substrate with protective sponge tools. Furthermore, when compared to the rest of the Shark Bay population, sponging dolphins appear to specialize in prey that lack swimbladders allowing them to occupy an empty ecological niche.
Alternatively, it is possible that dolphins could listen for soniferous benthic prey and simply use sponges for extraction. In fact, several species of echolocating bats passively listen for prey-generated sounds to detect insects in highly cluttered environments (Arlettaz et al.
2001). However, we believe this is unlikely for Shark Bay dolphins since only two prey families that were infrequently extracted during human Sponging are reported to be soniferous (Fish &
Mowbray 1970), and both possess swimbladders making them more detectable with echolocation anyway. The remaining swimbladderless prey are unlikely to be soniferous since the primary sonic mechanism in fishes is swimbladder movement (Fish & Mowbray 1970). Furthermore, fishes mainly produce sound for intraspecific communication (Fish & Mowbray 1970), not while hidden or buried in the substrate as observed in this study.
The predominant prey extracted during human Sponging was the barred sandperch whose behavior was strikingly consistent with dolphin sponging behavior (Mann et al. 2008). When barred sandperch were disturbed during human Sponging, they swam a few meters away and returned to the substrate often without reburying. This would give dolphins time to drop their sponges and quickly surface to breathe before diving back down in pursuit, as has been regularly documented during our long-term study (Mann et al. 2008). New photographs from 2010 of a sponger consuming a small red and brown fish provided further evidence that spongers prey on barred sandperch.
79 Sponging dolphins may gain several benefits from targeting these prey. First, barred sandperch exhibit consistent, predictable behavior enabling dolphins to employ a single stereotypic sponging behavior. If dolphins extracted a variety of prey species, all having different anti-predator tactics, a uniform sponging behavior would not be as effective. Second, similar to some foods extracted by primate tool users (Gibson 1986), barred sandperch are reliable and can easily and frequently be extracted with a sponge, one every nine minutes during human
Sponging. However, the average barred sandperch collected was small, only 12.6 ± 4.7 (SD) cm in length (Max = 23 cm, Min = 6 cm, N = 21), which may explain why spongers are more specialized and dedicate more time to foraging than other dolphins in Shark Bay (Mann et al.
2008). Finally, extracted foods are often high energy, premium foods (Gibson 1986). Since fishes with decreased swimbladder volumes typically have increased lipid content (Ona 1990), the barred sandperch may provide sponging dolphins with an energy-rich meal, similar to some insect larvae extracted by tool-using birds (Rutz et al. 2010). Accordingly, several barred sandperch have been stored for nutritional analysis. Thus, while requiring more effort than free- swimming prey, barred sandperch likely provide these tool users with a small, yet predictable, reliable, and possibly energy rich food source.
This highly specific tool use has implications for cognition and brain evolution among cetaceans and could even be considered a case of problem solving, a phenomenon difficult to document in the wild, but well established in studies of captive bottlenose dolphins (Herman
2010). Our study demonstrates how bottlenose dolphins might use these skills in their natural environment and provides insight into the ecological and evolutionary pressures that promote higher-level cognition. Spongers may have solved the problem of detecting and extracting swimbladderless prey from below a sharp and rough substrate by probing the seafloor with a soft
80 sponge tool. This solution appears to have been adopted at least twice, as unrelated dolphins residing 110 km away in the western gulf of Shark Bay also sponge forage (Bacher et al. 2010).
While this tool use requires sophisticated object manipulation, it appears to provide spongers with equal fitness compared to the rest the population (Mann et al. 2008).
Due to the difficulty of observing marine fauna, most studies of tool use focus on terrestrial organisms. Using novel underwater techniques, we show that sponge tool-using dolphins target buried prey that lack swimbladders, particularly barred sandperch. Such prey are difficult to detect with echolocation (Haslett 1962; McCartney & Stubbs 1971; Foote 1980,
1986; Edwards & Armstrong 1983a, 1983b; Foote & Ona 1985; Goddard & Welsby 1986; Ona
1990; Szczucka 2002), which, when paired with Shark Bay’s cluttered channel substrate, explains why dolphins probe the seafloor and benefit from using sponge tools. Similar to ant- fishing chimpanzees whose tool use is a function of prey type (Schöning et al. 2008), dolphin tool use directly relates to the physical characteristics of their prey. As such, this study emphasizes the critical role ecological factors play in explaining behavioral complexity.
MATERIALS AND METHODS
Study site
The Shark Bay Dolphin Research Project has an extensive 29-year database that includes demographic, genetic, association, life-history, ecological, and behavioral data on >1,400 dolphins (past and present) residential to a 500 km2 area (25° 47’ S, 113° 43’ E). Habitat consists of embayment plains (5 -13 m), shallow sand flats (0.5 – 4 m), seagrass beds (0.5 – 4 m), and
81 bisecting deep channels (7 -13 m). One area in particular, just north of Monkey Mia (Figure 2), is extremely well sampled because it is very close to our boat-launching site. As such, spatial patterns of dolphin foraging behavior in this location are unlikely the result of a bias in sampling effort. Using historical data, dolphin-sighting locations were projected using ArcGIS Map 9.3
(WGS 1984 UTM Zone 49S, ESRI 2011) to determine locations where dolphins sponge forage.
Seven semi-permanent transects were established using 1 m long star picket metal posts with attached buoys in locations with the highest numbers of recorded sponging observations. Each
100 m transect was ~ 200 m from adjacent transects and further split into two portions (NW and
SE) by a mid-stake. The NW portion of each transect was dedicated to systematic observational- video sampling, while the SE portion was designated for prey sample collection.
Data collection
For all 7 transects, two certified divers swam out a 50 m tape measure well above the substrate to connect the NW stake to the mid-stake for initial transect setup. After waiting several minutes, both divers then swam back towards the NW stake along one side of the transect line near the substrate and filmed a ~ 2 m wide belt transect (Hill & Wilkinson 2004; Eleftheriou &
McIntyre 2005) to determine prey availability near the seafloor without disturbing the substrate
(Non-Sponging). Next, divers swam back along the other side of the transect (~2 m to the side of the tape measure) towards the mid-stake with one diver pushing a 2 m long pole with a dead marine sponge attached along the substrate to ferret prey in the same manner as seen by sponging dolphins (Sponging), and the other diver filming this human Sponging with a Sony HDR-
XR500V HD video camera in an AquaticaHD housing (see Patterson & Mann 2011 video S1).
82 All dives were performed using an Airline Supply R360XL Hookah System by J. Sink, and were swam at a consistent speed of ~ 0.28 m/s. On the NW portion of each transect, all prey were simply filmed and allowed to swim away. However, on the SE side of the transects several sample specimens of all species encountered (except Dasyatidae and Sepiidae which all lack swimbladders) were collected using hand nets for identification and dissection. Transect sampling was performed on two different occasions for repeatability, but replicates were averaged to form a single transect value. To confirm that our fine scale study in this well sampled area was representative of greater patterns in the bay, in particular to be sure we had extracted all possible Sponging prey species, we also performed verification dives in all other general locations where sponging has been observed (Figure 2). On these verification dives no tape measure was laid, but divers performed and filmed both Non-Sponging and Sponging as described above. If any new species were encountered, sample specimens were collected for identification and dissection. Only one infrequent prey species was too fast for divers to catch,
Upeneus tragula, but other species in the same genus are known to have swimbladders
(Tominaga et al. 1996). Historical prey species were gathered from the long-term Shark Bay
Dolphin Research Project’s database and combined with families extracted during Sponging to create a total of 29 possible prey families. Families were used instead of individuals due to the potential biases in observing accurate quantities of species consumed (e.g. researchers observe dolphins consuming surface dwelling prey more often than benthic prey since dolphins regularly consume these prey near the surface, in plain sight). Using families allows us to avoid these potential biases and explore how dolphins hunt the richness of prey they encounter. Swimbladder status for prey families not collected and prey from video only identified to the level of family was determined using primary literature (Wittenberg et al. 1964; Fänge 1983; Green 1984;
83 Collette & Su 1986; McKay 1992; Battaglene et al. 1994; Tominaga et al. 1996; Treasurer 1997;
McCosker & Robertson 2001; Cole et al. 2003; McCune & Carlson 2004; Moravec & Justine
2007). All data in Table 1 and Table 2 follow the currently accepted scientific and common names according to Froese & Pauly (2010) and all analyses were performed at the family level.
Data processing and statistical analysis
Many of the species encountered were quite small, averaging less than 7 cm in length.
Such small prey are unlikely to be targeted during sponging because prey this size can easily be obtained at the surface in all habitats, even by young calves (Mann & Sargeant 2003; Sargeant et al. 2007). Thus, these prey were removed from prey abundance data. A total of 19 prey encountered could not be identified to the level of family; however, all were estimated to be less than 7 cm in length and thus excluded from the final data set. There is a chance prey were missed during video logging or were simply not captured on film; however, it is likely that all such prey would also be less than 7 cm in length and thus excluded since prey larger than this would be obvious to divers and in video and not overlooked.
The ratio of prey without swimbladders to those with swimbladders was compared between Sponging and Non-Sponging on transects using a Wilcoxon signed-rank test. Here, prey abundance data were transformed by adding one to all samples before ratios were calculated to correct for samples with zero individuals in either group, which results in an undefined ratio. We also compared the abundance of prey families extracted during Sponging on transects to the abundance of these same families during Non-Sponging on transects using a Wilcoxon signed- rank test. Finally, we compared data from Sponging dives to data from our long-term study. For
84 this final analysis we used a two-tailed permutation test to compare the observed number of families with and without swimbladders from Sponging to the expected number based on combined Sponging and historical prey data. We re-sampled (with replacement) 8 prey families
10,000 times from 29 possible families (Table 2) and determined the likelihood of obtaining our observed human Sponging data by chance. While Table 1 present abundance data pooled from all dives for descriptive purposes, all statistical analyses were performed only using the systematic transect data for which we could be sure that the amount of substrate traversed during
Sponging and Non-Sponging were equal. All statistical tests were performed in R 2.12.1 statistical environment (R Development Core Team 2011) and considered significant for p <
0.05.
ACKNOWLEDEMENTS
We thank our field assistants James Barnao, Anne Cotter, and Jacob Swartz and all members of the Mann lab, especially Ewa Krzyszczyk for providing new photos of a sponger with its prey. We thank all members, past and present, of the Shark Bay Dolphin Research
Project who contributed to the long-term data. Western Australian Museum’s Dr. Barry Hutchins and Ms. Sue Morrison helped with prey identification and provided advice on field sampling methods, while Dr. Jane Fromont identified sponge species. We also thank the Shark Bay
Ecosystem Research Project for help in the field and the Monkey Mia Dolphin Resort and the
Department of Environment and Conservation of Western Australia (DEC) for field and logistical support. All animal work was approved by the Georgetown University Animal Care and Use Committee (GUACUC) under permits 07-041 and 10-023. Observational and field
85 studies were approved by the Department of Environment and Conservation of Western
Australia (DEC) under permits SF007418 and SF006897. Support for this study was provided by
Georgetown University, the National Geographic Society Young Explorers Grant, the Explorers
Club Exploration Fund Grant, the Achievement Rewards for College Scientist, the Animal
Behavior Society Cetacean Behavior and Conservation Award, and the American Society of
Mammologists Grant in Aid of Research to EMP. Additional funding was provided by the
National Science Foundation Grants (NSF) 0847922, 0316800, 0918303 and Georgetown
University to JM. The orthophoto in Figure 1 was supported by NSF grant OCE0526065 to Dr.
Mike Heithaus and DEC. Field gear was supplied at discounted rates by Airline Supply by J.
Sink and Splash Dive Center of Alexandria, VA.
86 REFERENCES
Arlettaz, R., Jones, G. & Racey, P. 2001. Effect of acoustic clutter on prey detection by bats.
Nature, 414, 742–745.
Au, W. W. L. 2004. The dolphin sonar: excellent capabilities in spite of some mediocre
properties. In: High Frequency Ocean Acoustics Conference: AIP Conference Proceedings,
Vol 728, (Ed. by M. B. Porter, M. Siderius, & W. A. Kuperman), pp. 247–259. Melville:
American Institute of Physics.
Au, W. W. L., Branstetter, B. K., Benoit-Bird, K. J. & Kastelein, R. A. 2009. Acoustic basis
for fish prey discrimination by echolocating dolphins and porpoises. Journal of the
Acoustical Society of America, 126, 460–467.
Bacher, K., Allen, S. J., Lindholm, A. K., Bejder, L. & Krützen, M. 2010. Genes or culture:
are mitochondrial genes associated with tool use in bottlenose dolphins (Tursiops sp.)?
Behavior Genetics, 40, 706–714.
Battaglene, S. C., McBride, S. & Talbot, R. B. 1994. Swim bladder inflation in larvae of
cultured sand whiting, Sillago ciliata Cuvier (Sillaginidae). Aquaculture, 128, 177–192.
Bentley-Condit, V. K. & Smith, E. O. 2009. Animal tool use: current definitions and an
updated comprehensive catalog. Behaviour, 147, 185–221.
Bird, C. D. & Emery, N. J. 2009. Rooks use stones to raise the water level to reach a floating
worm. Current Biology, 19, 1410–1414.
Boonman, A. M., Boonman, M., Bretschneider, F. & van de Grind, W. A. 1998. Prey
detection in trawling insectivorous bats: duckweed affects hunting behaviour in
Daubenton’s bat, Myotis daubentonii. Behavioral Ecology and Sociobiology, 44, 99–107.
87 Clarke, M. R. 1996. Cephalopods as prey. III. Cetaceans. Philosophical Transactions of the
Royal Society of London. Series B: Biological Sciences, 351, 1053–1065.
Cole, R. G., Alcock, N. K., Handley, S. J., Grange, K. R., Black, S., Cairney, D., Day, J.,
Ford, S. & Jerrett, A. R. 2003. Selective capture of blue cod Parapercis colias by potting:
behavioural observations and effects of capture method on peri-mortem fatigue. Fisheries
Research, 60, 381–392.
Collette, B. B. & Su, J. 1986. The halfbeaks (Pisces, Beloniformes, Hemiramphidae) of the far
east. Proceedings of the Academy of Natural Sciences of Philadelphia, 138, 250–301.
Denton, E. J., Liddicoa, J. D. & Taylor, D. W. 1972. The permeability to gases of the
swimbladder of the conger eel (Conger conger). Journal of the Marine Biological
Association of the United Kingdom, 52, 727–746. dos Santos, M. E., Coniglione, C. & Louro, S. 2007. Feeding behaviour of the bottlenose
dolphin, Tursiops truncatus (Montagu, 1821) in the Sado estuary, Portugal, and a review of
its prey species. Revista Brasileira de Zoociências, 9, 31–39.
Edwards, J. I. & Armstrong, F. 1983a. Measurement of the target strength of live herring and
mackerel. FAO fisheries report, 300, 69–77.
Edwards, J. I. & Armstrong, F. 1983b. Target strength measurements on herring, sprat and
mackerel. International Council for the Exploration of the Sea C.M., B:23, 1–9.
Eleftheriou, A. & McIntyre, A. 2005. Methods for the Study of Marine Benthos. 3rd edn.
Oxford: Blackwell.
ESRI. 2011. ArcGIS Desktop. Version 9.3. Redlands, CA: Environmental Systems Research
Institute.
88 Fish, M. P. & Mowbray, W. H. 1970. Sounds of Western North Atlantic Fishes: a Reference
File of Biological Underwater Sounds. Baltimore: Johns Hopkins Press.
Foote, K. G. 1980. Importance of the swimbladder in acoustic scattering by fish: a comparison
of gadoid and mackerel target strengths. Journal of the Acoustical Society of America, 67,
2084–2089.
Foote, K. G. 1986. A critique of Goddard and Welsby’s paper: the acoustic target strength of
live fish. Journal du Conseil, 42, 212–220.
Foote, K. G. & Ona, E. 1985. Swimbladder cross sections and acoustic target strengths of 13
pollack and 2 saithe. In: Fiskeridirektoratets skrifter: Serie Havundersøkelser, Vol 18 pp.
1– 57. Bergen: Directorate of Fisheries.
Froese, R. & Pauly, D. 2010. FishBase World Wide Web electronic publication.
www.fishbase.org.
Fänge, R. 1983. Gas exchange in fish swim bladder. Reviews of Physiology Biochemistry and
Pharmacology, 97, 111–158.
Gibson, K. R. 1986. Cognition, brain size and the extraction of embedded food resources. In:
Primate Ontogeny, Cognition, and Social Behaviour, (Ed. by J. G. Else & P. C. Lee), pp.
93–103. Cambridge: Cambridge University Press.
Gibson, K. R. & Ingold, T. 1995. Tools, Language and Cognition in Human Evolution.
Cambridge: Cambridge University Press.
Goddard, G. C. & Welsby, V. G. 1986. The acoustic target strength of live fish. Journal du
Conseil, 42, 197–211.
Green, S. L. 1984. Ultrastructure and innervation of the swimbladder of Tetractenos glaber
(Tetraodontidae). Cell and Tissue Research, 237, 277–284.
89 Hansell, M. & Ruxton, G. D. 2008. Setting tool use within the context of animal construction
behaviour. Trends in Ecology & Evolution, 23, 73–78.
Haslett, R. W. G. 1962. The back-scattering of acoustic waves in water by an obstacle II:
eetermination of the reflectivities of solids using small specimens. Proceedings of the
Physiological Society, 79, 559–571.
Herman, L. M. 2010. What laboratory research has told us about dolphin cognition.
International Journal of Comparative Psychology, 23, 310–330.
Hill, J. & Wilkinson, C. 2004. Methods for Ecological Monitoring of Coral Reefs. Townsville:
Australian Institude of Marine Science.
Inoue-Nakamura, N. & Matsuzawa, T. 1997. Development of stone tool use by wild
chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 111, 159–173.
Jones, F. R. H. & Marshall, N. B. 1953. The structure and functions of the theleostean
swimbladder. Biological Reviews of the Cambridge Philosophical Society, 28, 16–82.
Krützen, M., Mann, J., Heithaus, M. R., Connor, R. C., Bejder, L. & Sherwin, W. B. 2005.
Cultural transmission of tool use in bottlenose dolphins. Proceedings of the National
Academy of Sciences of the United States of America, 102, 8939–8943.
Lancaster, J. B. 1968. On the evolution of tool using behavior. American Anthropologist, 70,
56–66.
Lefebvre, L., Reader, S. & Sol, D. 2004. Brains, innovations and evolution in birds and
primates. Brain, Behavior and Evolution, 63, 233–246.
Madsen, P. T., Willson, M., Johnson, M., Hanlon, R. T., Bocconcelli, A., Aguilar de Soto, N.
& Tyack, P. L. 2007. Clicking for calamari: toothed whales can echolocate squid Loligo
pealeii. Aquatic Biology, 1, 141–150.
90 Mann, J. & Sargeant, B. L. 2003. Like mother, like calf: the ontogeny of foraging traditions in
wild Indian ocean bottlenose dolphins (Tursiops sp.). In: The Biology of Traditions: Models
and Evidence, (Ed. by D. Fragaszy & S. Perry), pp. 236–266. Cambridge: Cambridge
University Press.
Mann, J., Sargeant, B. L., Watson-Capps, J. J., Gibson, Q. A., Heithaus, M. R., Connor, R.
C. & Patterson, E. M. 2008. Why do dolphins carry sponges? PloS ONE, 3, e3868.
Matsuzawa, T. 2001. Primate foundations of human intelligence: a view of tool use in
nonhuman primates and fossil hominids. In: Primate Origins of Human Cognition and
Behavior, (Ed. by T. Matsuzawa), pp. 3–25. Tokyo: Springer.
McCartney, B. S. & Stubbs, A. R. 1971. Measurements of the acoustic target strengths of fish
in dorsal aspect, including swimbladder resonance. Journal of Sound and Vibration, 15,
397–404.
McCosker, J. E. & Robertson, D. R. 2001. Aplatophis zorro, a new species of eastern Pacific
snake-eel, with comments on New World ophichthid distributions (Anguilliformes:
Ophichthidae). Revista de Biologia Tropical, 49, 13–20.
McCune, A. R. & Carlson, R. L. 2004. Twenty ways to lose your bladder: common natural
mutants in zebrafish and widespread convergence of swim bladder loss among teleost
fishes. Evolution & Development, 6, 246–259.
McKay, R. J. 1992. Sillaginid fishes of the world (family Sillaginidae). Rome: FAO.
Moore, P. W. 1997. Mine-hunting dolphins of the Navy. Proceedings of SPIE, 3079, 2–6.
Moravec, F. & Justine, J.-L. L. 2007. Redescription of Metabronema magnum (Nematoda:
Cystidicolidae), a swimbladder parasite of the carangid fish Gnathanodon speciosus off
New Caledonia. Folia Parasitologica, 54, 293–300.
91 Norman, J. R. & Greenwood, P. H. 1975. A History of Fishes. 3rd edn. New York: Wiley.
Ona, E. 1990. Physiological factors causing natural variations in acoustic target strength of fish.
Journal of the Marine Biological Association of the United Kingdom, 70, 107–127.
Patterson, E. M. & Mann, J. 2011. The ecological conditions that favor tool use and innovation
in wild bottlenose dolphins (Tursiops sp.). PloS ONE, 6, e22243.
R Development Core Team. 2011. R: A Language and Environment for Statistical Computing,
Version 2.12.1. Vienna, Austria: R Foundation for Statistical Computing.
Roitblat, H. L., Au, W. W. L., Nachtigall, P. E., Smzumura, R., Moons, G. & Shizumura, R.
1995. Sonar recognition of targets embedded in sediment. Neural Networks, 8, 1263–1273.
Roper, C. F. E. & Young, R. E. 1975. Vertical Distribution of Pelagic Cephalopods. In:
Smithsonian Contributions to Zoology, Vol 209. pp. 51. Washington: Smithsonian
Institution Press.
Rossbach, K. a. & Herzing, D. L. 1997. Underwater observations of benthic-feeding bottlenose
dolphins (Tursiops truncatus) near Grand Bahama Island, Bahamas. Marine Mammal
Science, 13, 498–504.
Rutz, C., Bluff, L. A., Reed, N., Troscianko, J., Newton, J., Inger, R., Kacelnik, A. &
Bearhop, S. 2010. The ecological significance of tool use in New Caledonian crows.
Science, 329, 1523–1526.
Sargeant, B. L., Wirsing, A. J., Heithaus, M. R. & Mann, J. 2007. Can environmental
heterogeneity explain individual foraging variation in wild bottlenose dolphins (Tursiops
sp.)? Behavioral Ecology and Sociobiology, 61, 679–688.
92 Schöning, C., Humle, T., Möbius, Y. & McGrew, W. C. 2008. The nature of culture:
technological variation in chimpanzee predation on army ants revisited. Journal of Human
Evolution, 55, 48–59.
Seed, A. & Byrne, R. 2010. Animal tool-use. Current Biology, 20, R1032–R1039.
Sen, S. N. 1990. Acoustics, Waves and Oscillations. Daryaganji: New Age International
Publishers.
Smolker, R. A., Richards, A. F., Connor, R. C., Mann, J. & Berggren, P. 1997. Sponge
carrying by dolphins (Delphinidae, Tursiops sp.): a foraging specialization involving tool
use? Ethology, 103, 454–465.
Szczucka, J. 2002. Acoustic properties of fish as related to external conditions. Acta Acustica
United with Acustica, 88, 735–738.
Tebbich, S. & Bshary, R. 2004. Cognitive abilities related to tool use in the woodpecker finch,
Cactospiza pallida. Animal Behaviour, 67, 689–697.
Thomas, J. A., Moss, C. F. & Vater, M. 2004. Echolocation in Bats and Dolphins. Chicago:
University of Chicago Press.
Tominaga, Y., Sakamoto, K. & Matsuura, K. 1996. Posterior Extension of the Swimbladder in
Percoid Fishes, with a Literature Survey of Other Teleosts. Tokyo: University Museum,
University of Tokyo.
Treasurer, J. W. 1997. Parasites of wrasse (Labridae) in inshore waters of west Scotland
stocked as cleaner fish of sea lice (Caligidae) on farmed Atlantic salmon. Journal of Fish
Biology, 50, 895–899.
93 Turl, C. W., Skaar, D. J. & Au, W. W. 1991. The echolocation ability of the beluga
(Delphinapterus leucas) to detect targets in clutter. The Journal of the Acoustical Society of
America, 89, 896–901. van Schaik, C. P., Deaner, R. O. & Merrill, M. Y. 1999. The conditions for tool use in
primates: implications for the evolution of material culture. Journal of Human Evolution,
36, 719–741.
Whitehead, H., MacLeod, C. D. & Rodhouse, P. 2003. Differences in niche breadth among
some teuthivorous mesopelagic marine mammals. Marine Mammal Science, 19, 400–406.
Wittenberg, J. B., Schwend, M. J. & Wittenberg, B. A. 1964. The secretion of oxygen into the
swim-bladder of fish. Journal of General Physiology, 48, 337–355.
94 FIGURES AND TABLES
Figure 1. Sponging in Shark Bay. (A) marine basket sponge (Echinodyctium mesenterinum), (B) dolphin wearing a sponge on its rostrum, (C) substrate littered with rock, shell, and debris, (D) hiding prey, barred sandperch (Parapercis nebulosa). All photographs taken by Eric M.
Patterson.
95
Figure 2. Sponging Map. Boat launch site (Monkey Mia), dolphin sponge foraging sightings, transects, and verification dive sites in Shark Bay, Western Australia.
96
Figure 3. Ratio of prey without swimbladders (SB) to prey with swimbladders during both
Sponging and Non-Sponging on transects. Data were transformed (+1) before ratios were calculated to correct for undefined ratios in samples with zero individuals in either group.
Wilcoxon signed-rank test W = 28, *p = 0.016
97
Figure 4. Abundance of prey species extracted during Sponging and abundance of these same families during Non-Sponging on transects. Wilcoxon signed-rank test W = 28, *p = 0.016
98 Table 1. Prey abundance from Sponging and Non-Sponging, pooled from both transects and verification dives. Numbers represent reference(s) used to determine swimbladder status: 1Cole et al., 2003; 2Tominaga, Sakamoto, & Matsuura, 1996; 3McCune & Carlson, 2004; 4Treasurer,
1997; 5Green, 1984. *Dissected in this study, +swimbladder well known to be absent in entire family, #additional family encountered on verification dives.
Sponging Non-Sponging Common Name Family Species (if known) Swimbladder Abundance Abundance barred sandperch Pinguipedidae Parapercis nebulosa 87 15 N1* sand lizardfish Synodontidae Synodus dermatogenys 9 0 N2* cuttlefishes Sepiidae 5 0 N+ stingrays Dasyatidae 1 1 N+ lefteye flounders Bothidae 1 0 N3* painted maskray Dasyatidae Neotrygon leylandi 1 0 N+ tasselsnout flathead Platycephalidae Thysanophrys cirronasa 1 0 N2* purple tuskfish Labridae Choerodon cephalotes 23 23 Y2,4* freckled goatfish Mullidae Upeneus tragula 4 0 Y2 wrasses Labridae 2 0 Y2,4* striped whiptail Nemipteridae Pentapodus vitta 0 29 Y* margined coralfish Chaetodontidae Chelmon marginalis 0 6 Y2# blackspot tuskfish Labridae Choerodon schoenleinii 0 5 Y2,4* bluntheaded wrasse Labridae Thalassoma amblycephalum 0 2 Y2,4* humpback batfish Ephippidae Platax batavianus 0 2 Y2# yellowtail clownfish Pomacentridae Amphiprion clarkii 0 1 Y2# puffers Tetraodontidae 0 1 Y5*
99 Table 2: Sponging and historical prey families. Numbers represent reference(s) used to determine swimbladder status: 1McCune & Carlson, 2004; 2Cole et al., 2003; 3Tominaga et al.,
1996; 4Treasurer, 1997; 5Moravec & Justine, 2007; 6Fänge, 1983; 7Collette & Su, 1986;
8McCosker & Robertson, 2001; 9Wittenberg et al. 1964; 10Battaglene et al. 1994; 11McKay,
1992; 12Green, 1984. *Dissected in this study, +swimbladder well known to be absent in entire family.
Extracted Common Name Family Swimbladder During Sponging lefteye flounders Bothidae Y N1* stingrays Dasyatidae Y N+ sandperches Pinguipedidae Y N2* flatheads Platycephalidae Y N3* cuttlefishes Sepiidae Y N+ lizardfishes Synodontidae Y N3* octopuses unidentified Octopoda families N N+ squids unidentified Teuthida families N N+ wrasses Labridae Y Y3,4* goatfishes Mullidae Y Y3 bonefishes Albulidae N Y3 needlefishes Belonidae N Y3 jacks and pompanos Carangidae N Y3,5 herrings, shads, sardines, menhadens Clupeidae N Y3,6 halfbeaks Hemiramphidae N Y3,7 filefishes Monacanthidae N Y3,6 mullets Mugilidae N Y3* snake eels Ophichthidae N Y3,8 eeltail catfishes Plotosidae N Y3 bluefishes Pomatomidae N Y3,6,9 dottybacks Pseudochromidae N Y3 scats Scatophagidae N Y3 tuna, mackerels, and bonitos Scombridae N Y3 sea basses Serranidae N Y3 rabbitfishes Siganidae N Y* smelt-whitings Sillaginidae N Y3,10,11 porgies Sparidae N Y3 grunters or tigerperches Terapontidae N Y3 puffers Tetraodontidae N Y12*
100 CHAPTER III
Becoming an Expert: Lifelong Learning in Tool-using Wild Bottlenose Dolphins
(Tursiops sp.)*
SUMMARY
The ability to learn may be universal among animals (Dukas 2009), but the extent to which non-human animals demonstrate sustained, lifelong learning of a skill akin to the acquisition of expertise in humans, remains unknown. Here we examined whether wild female bottlenose dolphins improved upon their sponge tool-use proficiency with age. Similar to chimpanzees (Pan troglodytes) and human forager societies (Homo sapiens) (Helton 2008), we find that tool use efficiency follows an inverted U function of age; dolphins peak in tool use performance at around 23 of age, but can live into their early 40s. Additionally, by their early 20s dolphins spend little time locating sponge tools, presumably because they have substantial knowledge concerning the distribution of sponges within their home ranges. Sponging performance peaks when females are most likely to have dependent offspring and increased foraging efficiency would help them meet the demands of lactation. The inverted U-shaped pattern with age seen across taxa and in multiple domains indicates that similar underlying developmental and senescent mechanisms are involved. Our results show that the development of expertise plays an important role in non-human animal life histories and has implications for behavioral specialization, social learning, culture, and fitness.
*Authorship for paper: Eric M. Patterson, Ewa Krzyszczyk & Janet Mann 101 RESULTS AND DISCUSSION
From C. elegans (Wen et al. 1997) to humans (Ormrod 2004), learning enables animals to adjust their behavior to meet the demands of their unique life situations (Dukas 1998, 2009).
Extended learning periods are often associated with cognitively complex tasks (McGrew 1992;
Inoue-Nakamura & Matsuzawa 1997; Dukas 1998; Holzhaider et al. 2010), and in humans coincide with the development of expertise (Ericsson et al. 2006). In fact, there may not be any qualitative difference between humans and non-human animals in regards to expertise development (Helton 2005). However, unlike human expertise, which extends into many domains, animal expertise, if it exists, likely occurs in behaviors directly tied to fitness like foraging. Thus, by examining lifetime learning in cognitively demanding foraging behaviors, like those involving tool use, we can gain insight into the evolution and development of expertise in animals.
Bottlenose dolphins are well known for their cognitive and learning abilities both in captivity and in the wild (Herman 2002; Marino et al. 2007). In Shark Bay, Australia, a subset of wild bottlenose dolphins (Tursiops sp., primarily females, hereafter spongers) use marine basket sponge tools to forage (Fig. 1, Smolker et al. 1997; Mann et al. 2008), a behavior that is vertically transmitted from mother to offspring (Sargeant & Mann 2009). Previous data suggest that sponge foraging (hereafter sponging) is difficult to learn, with late onset compared to other foraging behaviors (Mann & Sargeant 2003; Sargeant & Mann 2009). However, beyond this we know very little about how, or if, spongers improve upon their tool-using skills with experience.
Here we tested whether wild female bottlenose dolphins develop tool use expertise in a way similar to chimpanzees and human forager societies (Kaplan et al. 2000; Walker et al. 2002;
102 Lonsdorf 2005; Gurven et al. 2006; Helton 2008) while controlling for ecological factors and foraging effort.
Sponging experience
Improved performance in any skill is likely the result of substantial experience, or in the human literature, practice. For expertise, many authors note the importance of an ‘early start’ followed by sustained lifetime practice (Ericsson et al. 2006). Previous data demonstrated that dolphins who become spongers learn the behavior early on and specialize in this foraging tactic for life (Mann & Sargeant 2003; Mann et al. 2008; Sargeant & Mann 2009). Our results confirm these findings. After dolphins initially begin sponging around age 2, time spent sponging continues to increase until 8–10 years of age when it consumes over 50 % of their diurnal activity budget (Fig. 2). Females reach sexual maturity at 10–12 years (Mann et al. 2000;
Krzyszczyk & Mann 2012), which may explain the increased foraging effort and presumably energy intake. Beyond 8–10 years of age, time spent sponging appears fairly stable, but may decrease late in life (Fig. 2).
Tool use efficiency
There are three main stages of sponging. First, a dolphin must locate a sponge tool. We regularly observe spongers traveling from place to place without a sponge, intermittently performing long deep dives in what appeared to be searching behavior. Second, after a sponge is located, a dolphin must detach the tool, which is anchored to the substrate by a strong basal plate
103 (Hooper & Van Soest 2002). During this stage, dolphins repeatedly dive in the same location for an extended period until they surface with a new sponge tool. Finally, the third and last stage, foraging, can begin. With their sponge tools ‘on beak’ for protection (Fig. 1), dolphins probe the seafloor riddled with sharp debris in search of hidden and otherwise difficult to detect prey
(Patterson & Mann 2011).
Previous work demonstrated that the majority of prey spongers encounter (barred sandperch, Parapercis nebulosa) are small, abundant, and easy to catch, making them a reliable food source (Patterson & Mann 2011). Here we further confirm barred sandperch in the diet of spongers. Using custom designed primers for the cytochrome oxidase subunit I gene (cox1) of P. nebulosa, we amplified DNA from a known sponger’s fecal sample and obtained an exact match
(Ward et al. 2005, Table 1–2). As has previously been noted (Mann et al. 2008; Patterson &
Mann 2011), spongers swallow their prey quickly, usually before resurfacing, indicating very little handling time and energy are required after prey capture. This is in striking contrast to some other foraging behaviors in Shark Bay. For example, during beach hunting dolphins spend considerable energy reaching high speeds in order hydroplane onshore in an attempt to catch a single large fish (Sargeant et al. 2005), which may require processing to consume. In contrast then to some other foraging behaviors, the most difficult and costly aspects of sponging are likely locating and detaching an appropriate tool. Thus, if dolphins are to improve their sponging skills, they are most likely to do so during these first two stages of sponging.
One way of looking at sponging performance is to examine efficiency (energy gain per unit energy expenditure, Stephens & Krebs 1986; Ydenberg 1998). To maximize sponging efficiency dolphins should decrease time spent locating and detaching sponges and increase time actually foraging. The simplest way to achieve this is to use any single sponge tool for an
104 extended period. Spongers invest time and energy locating and detaching their tools, so abandoning a sponge that is still functional is costly. However, since sponges cover dolphins’ beaks (Fig. 1), dolphins must drop their tools to consume prey, but can, and are regularly observed, retrieving their tools to continue foraging (Mann et al. 2008; Patterson & Mann 2011).
This tool re-use means dolphins can spend more time foraging and less time obtaining tools, which increases their tool bout duration (amount of time foraging with a single sponge tool) and ultimately their sponging efficiency. Thus, to examine lifetime changes in sponging efficiency we explored the relationship between tool bout duration and age, percent time sponging, and basket sponge density. We found that tool use efficiency follows a negative quadratic relationship with age (MCMC glmm, Table 3 & 7, Fig. 3). Dolphins increase tool bout duration until a peak at midlife (predicted peak of 85.3 minutes at 22.7 years of age) after which tool bout duration declines. Such results are remarkably similar to the patterns seen in honey bees (Apis mellifera), oystercatchers (Haematopus ostralegus), chimpanzees, and human forager societies
(Helton 2008). In addition, both the amount of time a female dedicates to sponge foraging at a given age and the sponge density in her home range have a positive linear effect on her tool bout duration (Table 3 & 7).
The increase in tool use efficiency until midlife may be the result of several, non- mutually exclusive mechanisms. First, as females gain more experience, they may learn which tools last longer and consequently, show greater tool selectivity. We found that dolphins use at least three different sponge types (Echinodictyum sp., Ircinia sp. Fig. 1, and Pseudoceratina sp.) that have diverse physical properties (Hooper & Van Soest 2002). Perhaps it takes years of experience to learn which tools work best. It could also be that the best sponge tools are the most difficult to detach and either it takes time to learn these skills, or younger animals are not
105 physically able to do so. However, after females reach physical maturity at 16 years (Cheal &
Gales 1992), they still spend over six years or ~ 27 % of their average lifespan (25 years, Wells
& Scott 1990) fine-tuning their tool-use skills, so physical ability is unlikely to fully explain our results. Another possibility is that middle-aged females use their sponges in a way that promotes extended tool use. For example, by probing the substrate only where prey are likely to be hiding, abrasive contact between the sponge and the seafloor can be minimized allowing a single sponge tool to last longer. Dolphins could also be improving their ability to retrieve sponges after having dropped them to consume prey. The channels in Shark Bay experience fairly constant, strong tidal current (Burling et al. 2003, Chapter 1) that likely washes away any unattached sponges. It may be that after years of sponging, a dolphin becomes skilled at predicting where her sponge might be based on her last location and the direction of water flow. Finally, middle-aged females may be more efficient mothers, which in turn might allow them to be more efficient spongers.
Future studies investigating female tool selectivity, tool-use technique, time to sponge retrieval, and lifetime changes in maternal care may shed light on the mechanism behind the patterns observed here.
The positive relationship between the proportion of time a dolphin spends sponging and the length of time she uses a single sponge tool is certainly not surprising and supports the idea that prolonged use of a single sponge tool is efficient. Regardless of age, the most efficient way to increase time spent sponging is simply to hold onto a sponge and keep foraging. In this sense, one could argue that as an animal increases her sponging effort (percent time sponging), efficiency might just scale accordingly. This is precisely why we included age-specific percent time sponging in our analysis: to examine lifetime changes in efficiency while controlling for any effects related to lifetime changes in foraging effort.
106 Initially we expected a negative relationship between the sponge density in an animal’s home range and the length of time she uses a single sponge tool. If more tools are available, the cost of finding a replacement tool is low and dolphins might be less likely to retrieve their tools between prey captures. However, the opposite pattern was observed, possibly because an increase in sponge abundance allows dolphins to be more selective and pick superior tools that last longer. Such an interpretation would support the notion that not all basket sponges are created equal; perhaps tool selectivity is the result of a dolphin’s lifetime experience, but ecologically constrained by the available tools. Future work exploring tool use selectivity as a function of sponge density will help shed light on this subject.
Traveling with a tool
Previous studies have noted that dolphins sometimes transport their tools to different areas to forage (Mann et al. 2008). We found that the amount of time dolphins travel with a single sponge also shows a negative quadratic relationship with age while controlling for age- specific time spent traveling and sponge density (Table 4 & 8, Fig. 4), although large confidence intervals indicate poor reliability in parameter estimates. Here, the predicted peak of 53 minutes occurs at age 26.6.
Increased traveling with a sponge may indicate more knowledge concerning potential foraging locations, the distribution of quality sponge tools, and/or be a sign of increased forethought. After foraging in one location, experienced females may plan ahead and transport their tools to new locations with a greater abundance of prey. Females may also retain sponges when traveling to new foraging areas if they know that no suitable sponges exist at their next
107 intended location. Similarly, when searching for sponges females may travel to locations with superior tools and then transport these to foraging areas, rather than using inferior sponges near their foraging sites. Regardless, traveling with one’s tool means a single sponge can be used for more than one foraging bout and in more than one location, which ultimately increases foraging efficiency.
Locating and detaching sponges
Dolphins appear to reduce their effort dedicated to locating and detaching tools with age
(Fig. 5). To test the significance of this relationship, we analyzed the effects of age and sponge density on the ratio of sponge locating and detaching, to sponge foraging (hereafter the search to forage ratio) before and after the peak in tool use efficiency. We found that the ratio decreases with age before 22.7 years, but shows no significant relationship afterward (Fig. 6, Table 5 & 9).
Of the total time older females (> 22.7 years) spend on sponging related activities (time locating, detaching, and foraging with sponges), on average less than 7% (mean = 6.33%, SD = 5.89%) is dedicated to locating and detaching behavior. Sponge density does not appear to have an effect regardless of age period (Table 5 & 9).
Both Sargeant et al. (2007) and Tyne et al. (2011) found that basket sponges are not equally distributed throughout Shark Bay and this partially explains why sponging generally only occurs in deep channel habitat (Sargeant et al. 2007; Tyne et al. 2012). However, here we found that even within deep channel habitat, sponges grow in patches (Fig. 7). Furthermore, while many spongers’ home ranges overlap, they differ in sponge density (Fig. 7), which seems to partially explain some tool use behavior (Table 3 & 7). Despite this, sponge density does not
108 seem to affect the search to forage ratio. One might predict that higher sponge densities should result in a lower search to forage ratio and indeed the sponge density coefficient estimates are negative, but not significant (Table 5 & 9). However, sponge density per se may not matter if sponges are distributed in patches and young dolphins with little experience have little knowledge of their locations. Dolphins in Shark Bay have stable home ranges throughout their lives (Tsai & Mann 2012), so older individuals should be more familiar with their home range and the sponge distribution within. Thus, dolphins may increase their spatial knowledge concerning the distributions of tools until midlife, when perhaps they know their home ranges as well as they ever will. This spatial learning ultimately means less time is spent locating sponges and inevitably frees up more time and energy for other activities. Alternatively, one could argue that as dolphins increase the length of time they use a single tool, search effort should decrease even in the absence of improved spatial memory since reusing one’s tool negates the need to find another. Indeed this is why using a single tool for longer is more efficient. Yet, if our result of a decrease in the search to forage ratio up until midlife are driven solely an age related increase in tool bout duration, we would expect the search to forage ratio to increase again late in life in accordance with a decrease in tool bout duration, which is the case (Fig. 3 & 6). Thus, it appears likely that dolphins become more knowledgeable of sponge distributions with experience.
Energetic demands
Reducing energetic costs and increasing foraging efficiency may be particularly important for females with a dependent offspring. Not surprisingly then, we find a negative quadratic relationship between age and the probability a female is lactating, with a occurring at
109 an age very near the peak in tool use efficiency (Table 6 & 10, 0.80 probability of lactating at age 24, Fig. 8, 85.3 minutes tool bout at age 22.7, Fig. 3). Thus, a female sponger in her mid 20s uses sponge tools more efficiently, transports her tools from place to place, and may have valuable knowledge of the distributions of prey, sponges, or both, during the period of her life when she is most likely to be nursing a calf.
The convergence of improved sponging performance and increased lactation in midlife suggests that lifetime learning may help female bottlenose dolphins cope with the energetic costs of maternal care. Before sexual maturity females increase their sponging effort with age (Fig. 2), which may help meet the energy requirements of growth and perhaps their first impending calf.
However, beyond this, time spent sponging remains fairly stable (Fig. 2), so perhaps improving efficiency allows females to meet their energetic demands thereafter (Fig. 3). Of course, given the observational nature of our study, we cannot say whether becoming an expert sponger drives increases in reproductive output, or if a peak in reproduction relating to physiological factors simply drives females to be more efficient at this age, but in either case, being an efficient forager at this time is certainly advantageous. Indeed, reproductive output in many animals (e.g. chimpanzees and humans, Thompson et al. 2007; red deer, Cervus elaphus, Clutton-Brock et al.
1982; meerkats, Suricata suricatta, Sharp & Clutton-Brock 2010; short-tailed shearwaters,
Puffinus tenuirostris, Wooller et al. 1990; wandering albatross, Diomedea exulans, Lecomte et al. 2010) follows an inverted-U pattern with age, which in some cases may also be linked to lifetime learning.
Whether or not sponging efficiency correlates with female reproductive success remains unknown. Many offspring do not survive to weaning (Mann et al. 2000), meaning that a peak in the probability of lactating doesn’t necessarily translate to a peak in age specific reproductive
110 success. In fact, meerkats also show a similar negative quadratic relationship between age and reproductive output, but this does not correspond to increased rates of pup survival to independence (Sharp & Clutton-Brock 2010). We are just now in a position to examine such lifetime patterns, having followed individual females from sexual maturity to death, so future studies will certainly shed light on how sponging efficiency relates to fitness. Nevertheless, we do know that calving success of spongers does not significantly differ from their non-tool-using counterparts (Mann et al. 2008). In fact, in many ways sponging seems more costly compared to other foraging behaviors, but perhaps by continuing to improve sponging efficiency females are able to mitigate these costs and achieve calving success similar to the rest of the Shark Bay population.
Implications
The development of expertise in non-human animals has broad implications concerning life history, foraging theory, individual specialization, and may influence social learning and cultural transmission.
First, the inverted U function seen here has been previously reported for lifetime performance in many domains across taxa including reproduction (Clutton-Brock et al. 1982;
Sydeman et al. 1991; Derocherand & Stirling 1994; Lunn et al. 1994; Black et al. 1996; Ericsson et al. 2001; Krüger & Lindström 2001; Thompson et al. 2007; Sharp & Clutton-Brock 2010), sociality (Watts & Mitani 2001; Robbins & Robbins 2005), foraging (Heiling & Herberstein
1999; Helton 2008; MacNulty et al. 2009), memory (Connor 2001; Tapp & Siwak 2003; Lyons-
Warren et al. 2004; Cowan & Naveh-Benjamin 2006; Brehmer et al. 2008; Nagel et al. 2009;
111 Castel et al. 2011), as well as other cognitive abilities like reaction time (Williams et al. 2005), inhibitory control (Williams et al. 1999; Bedard & Nichols 2002), executive function (Kray et al.
2004; Zelazo et al. 2004), problem solving (Denney & Palmer 1981), and tool use (Helton 2008).
Many of these patterns may represent sustained learning (Ericsson et al. 2006; Helton 2008;
Dukas 2008a, 2008b), but are also likely due to similar underlying developmental and senescent processes across species (Schulz & Heckhausen 1996; Craik & Bialystok 2006; Dukas 2008b).
In general, early life is characterized by a continuous increase in both cognitive and physical abilities until growth levels off. In some species, learning and performance extends beyond physical maturation. Much of this learning is probably achieved through extensive practice, in what has been called ‘learning by doing’ (Arrow 1962). Once in their prime, individuals may sustain high performance until both physical and cognitive function decline as a result of aging.
It is not known exactly how spongers improve their tool use proficiency, but long-term experience likely allows dolphins to make subtle improvements in multiple aspects of sponging
(Ericsson et al. 2006). In particular, one area suspected to be extremely important for human experts is memory (Ericsson & Kintsch 1995). Measures of short-term memory, particularly working memory, consistently show a negative quadratic relationship with age (Tapp & Siwak
2003; Lyons-Warren et al. 2004; Cowan & Naveh-Benjamin 2006; Nagel et al. 2009). For dolphins, an apt working memory is probably vital for retrieving sponges in between prey captures, which could easily explain our results in Figure 4. Long-term memory on the other hand allows information to be stored more permanently and may to be more robust to the effects of aging (Connor 2001; Lyons-Warren et al. 2004). Perhaps this is why, even though we see an age related decline the tool bout duration (Fig. 3), we do not see an increase in time spent locating and detaching sponges late in life (Fig. 6). Thus, older females with an aging memory
112 may have difficulty retrieving sponges during foraging, but maintain their knowledge of the distribution of sponges.
Second, our results support classical and more recent theoretical foraging models.
Classical foraging theory predicts that animals should maximize efficiency (energy gained per unit energy expenditure) when there is a shortage of energy, but rate (energy gained per unit time) when there is a shortage of time (Stephens & Krebs 1986; Ydenberg 1998). Consistent with this, female dolphins appear to maximize efficiency at a time when they are energetically stressed (have nursing offspring). More recently, Dukas and Clark (1995) predicted that animals should employ long foraging bouts when resources are easy to obtain and/or when the cost of switching to non-foraging behaviors is high. As predicted then, spongers, who hunt easy to catch prey and have a high cost of activity switching due to the loss of a tool, appear to learn this optimal strategy and increase the time they forage with a single sponge tool.
Third, our results support existing theory predicting that complex foraging behaviors requiring extensive learning can lead to individual specialization (Hughes 1979; Tinker et al.
2009). Spongers spend over 20 years of their lives improving their tool use performance. This extensive learning may indicate individuals must specialize in sponging for it to be beneficial. In humans, the development of expertise leads to extreme specialization and the use of a narrow, unique niche (Agnew et al. 1997; Ericsson et al. 2006). Fittingly, adult female spongers spend ~
96 % of their foraging time sponging (Mann et al. 2008) and appear to target few prey species that may be inaccessible to other non-tool-using dolphins (Patterson & Mann 2011).
This lifelong learning may preclude the existence of part-time female spongers, and ultimately male spongers. For example, an occasional female sponger would have an experience level similar to young dolphins, meaning she would use tools inefficiently and have to spend
113 considerable effort locating and detaching sponges, which would ultimately make sponging a costly foraging tactic for her. Females may solve this problem by specializing, and indeed all females that sponge forage are highly specialized, but males may not have this option.
Specializing in sponging would restrict males ranging and habitat use (Randic et al. 2012,
Chapter 1), and their ability to form alliances necessary for gaining access to females (Connor et al. 2006; Mann et al. 2008). Unfavorably then, males can only sponge forage part-time, which may explain why just seven males use sponge tools, and do so less frequently than females
(unpublished data). In fact, it is for this reason that we were unable to include males in this study.
However, male spongers should not forever be neglected on the basis of small sample size since the infrequency of sponging by males in its self is interesting, and future studies of male spongers are certain to provide insight into many aspects of the behavior otherwise inaccessible.
Finally, our data suggest the mode of cultural transmission (vertical, horizontal, oblique) of socially learned behaviors may depend on the time required to make the behavior beneficial.
For example, the cost of learning to sponge forage late in life may be too high since the efficiency required to meet ones reproductive demands might not be reached until well after a dolphin’s average lifespan. Exposure to sponging early in life may be critical for adopting the behavior which, given an extended period of maternal dependency, means a maternal demonstrator is necessary. While it is true that calves of non-spongers sometimes associate with sponging females, such calves have little exposure to sponging since it is a solitary behavior
(Mann et al. 2008; Sargeant & Mann 2009). Therefore, only calves of spongers have sufficient exposure to sponging during this formative period. Of course at least one dolphin must have learned the behavior without a sponger mother in order for sponging to be established in the population, and genetic data suggest either multiple innovation events or at least some horizontal
114 transmission has occurred (Krützen et al. 2005; Ackermann 2008). However, in modeling the emergence of sponging, Kopps & Sherwin (2012) suggest that at most, horizontal transmission and/or repeated innovation events occur at very low frequencies. Thus, under certain conditions the existence of vertical cultural transmission and lack of horizontal may simply be the result of timing. Nevertheless, this does not preclude horizontal or oblique transmission between spongers. Spongers show strong homophily (tendency to associate with other spongers) outside their matriline (Mann et al. 2012), which may facilitate shared knowledge concerning sponge and/or prey distributions, superior sponge types, tool using techniques, and/or other aspects of sponging. In particular, oblique social learning may be adaptive as older individuals demonstrate superior performance. Examining the relationship between sponging performance and association with other non-maternal spongers may reveal if dolphins, like humans, learn from the experts in their field.
EXPERIMENTAL PROCEDURES
Study site and data collection
The Shark Bay Dolphin Research Project has an extensive 29-year database that includes demographic, genetic, association, life-history, ecological, and behavioral data on 1,400 dolphins residential to a 500 km2 area off shore of Monkey Mia (25° 47’ S, 113° 43’ E, Fig. 7). Habitat consists of embayment plains (5–13 m), shallow sand flats (0.5–4 m), seagrass beds (0.5–4 m), and bisecting deep channels (7–13 m). Behavioral data for this study were collected during focal follows of individual females or mother-calf pairs from 1989-2012 in which systematic point and
115 continuous samples on behavior were collected (Altmann 1974; Mann 1999). All dolphin identities were determined using standard Photo ID methods (Würsig & Würsig 1977). We classified any female seen carrying a sponge on more than one day as a known sponger (Mann et al. 2008). Age was determined by either known birth year or first reproductive event. In the case these data were not available, age was estimated using degree of ventral speckling (Krzyszczyk
& Mann 2012) and/or size estimates. Sex was determined by either the presence of a dependent calf or by observing a female’s mammary slits. Whether or not a female was lactating was determined by the presence of a dependent calf (see Mann et al. 2000 for further details). For all analyses, only individuals known to be spongers, with an estimated age, and at least three hours of focal follow data per age were included. If a female had three hours or more of focal data at some ages, but less at others, only her data from ages with at least 3 hours of observations were included.
Observers recorded when a female obtained and dropped sponge tools. In general, most sponges are uniquely identifiable by their color and shape allowing observers to easily determine when a dolphin changes sponges, but whenever possible we confirmed each sponge’s uniqueness using photographs (e.g. Fig. 1). If a dolphin performed non-sponging behavior for at least five minutes and subsequently obtained a sponge, the sponge was considered to be new and a tool bout began. Because dolphins regularly drop sponges to consume prey, the final stop-time of a sponge tool was considered to occur either at the last minute a dolphin had a particular sponge before switching to a new one, or when a dolphin dropped their sponge and performed non- sponging behavior for at least five minutes. Because our focal follows did not always capture the full duration a single sponge tool was used, many tool bout durations are censored. However, we adjusted all analyses to handle censored data where appropriate.
116 To obtain an estimate of basket sponge density, divers performed 39 video belt transects
(Hill & Wilkinson 2004; Eleftheriou & McIntyre 2005; Patterson & Mann 2011) randomly located in two sampling areas that contain ~78% of all historical sponge foraging sightings (Fig.
7). On each transect, which was at least 200 m from all other transects, divers filmed a 50 m long, 2 m wide area of the substrate using a Sony HDR-XR500V HD video camera in an
AquaticaHD housing. Divers initially dove down and weighted one end of a 50 m tape measure and then swam away at a bearing consistent with the direction of the current filming the substrate. Swimming in this direction minimizes within transect heterogeneity as different habitat types within channels tended to occur in strips in line with the direction of tidal flow. All dives were performed using an Airline Supply R360XL Hookah System by J. Sink and were swam at a consistent speed of ~ 0.28 m/s. All video footage was logged by E.M.P. and only basket sponges of an appropriate size and shape were included in sponge abundance estimates.
To identify sponge species, divers opportunistically collected small samples from the basal attachment to the outermost edge of all sponge types used by dolphins and preserved them in 100% ethanol. Sponge taxonomy was determined by Dr. Fromont at the Western Australian
Museum. Briefly, skeletal structure and spicule morphology were examined using light microscopy. Spicules and skeletons were prepared by hand cutting thin sections of sponge
(including the ectosome and choanosome), clearing tissue in household bleach, rinsing with distilled water, and mounting the material on a glass slide. Spicule dimensions were determined using an eyepiece graticule with an Olympus BX50 microscope.
To obtain home range estimates, we employed a new adaptive local convex hull method
(a-LoCoH) that is known to outperform several other home range estimators, especially in areas with hard boundaries such as shoreline (Getz & Wilmers 2004; Calenge 2006; Getz et al. 2007,
117 Chapter 1). Spatial data for all females were collected during surveys, which consisted of short
5–60 minute behavioral observations in which locational, behavioral, and ecological data were collected for all dolphins (Gibson & Mann 2009). If an animal was sighted more than once on the same day, only the last location point was used to minimize biases associated with our boat- launching site (Monkey Mia, Fig. 7). Spatial data were compiled across ages for all animals as
Shark Bay dolphins have stable home ranges throughout their lives (Tsai & Mann 2012).
Following the heuristic rule proposed by Getz et al. (2007), we used the average maximum distance between any two location points for each animal as the adaptive a parameter in constructing a-LoCoH estimates (Chapter 1). The 90% isopleths were then used as the boundary of an animal’s home range since these are known to be less biased to differences in sample size compared to the more standard 95% isopleths (Fig. 7; Börger et al. 2006; Getz et al. 2007).
Finally, we calculated an estimate of the sponge density for each animal separately using data from only those transects that fell within that animal’s home range (Fig. 7). Because we used all available data to estimate home ranges, differences in home range size between animals could simply be due to an unequal number of location points (Urian et al. 2009). However, because we are only interested in obtaining an estimate of sponge density, we are less concerned with home range size and the effect sampling effort has on it.
Genetic analysis
During a focal follow in 2010, we collected a sponger fecal sample and stored it at –20
°C in 70% ethanol. A genetic diet analysis of this fecal sample was later performed by Yvette
Hitchen and Dr. Jason Kennington of Helix Molecular Solutions Pty Ltd. Briefly DNA was
118 extracted from dolphin feces using Qiagen QIAamp DNA Stool Mini Kit. A fecal pellet was formed via centrifuging 2 ml of the sample at 14,000 rpm (8,500G) for 1 min and decanting ethanol. This was repeated in the same tube until the sample completely pelleted, which was then eluted into two volumes of 30 µl and 100 µl. In a separate lab, DNA from P. nebulosa (collected for Patterson & Mann 2011 and stored at –80 °C) was extracted using the Qiagen DNeasy Blood and Tissue Kit and was eluted into two volumes of 30 µl and 100 µl. The mitochondrial cytochrome oxidase subunit I gene (cox1) of P. nebulosa was amplified using primers FishF2 and FishR2 (Ward et al. 2005, Table 1) resulting in a 679 bp fragment. From this, a new primer set was designed (Oligo version 6, Molecular Biology Insights Incorporated 2011) to amplify a shorter 179 bp fragment used to amplify prey DNA extracted from the dolphin fecal sample
(Table 1). Samples were sequenced using capillary separation on the AB 3730xl through AGRF
(Australian Genome Research Facility) and resulting fragments were edited using Sequencher
4.10.1 (Gene Codes Corporation 2011) and aligned using Mega 5.05 (Tamura et al. 2011).
Statistical analyses
To examine changes in the percent time sponging and in the ratio of locating and detaching to sponging as a function of age, we applied LOESS curves to visual data from 19 females observed at multiple ages. Because it is often difficult in the field to determine the precise moment that an animal has found a sponge, and thus switches from locating to detaching, we do not distinguish between the two behaviors in any analyses. We used a span value of = 0.5 and a second order polynomial with a redescending M-estimator with Tukey's biweight function and also plotted the square root of the variance function (Figs. 2 and 6). These methods were
119 employed to help visualize data and describe possible patterns, not to test for statistical significance.
After initial visual inspection of the data, we employed the use of several Markov chain
Monte Carlo generalized linear mixed models (MCMC glmm) in a Bayesian framework. To examine the development of tool use efficiency we used a MCMC glmm with a censored
Poisson error structure, the log link function, and additive overdispersion. Age, age2, age-specific percent time sponge foraging, and the sponge density in an animal’s home range were treated as fixed effects, while individual identity and day of observation were included as random factors.
The response variable here was time foraging (minutes) with a single sponge tool, or tool bout duration. The same model specification was used to assess the effects of age, age2, age-specific percent time traveling, and the sponge density on the time traveling with a single sponge.
To examine the effect of age on the ratio of time spent locating for and detaching sponges, to time spent foraging with sponges we utilized a MCMC glmm with a binomial error structure, logit link function, and additive overdispersion. Here, we created two models, one before and one after the peak in tool use efficiency observed at age 22.7. Both models included age and sponge density as fixed effects, individual identity as a random effect, and the response variable was the ratio of sponge locating and detaching to foraging. Here, a quadratic term was not used since only a linear trend was visible from the LOESS smoothing curve.
Finally, to explore how lifetime performance in tool use relates to the costs of reproduction, we used a MCMC glmm with a binomial error structure and the logit link function to examine the effects of age and age2 on female lactation. Individual identity was again controlled for as a random effect. Here the response variable was binary, 0 – not lactating, 1 – lactating so overdispersion cannot occur (Collett 2003) and the residual variance was fixed at 1
120 to allow proper mixing (Hadfield 2010a). We did not include an intercept in this model as by definition an unborn animal has zero probability of lactating.
All models were run using the MCMCglmm package in R (Hadfield 2010b; R
Development Core Team 2011), with weak priors, 5,000,000 iterations, a burn-in of 10,000, and a 1,000 thinning interval. Because the linear and quadratic age terms by definition are highly correlated, we employed orthogonal polynomials in all models with age and age2 to reduce problems associated with multicollinearity (Narula 1979). Proper mixing and model convergence were accessed through trace and posterior distributions plots, as well as Raftery and Lewis's diagnostic (Raftery & Lewis 1992), Heidelberger and Welch's diagnostic (Heidelberger & Welch
1983), and Geweke's diagnostic (Geweke 1991) from the coda package in R (Plummer et al.
2006; R Development Core Team 2011).
MCMC glmms have several advantages over maximum likelihood (ML) techniques. For example, MCMC glmm methods easily extend to incorporate multiple random factors for non-
Gaussian models. Parameter estimates are also often more accurate and testing the significance of these using the posterior distribution and confidence intervals is fairly straightforward when compared to a ML framework (Thomas et al. 2006; Bolker et al. 2009; Hadfield 2010b; Wilson et al. 2010). However, for comparison we also performed all analyses using ML techniques and
Cox proportional-hazards models for censored data. Glmm analyses were performed using the lme4 package in R (Bates et al. 2011; R Development Core Team 2011), while Cox proportional-hazards models were created using the coxme package (Therneau 2012). Because
Cox proportional-hazard models estimate an underlying hazard function and not a function of the original data, parameter estimates are expected to have the opposite sign when compared to the
MCMC glmm results. Results from both the MCMC and ML methods were nearly identical,
121 which is often the case when using weak priors (Bolker et al. 2009). As such, only the results from the MCMC glmms are discussed and were used for figures, but statistical tables from both methods are included.
ACKNOWLEDGEMENTS
We thank our field assistants Kippy Searles, Jenny Smith, and Katie Gill, all members of the Mann lab past and present, and all members of the Shark Bay Dolphin Research Project
(SBDRP) who contributed to the long-term data. We are grateful Dr. Jane Fromont for determining sponge taxonomy and also for her guidance in designing transects. We also thank the Shark Bay Ecosystem Research Project for help in the field, and the Monkey Mia Dolphin
Resort and the Department of Environment and Conservation of Western Australia (DEC) for field and logistical support. All animal work was approved by the Georgetown University
Animal Care and Use Committee (GUACUC) under permits 07–041 and 10–023. Observational and field studies were approved by DEC under permits SF007418, SF007975, SF006897, and
SF007457. Support for this study was provided by Georgetown University, the National
Geographic Society Young Explorers Grant, the Explorers Club Exploration Fund Grant, the
Achievement Rewards for College Scientist, the Animal Behavior Society Cetacean Behavior and Conservation Award, and the American Society of Mammologists Grant in Aid of Research to EMP. Additional funding was provided by the National Science Foundation Grants (NSF)
0847922, 0316800, 0918303, 0918308, Georgetown University, and Office of Naval Research
Grant 10230702 to JM. The orthophoto in Figure 1 was supported by NSF grant OCE0526065 to
122 Dr. Mike Heithaus and DEC. Field gear was supplied at discounted rates by Airline Supply by J.
Sink and Splash Dive Center of Alexandria, VA.
123 REFERENCES
Ackermann, C. 2008. Contrasting vertical skill transmission patterns of a tool use behaviour in
two groups of wild bottlenose dolphins (Tursiops Sp.) as revealed by molecular cenetic
analyses. M.Sc. thesis. Zurich, Switzerland: University of Zurich, Anthropological Institute
and Museum.
Agnew, N., Ford, K. & Hayes, P. 1997. Expertise in context: personally constructed, socially
selected and reality-relevant? In: Expertise in Context: Human and Machine, (Ed. by P.
Feltovich, K. M. Ford, & R. R. Hoffman), pp. 219–244. Menlo Park: MIT Press.
Altmann, J. 1974. Observational study of behavior: sampling methods. Behaviour, 49, 227–267.
Arrow, K. 1962. The economic implications of learning by doing. The Review of Economic
Studies, 29, 155–173.
Bates, D., Maechler, M. & Bolker, B. 2011. lme4: Linear mixed-effects models using S4
classes. R package version 0.999999-0, http://cran.r-project.org/package=lme4.
Bedard, A. & Nichols, S. 2002. The development of selective inhibitory control across the life
span. Developmental Neuropsychology, 21, 93–111.
Black, J. M., Choudhury, S. & Owen, M. 1996. Do barnacle geese benefit from lifelong
monogamy? In: Partnerships in Birds: The Study of Monogamy, (Ed. by J. M. Black), pp.
91–117. Oxford: Oxford University Press.
Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H.
& White, J.-S. S. 2009. Generalized linear mixed models: a practical guide for ecology and
evolution. Trends in Ecology & Evolution, 24, 127–135.
124 Brehmer, Y., Li, S.-C., Straube, B., Stoll, G., von Oertzen, T., Müller, V. & Lindenberger,
U. 2008. Comparing memory skill maintenance across the life span: preservation in adults,
increase in children. Psychology and Aging, 23, 227–238.
Burling, M. C., Pattiaratchi, C. B. & Ivey, G. N. 2003. The tidal regime of Shark Bay,
Western Australia. Estuarine, Coastal and Shelf Science, 57, 725–735.
Börger, L., Franconi, N., De Michele, G., Gantz, A., Meschi, F., Manica, A., Lovari, S. &
Coulson, T. 2006. Effects of sampling regime on the mean and variance of home range size
estimates. The Journal of Animal Ecology, 75, 1393–1405.
Calenge, C. 2006. The package “adehabitat” for the R software: a tool for the analysis of space
and habitat use by animals. Ecological Modelling, 197, 516–519.
Castel, A. D., Humphreys, K. L., Lee, S. S., Galván, A., Balota, D. A. & McCabe, D. P.
2011. The development of memory efficiency and value-directed remembering across the
life span: a cross-sectional study of memory and selectivity. Developmental Psychology, 47,
1553–1564.
Cheal, A. & Gales, N. 1992. Growth, sexual maturity and food intake of Australian Indian
Ocean bottlenose dolphins, Tursiops truncatus, in captivity. Australian Journal of Zoology,
40, 215–223.
Clutton-Brock, T. H., Guinness, F. E. & Albon, S. D. 1982. Red Deer: Behavior and Ecology
of Two Sexes. Chicago: University of Chicago Press.
Collett, D. 2003. Modelling Binary Data. 2nd edn. Boca Raton: Chapman & Hall/CRC.
Connor, L. 2001. Memory in old age: patterns of decline and preservation. Seminars in Speech
and Language, 22, 117–125.
125 Connor, R. C., Smolker, R. & Bejder, L. 2006. Synchrony, social behaviour and alliance
affiliation in Indian Ocean bottlenose dolphins, Tursiops aduncus. Animal Behaviour, 72,
1371–1378.
Cowan, N. & Naveh-Benjamin, M. 2006. Life-span development of visual working memory:
when is feature binding difficult? Developmental Psychobiology, 42, 1089–1102.
Craik, F. I. M. & Bialystok, E. 2006. Cognition through the lifespan: mechanisms of change.
Trends in Cognitive Sciences, 10, 131–138.
Denney, N. W. & Palmer, A. M. 1981. Adult age differences on traditional and practial
problem-solving measures. Journal of Gerontology, 36, 323–328.
Derocherand, A. E. & Stirling, I. 1994. Age-specific reproductive performance of female polar
bears (Ursus maritimus). Journal of Zoology, 234, 527–536.
Dukas, R. 1998. Evolutionary ecology of learning. In: Cognitive Ecology: the Evolutionary
Ecology of Information Processing and Decision Making, (Ed. by R. Dukas), pp. 129–174.
Chicago: University of Chicago Press.
Dukas, R. 2008a. Life history of learning: performance curves of honeybees in settings that
minimize the role of learning. Animal Behaviour, 75, 1125–1130.
Dukas, R. 2008b. Life history of learning: performance curves of honeybees in the wild.
Ethology, 114, 1195–1200.
Dukas, R. 2009. Learning mechanisms, ecology, and evolution. In: Cognitive Ecology II, (Ed.
by R. Dukas & J. Ratcliffe), pp. 7–26. Chicago: University of Chicago Press.
Eleftheriou, A. & McIntyre, A. 2005. Methods for the Study of Marine Benthos. 3rd edn.
Oxford: Blackwell.
126 Ericsson, K. A. & Kintsch, W. 1995. Long-term working memory. Psychological Review, 102,
211–245.
Ericsson, G., Wallin, K., Ball, J. & Broberg, M. 2001. Age-related reproductive effort and
senescence in free-ranging moose, Alces alces. Ecology, 82, 1613–1620.
Ericsson, K. A., Charness, N., Feltovich, P. J. & Hoffman, R. R. 2006. The Cambridge
Handbook of Expertise and Expert Performance. Cambridge: Cambridge University Press.
Gene Codes Corporation. 2011. Sequencher® version 5.1 sequence analysis software. Ann
Arbor: Gene Codes Corporation.
Getz, W. M. & Wilmers, C. C. 2004. A local nearest-neighbor convex-hull construction of
home ranges and utilization distributions. Ecography, 4, 489–505.
Getz, W. M., Fortmann-Roe, S., Cross, P. C., Lyons, A. J., Ryan, S. J. & Wilmers, C. C.
2007. LoCoH: nonparameteric kernel methods for constructing home ranges and utilization
distributions. PloS ONE, 2, e207.
Geweke, J. 1991. Evaluating the accuracy of sampling-based approaches to calculating posterior
moments. In: Bayesian Statistics: Proceedings of the Fourth Valencia International
Meeting, 4th edn. (Ed. by J. Bernado, J. Berger, A. Dawid, & A. Smith), pp. 169–193.
Oxford: Clarendon Press.
Gibson, Q. A. & Mann, J. 2009. Do sampling method and sample size affect basic measures of
dolphin sociality? Marine Mammal Science, 25, 187–198.
Gurven, M., Kaplan, H. & Gutierrez, M. 2006. How long does it take to become a proficient
hunter? Implications for the evolution of extended development and long life span. Journal
of Human Evolution, 51, 454–470.
127 Hadfield, J. D. 2010a. MCMCglmm: Markov chain Monte Carlo methods for generalised linear
mixed models. Tutorial for MCMCglmm package in R, 1–25.
Hadfield, J. D. 2010b. MCMC methods for multi-response generalized linear mixed models: the
MCMCglmm R package. Journal of Statistical Software, 33, 1022.
Heidelberger, P. & Welch, P. D. 1983. Simulation run length control in the presence of an
initial transient. Operations Research, 31, 1109–1144.
Heiling, A. M. & Herberstein, M. E. 1999. The role of experience in web-building spiders
(Araneidae). Animal Cognition, 2, 171–177.
Helton, W. S. 2005. Animal expertise, conscious or not. Animal Cognition, 8, 67–74.
Helton, W. S. 2008. Expertise acquisition as sustained learning in humans and other animals:
commonalities across species. Animal Cognition, 11, 99–107.
Herman, L. 2002. Exploring the cognitive world of the bottlenosed dolphin. In: The Cognitive
Animal: Empirical and Theoretical Perspectives on Animal Cognition, (Ed. by M. Bekoff,
C. Allen, & G. M. Burghardt), pp. 275–283. Cambridge: MIT Press.
Hill, J. & Wilkinson, C. 2004. Methods for Ecological Monitoring of Coral Reefs. Townsville:
Australian Institude of Marine Science.
Holzhaider, J. C., Hunt, G. R. & Gray, R. D. 2010. The development of pandanus tool
manufacture in wild New Caledonian crows. Behaviour, 147, 553–586.
Hooper, J. N. A. & Van Soest, R. W. M. 2002. Systema Porifera: a Guide to the Classification
of Sponges. New York: Kluwer Academic.
Hughes, R. 1979. Optimal diets under the energy maximization premise: the effects of
recognition time and learning. The American Naturalist, 113, 209–221.
128 Inoue-Nakamura, N. & Matsuzawa, T. 1997. Development of stone tool use by wild
chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 111, 159–173.
Kaplan, H., Hill, K., Lancaster, J. B. & Hurtado, A. M. 2000. A theory of human life history
evolution: diet, intelligence, and longevity. Evolutionary Anthropology, 9, 156–185.
Kopps, A. M. & Sherwin, W. B. 2012. Modelling the emergence and stability of a vertically
transmitted cultural trait in bottlenose dolphins. Animal Behaviour, doi:
10.1016/j.anbehav.2012.08.029.
Kray, J., Eber, J. & Lindenberger, U. 2004. Age differences in executive functioning across
the lifespan: the role of verbalization in task preparation. Acta Psychologica, 115, 143–165.
Krzyszczyk, E. & Mann, J. 2012. Why become speckled? Ontogeny and function of speckling
in Shark Bay bottlenose dolphins (Tursiops sp.). Marine Mammal Science, 28, 295–307.
Krüger, O. & Lindström, J. 2001. Lifetime reproductive success in common buzzard, Buteo
buteo: from individual variation to population demography. Oikos, 2, 260–273.
Krützen, M., Mann, J., Heithaus, M. R., Connor, R. C., Bejder, L. & Sherwin, W. B. 2005.
Cultural transmission of tool use in bottlenose dolphins. Proceedings of the National
Academy of Sciences of the United States of America, 102, 8939–8943.
Lecomte, V. J., Sorci, G., Cornet, S., Jaeger, A., Faivre, B., Arnoux, E., Gaillard, M.,
Trouvé, C., Besson, D., Chastel, O. & Weimerskirch, H. 2010. Patterns of aging in the
long-lived wandering albatross. Proceedings of the National Academy of Sciences of the
United States of America, 107, 6370–6375.
Lonsdorf, E. V. 2005. Sex differences in the development of termite-fishing skills in the wild
chimpanzees, Pan troglodytes schweinfurthii, of Gombe National Park, Tanzania. Animal
Behaviour, 70, 673–683.
129 Lunn, N. J., Boyd, I. L. & Croxall, J. P. 1994. Reproductive performance of female antarctic
fur seals: the influence of age, breeding experience, environmental bariation and individual
quality. The Journal of Animal Ecology, 63, 827–840.
Lyons-Warren, A., Lillie, R. & Hershey, T. 2004. Short- and long-term spatial delayed
response performance across the lifespan. Developmental Neuropsychology, 26, 661–678.
MacNulty, D. R., Smith, D. W., Vucetich, J. A., Mech, L. D., Stahler, D. R. & Packer, C.
2009. Predatory senescence in ageing wolves. Ecology Letters, 12, 1347–1356.
Mann, J. 1999. Behavioral sampling methods for cetaceans: a review and critique. Marine
Mammal Science, 15, 102–122.
Mann, J. & Sargeant, B. L. 2003. Like mother, like calf: the ontogeny of foraging traditions in
wild Indian ocean bottlenose dolphins (Tursiops sp.). In: The Biology of Traditions: Models
and Evidence, (Ed. by D. Fragaszy & S. Perry), pp. 236–266. Cambridge: Cambridge
University Press.
Mann, J., Connor, R. C., Barre, L. M. & Heithaus, M. R. 2000. Female reproductive success
in bottlenose dolphins (Tursiops sp.): life history, habitat, provisioning, and group-size
effects. Behavioral Ecology, 11, 210–219.
Mann, J., Sargeant, B. L., Watson-Capps, J. J., Gibson, Q. A., Heithaus, M. R., Connor, R.
C. & Patterson, E. M. 2008. Why do dolphins carry sponges? PloS ONE, 3, e3868.
Mann, J., Stanton, M. A., Patterson, E. M., Bienenstock, E. J. & Singh, L. O. 2012. Social
networks reveal cultural behaviour in tool-using using dolphins. Nature Communications, 3,
980.
Marino, L., Connor, R. C., Fordyce, R. E., Herman, L. M., Hof, P. R., Lefebvre, L.,
Lusseau, D., McCowan, B., Nimchinsky, E. A., Pack, A. A., Rendell, L., Reidenberg, J.
130 S., Reiss, D., Uhen, M. D., Van der Gucht, E. & Whitehead, H. 2007. Cetaceans have
complex brains for complex cognition. PLoS Biology, 5, e139.
McGrew, W. C. 1992. Chimpanzee Material Culture: Implications for Human Evolution.
Cambridge: Cambridge University Press.
Molecular Biology Insights Incorporated. 2011. OLIGO Primer Analysis Software Version 6.
Cascade: Molecular Biology Insights Incorporated.
Nagel, I. E., Preuschhof, C., Li, S.-C., Nyberg, L., Bäckman, L., Lindenberger, U. &
Heekeren, H. R. 2009. Performance level modulates adult age differences in brain
activation during spatial working memory. Proceedings of the National Academy of
Sciences of the United States of America, 106, 22552–22557.
Narula, S. 1979. Orthogonal polynomial regression. International Statistical Review, 47, 31–36.
Ormrod, J. E. 2004. Human Learning. 6th edn. Upper Saddle River: Prentice Hall.
Patterson, E. M. & Mann, J. 2011. The ecological conditions that favor tool use and innovation
in wild bottlenose dolphins (Tursiops sp.). PloS ONE, 6, e22243.
Plummer, M., Best, N., Cowles, K. & Vines, K. 2006. CODA: convergence diagnosis and
output analysis for MCMC. R News, 6, 7–11.
R Development Core Team. 2011. R: A Language and Environment for Statistical Computing,
Version 2.12.1. Vienna, Austria: R Foundation for Statistical Computing.
Raftery, A. & Lewis, S. 1992. Comment: one long run with diagnostics: implementation
strategies for Markov chain Monte Carlo. Statistical Science, 7, 493–497.
Randic, S., Connor, R. C., Sherwin, W. B. & Krützen, M. 2012. A novel mammalian social
structure in Indo-Pacific bottlenose dolphins (Tursiops sp.): complex male alliances in an
131 open social network. Proceedings of the Royal Society B: Biological Sciences, 279, 3083–
3090.
Robbins, M. & Robbins, A. 2005. Long-term dominance relationships in female mountain
gorillas: strength, stability and determinants of rank. Behaviour, 142, 779–809.
Sargeant, B. L. & Mann, J. 2009. Developmental evidence for foraging traditions in wild
bottlenose dolphins. Animal Behaviour, 78, 715–721.
Sargeant, B. L., Mann, J., Berggren, P. & Krützen, M. 2005. Specialization and development
of beach hunting, a rare foraging behavior, by wild bottlenose dolphins (Tursiops sp.).
Canadian Journal of Zoology, 83, 1400–1410.
Sargeant, B. L., Wirsing, A. J., Heithaus, M. R. & Mann, J. 2007. Can environmental
heterogeneity explain individual foraging variation in wild bottlenose dolphins (Tursiops
sp.)? Behavioral Ecology and Sociobiology, 61, 679–688.
Schulz, R. & Heckhausen, J. 1996. A life span model of successful aging. The American
Psychologist, 51, 702–714.
Sharp, S. P. & Clutton-Brock, T. H. 2010. Reproductive senescence in a cooperatively
breeding mammal. The Journal of Animal Ecology, 79, 176–183.
Smolker, R. A., Richards, A. F., Connor, R. C., Mann, J. & Berggren, P. 1997. Sponge
carrying by dolphins (Delphinidae, Tursiops sp.): a foraging specialization involving tool
use? Ethology, 103, 454–465.
Stephens, D. W. & Krebs, J. R. 1986. Foraging Theory. Princeton: Princeton University Press.
Sydeman, W. J., Huber, H. R., Emslie, S. D., Ribic, C. A. & Nur, N. 1991. Age-specific
weaning success of northern elephant seals in relation to previous breeding experience.
Ecology, 72, 2204–2217.
132 Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M. & Kumar, S. 2011. MEGA5:
molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance,
and maximum parsimony methods. Molecular Biology and Evolution, 28, 2731–2739.
Tapp, P. & Siwak, C. 2003. Effects of age on measures of complex working memory span in
the beagle dog (Canis familiaris) using two versions of a spatial list learning paradigm.
Learning & Memory, 10, 148–160.
Therneau, T. 2012. coxme: Mixed effects Cox models. R package version, 2, http://cran.r-
project.org/package=coxme.
Thomas, D., Johnson, D. & Griffith, B. 2006. A bayesian random effects discrete-choice
model for resource selection: population-level selection inference. Journal of Wildlife
Management, 70, 404–412.
Thompson, M. E., Jones, J. H., Pusey, A. E., Brewer-Marsden, S., Goodall, J., Marsden, D.,
Matsuzawa, T., Nishida, T., Reynolds, V., Sugiyama, Y. & Wrangham, R. W. 2007.
Aging and fertility patterns in wild chimpanzees provide insights into the evolution of
menopause. Current Biology, 17, 2150–2156.
Tinker, M. T., Mangel, M. & Estes, J. A. 2009. Learning to be different: acquired skills, social
learning, frequency dependence, and environmental variation can cause behaviourally
mediated foraging specializations. Evolutionary Ecology Research, 11, 841–869.
Tsai, Y.-J. J. & Mann, J. 2012. Dispersal, philopatry, and the role of fission-fusion dynamics in
bottlenose dolphins. Marine Mammal Science, 28, doi: 10.1111/j.1748–7692.2011.00559.x.
Tyne, J. A., Loneragan, N. R., Kopps, A. M., Allen, S. J., Krützen, M. & Bejder, L. 2012.
Ecological characteristics contribute to sponge distribution and tool use in bottlenose
dolphins Tursiops sp. Marine Ecology Progress Series, 444, 143–153.
133 Urian, K. W., Hofmann, S., Wells, R. S. & Read, A. J. 2009. Fine-scale population structure
of bottlenose dolphins (Tursiops truncatus) in Tampa Bay, Florida. Marine Mammal
Science, 25, 619–638.
Walker, R., Hill, K., Kaplan, H. & McMillan, G. 2002. Age-dependency in hunting ability
among the Ache of eastern Paraguay. Journal of Human Evolution, 42, 639–657.
Ward, R. D., Zemlak, T. S., Innes, B. H., Last, P. R. & Hebert, P. D. N. 2005. DNA
barcoding Australia’s fish species. Philosophical Transactions of the Royal Society B:
Biological Sciences, 360, 1847–1857.
Watts, D. P. & Mitani, J. C. 2001. Boundary patrols and intergroup encounters in wild
chimpanzees. Behaviour, 138, 299–327.
Wells, R. S. & Scott, M. D. 1990. Estimating bottlenose dolphin population parameters from
individual identification and capture-release techniques. In: Individual Recognition of
Cetaceans: Use of Photo-Identification and Other Techniques to Estimate Population
Parameters, (Ed. by P. S. Hammond, S. A. Mizroch, & G. P. Donovan), pp. 407–415.
Cambridge: Report of the International Whaling Commission, Special Issue 12.
Wen, J. Y., Kumar, N., Morrison, G., Rambaldini, G., Runciman, S., Rousseau, J. & van
der Kooy, D. 1997. Mutations that prevent associative learning in C. elegans. Behavioral
Neuroscience, 111, 354–368.
Williams, B. R., Ponesse, J. S., Schachar, R. J., Logan, G. D. & Tannock, R. 1999.
Development of inhibitory control across the life span. Developmental Psychology, 35,
205–213.
Williams, B. R., Hultsch, D. F., Strauss, E. H., Hunter, M. a & Tannock, R. 2005.
Inconsistency in reaction time across the life span. Neuropsychology, 19, 88–96.
134 Wilson, A. J., Réale, D., Clements, M. N., Morrissey, M. M., Postma, E., Walling, C. A.,
Kruuk, L. E. B. & Nussey, D. H. 2010. An ecologist’s guide to the animal model. The
Journal of Animal Ecology, 79, 13–26.
Wooller, R., Bradley, J., Skira, I. & Serventy, D. 1990. Reproductive success of short-tailed
shearwaters Puffinus tenuirostris in telation to their age and breeding experience. The
Journal of Animal Ecology, 59, 161–170.
Würsig, B. & Würsig, M. 1977. The photographic determination of group size, composition,
and stability of coastal porpoises (Tursiops truncatus). Science, 198, 755–756.
Ydenberg, R. C. 1998. Behavioral decisions about foraging and predator avoidance. In:
Cognitive Ecology: the Evolutionary Ecology of Information Processing and Decision
Making, (Ed. by R. Dukas), pp. 343–378. Chicago: University of Chicago Press.
Zelazo, P. D., Craik, F. I. M. & Booth, L. 2004. Executive function across the life span. Acta
Psychologica, 115, 167–183.
135 FIGURES AND TABLES
Figure 1. Female bottlenose dolphin (Tursiops sp.) with marine sponge tool (Ircinia sp.) in
Shark Bay, Australia.
136 Observed data LOESS curve
8
. Standard deviation
0
)
t
e
g
d
u
b
6
y
.
t
i
0
v
i
t
c
a
f
o
%
(
4
g
.
n
i
0
g
a
r
o
f
e
g
n
o
2
p
.
S
0
0
.
0
0 10 20 30 40 Age (yrs)
Figure 2. Percent time spent sponging as a function of age. Data are from 19 female bottlenose dolphins observed for at least 3 hours at a given age. Many females were observed over multiple years of their life (mean=3.53 years ± 1.87 SD). Each data point represents a single female at a specific age. The solid line is a LOESS smoothing curve, while the dashed line is an estimate of the square root of the variance function.
137 Predictions Partial effect of age 95% Confidence intervals
0 Maximum of 85.3 min at 22.7 yrs
5
1
)
n
i
m
(
t
u
o
b
l
o
0
o
t
0
1
g
n
i
r
u
d
g
n
i
g
a
r
o
F
0
5
0 10 20 30 Age (yrs)
Figure 3. Partial effect of age on the amount of foraging with a single sponge.. Data are from 22 female spongers observed multiple times at different ages. Mean number of tool observations per female was 14.86 ± 12.33 SD. Each data point represents the predicted amount of time each sponge was used to forage from the full MCMC generalized linear mixed model (Table 1, 327 total tool observations). The solid line represents the partial effect of age after marginalizing across all other factors. The dashed lines denote the 95% confidence intervals, while the triangle represents the predicted peak based the partial effect of age.
138 0 Predictions
5
1 Partial effect of age 95% Confidence intervals Maximum of 53 min at 26.6 yrs
)
n
i
m
(
0
t
0
u
1
o
b
l
o
o
t
g
n
i
r
u
d
g
n
i
l
0
e
5
v
a
r
T
0
0 10 20 30 Age (yrs)
Figure 4. Partial effect of age on the amount of time females traveled with a single sponge. Data are from 22 female spongers observed multiple times at different ages. Mean number of tool observations per female was 14.86 ± 12.33 SD. Each data point represents the predicted amount of time each sponge was traveled with from the full MCMC generalized linear mixed model
(Table 2, 327 total tool observations). The solid line represents the partial effect of age after marginalizing across all other factors. The dashed lines denote the 95% confidence intervals, while the triangle represents the predicted peak based the partial effect of age.
139 Observed Data LOESS curve Standard Deviation
8
.
0
g
n
i
g
a
r
o
F
6
.
e
0
g
n
o
p
S
/
g
n
i
h
4
c
.
r
0
a
e
S
e
g
n
o
p
2
S
.
0
0
.
0
0 10 20 30 40 Age (yrs)
Figure 5. Ratio of time spent locating and detaching sponges, to time spent sponging as a function of age. Data are from 19 female bottlenose dolphins observed for at least 3 hours at a given age. Many females were observed over multiple years of their life (mean=3.53 years ±
1.87 SD). Each data point represents a single female at a specific age. The solid line is a LOESS smoothing curve, while the dashed line is an estimate of the square root of the variance function.
140 0
3 . Predictions
0 Partial Effect of Age 95% Confidence Intervals
5
2
.
0
g
n
i
g
a
r
0
o
2
.
F
0
e
g
n
o
p
S
5
/
1
g
.
n
0
i
h
c
r
a
e
0
S
1
.
e
0
g
n
o
p
S
5
0
.
0
0
0
.
0 0 10 20 30 40 Age (yrs)
Figure 6. Partial effect of age on the ratio of time spent locating and detaching sponges, to time spent sponging both before and after the peak in tool use efficiency separately. Data are from 19 female bottlenose dolphins observed for at least 3 hours at a given age. Many females were observed over multiple years of their life (mean=3.53 years ± 1.87 SD). Each data point represents the predicted ratio from the full MCMC generalized linear mixed models (Table 3).
The vertical line indicates the location of the peak in tool use efficiency at 22.7 years of age, where the model was split. The solid line represents the partial effect of age after marginalizing across all other factors and the dashed lines denote the 95% confidence intervals.
141
Figure 7. Sponge density. Hatched polygons designate two sampling areas where ~78 % of sponge foraging sightings were located. The green and orange polygons represent example 90% home range contours from a-LoCoH estimates for two female spongers. Points represent estimates of basket sponges densities (sponges/m2) from 39 randomly located video belt transects.
142 0
.
1 Predictions Partial Effect of Age 95% Confidence Intervals Maximum of 0.8 at 24 yrs
8
.
0
g
n
i
t
6
a
.
t
c
0
a
L
f
o
y
t
i
l
i
b
4
.
a
0
b
o
r
P
2
.
0
0
.
0
0 10 20 30 40 Age (yrs)
Figure 8. Partial effect of age on probability of lactating. Data are from 27 sponger females with known reproductive status. Many females were observed over multiple years of their life
(mean=15.07 years ± 7.47 SD). Each data point represents the predicted probability a particular female will be lactating based on the full MCMC generalized linear mixed model (Table 4.). The solid line represents the partial effect of age after marginalizing across all other factors. The dashed lines denote the 95% confidence intervals, while the triangle represents the predicted peak based the partial effect of age.
143 Table 1. Primers used in genetic analysis of dolphin fecal sample.
Primer Primer Sequence (5'-3') Reference FishF2 TCGACTAATCATAAAGATATCGGCAC Ward et al. 2005 FishR2 ACTTCAGGGTGACCGAAGAATCAGAA Ward et al. 2005 FishF2 short TGCTATAGTAGGCACAGCAT Helix Molecular Solutions Pty Ltd 2011 FishR2 short CAGAGGGATCAATCAATTACC Helix Molecular Solutions Pty Ltd 2011
144 Table 2. Results from genetic analysis of sponger fecal sample and Parapercis nebulosa. The two sequences below are identical.
Sample Sequence (5'-3') Sponger feces TGCTATAGTAGGCACAGCATTAAGCCTGCTAATCCGCGCTGAACTA AGTCAACCTGGTGCCCTTCTTGGCGACGACCAGATCTACAACGTAA TCGTTACAGCACATGCTTTCGTGATAATCTTTTTTATAGTAATGCCC ATCATGATTGGAGGATTCGGTAATTGATTGATCCCTCTGA Parapercis nebulosa TGCTATAGTAGGCACAGCATTAAGCCTGCTAATCCGCGCTGAACTA AGTCAACCTGGTGCCCTTCTTGGCGACGACCAGATCTACAACGTAA TCGTTACAGCACATGCTTTCGTGATAATCTTTTTTATAGTAATGCCC ATCATGATTGGAGGATTCGGTAATTGATTGATCCCTCTGA
145 Table 3. MCMC glmm for tool bout duration. The model was created using a censored Poisson error structure and the log link function using additive overdispersion. Fixed effects were age, age2 (using orthogonal polynomials to reduce multicollinearity), sponge density, and age-specific percent time sponging, and the response was time spent foraging with a single sponge. Individual identity and sampling day were controlled for as random factors.
Fixed Effect Coefficient Lower 95% CI Upper 95% CI Effective sample Size MCMC p-value Intercept 2.9404 2.4419 3.5206 5123 0.0002 Age 4.7884 1.6047 7.9621 5125 0.0056 Age2 -4.8204 -8.1107 -1.8696 4990 0.0024 Sponge density 0.009 0.0001 0.0184 4990 0.0497 % sponging 1.3717 0.6045 2.2158 4990 0.0004
146 Table 4. MCMC glmm for time traveled with a single sponge tool. The model was created using a censored Poisson error structure and the log link function using additive overdispersion. Fixed effects were age, age2 (using orthogonal polynomials to reduce multicollinearity), sponge density, and age-specific percent time traveling, and the response was time spent traveling with a single sponge. Individual identity and sampling day were controlled for as random factors.
Fixed Effect Coefficient Lower 95% CI Upper 95% CI Effective sample size MCMC p-value Intercept -0.4908 -1.5174 0.5920 4990 0.3447 Age 19.1464 10.3303 27.5917 4990 0.0002 Age2 -11.1812 -18.6655 -4.4979 4990 0.0040 Sponge density 0.0108 -0.0119 0.0345 4990 0.3567 % traveling 7.3643 3.6428 11.1538 4990 0.0002
147 Table 5. MCMC glmm for the ratio of locating and detaching to sponging. The model was created using a binomial error structure and the logit function using additive overdispersion.
Fixed effects were age, age2 (using orthogonal polynomials to reduce multicollinearity), and sponge density, and the ratio of sponge locating and detaching to sponging was the response.
Individual identity was controlled for as a random factor. Separately models were created for before and after the peak of tool use efficiency at age 22.7.
Fixed Effect Coefficient Lower 95% CI Upper 95% CI Effective sample size MCMC p-value Intercept before -0.9227 -1.9850 0.1706 4990 0.0914 Age before -0.0788 -0.1442 -0.0144 4990 0.0257 Sponge density before -0.0093 -0.0224 0.0044 4990 0.1667 Intercept after -2.1288 -7.6249 3.1848 4990 0.4190 Age after -0.0206 -0.1852 0.1347 4990 0.7860 Sponge density after -0.0129 -0.0466 0.0193 4990 0.4210
148 Table 6. MCMC glmm for the probability of lactating. The model was created with a binomial error structure and the logit link function. Fixed effects were age, age2 (using orthogonal polynomials to reduce multicollinearity), and the response was the probability of lactating.
Individual identity was controlled for as a random factor.
Fixed Effect Coefficient Lower 95% CI Upper 95% CI Effective sample size MCMC p-value Age 19.0000 11.5000 26.6000 4990 0.0002 Age2 -28.8200 -37.3900 -21.7100 5356 0.0002
149 Table 7. Cox-proportional hazards model for tool bout duration. Fixed effects were age, age2
(using orthogonal polynomials to reduce multicollinearity), sponge density, and age specific percent time sponging, and the response was time spent foraging with a single sponge. Individual identity and sampling day were controlled for as random factors.
Fixed Effect Coefficient SE Coefficient Z p-value Age -4.4463 1.7680 -2.5100 0.0120 Age2 6.4094 1.7708 3.6200 0.0003 Sponge density -0.0079 0.0053 -1.4900 0.1400 % sponging -1.4826 0.4710 -3.1500 0.0016
150 Table 8. Cox-proportional hazards model for the time traveled with a single sponge. Fixed effects were age, age2 (using orthogonal polynomials to reduce multicollinearity), sponge density, and age specific percent time traveling, and the response was time spent traveling with a single sponge. Individual identity and sampling day were controlled for as random factors.
Fixed Effect Coefficient SE Coefficient Z p-value Age -8.3920 1.9634 -4.2700 < 0.0001 Age2 6.9822 1.6404 4.2600 < 0.0001 Sponge density -0.0058 0.0052 -1.1200 0.2600 % traveling -2.9974 1.0180 -2.9400 0.0032
151 Table 9. Maximum likelihood glmm for the ratio of locating and detaching to sponging. The model was created with a Poisson error structure and the log link function using additive overdispersion and was fit by Laplace approximation. Fixed effects were age, age2 (using orthogonal polynomials to reduce multicollinearity), and sponge density, and the response was the ratio was locating and detaching to sponging. Separate models were created for before and after the peak of tool use efficiency at age 22.7. Individual identity was controlled for as a random factor.
Significance was tested using Wald Z-tests.
Fixed Effect Coefficient SE Coefficient Z p-value Intercept before -0.8932 0.4760 -1.8760 0.0606 Age before -0.0810 0.0285 -2.8460 0.0044 Sponge density before -0.0093 0.0060 -1.5620 0.1183 Intercept after -2.1288 2.3163 -0.9110 0.3620 Age after -0.0201 0.0686 -0.2940 0.7690 Sponge density after -0.0128 0.0139 -0.9260 0.3540
152 Table 10. Maximum likelihood glmm for the probability of lactating. The model was created with a binomial error structure and the logit link function and was fit by Laplace approximation. Fixed effects were age, age2 (using orthogonal polynomials to reduce multicollinearity), and the response was the probability of lactating. Individual identity was controlled for as a random factor. Significance was tested using Wald Z-tests.
Fixed Effect Coefficient SE Coefficient Z p-value Age 15.9570 3.2320 4.9380 < 0.0001 Age2 -25.0540 3.4150 -7.3360 < 0.0001
153 DISSERTATION CONCLUSION
Bottlenose dolphins are extraordinarily plastic in their behavior, which undoubtedly contributes to their success as a species worldwide. This plasticity allows individuals to learn, adjust their behavior, and ultimately meet the demands of their unique social and ecological situation. In areas containing within population habitat heterogeneity, this plasticity may lead to pronounced intrapopulation behavioral variation. Shark Bay and its great diversity of habitats provides a unique system in which to study intrapopulation variation, and because of the long- term data from the Shark Bay Dolphin Research Project, the influence of various factors on this variation can be explicitly examined. In this study I examined how ecology and life history relate to intrapopulation behavioral variation.
In its most simple form, ecology is the study of the dynamic interactions between an organism and its environment. One of the most basic ways an animal can interact with its environment is to simply occupy it, or use it. For bottlenose dolphins, I found that this use should not be described at the population level without first examining and describing inter-individual variation. Such individual level information joins together knowledge from previous studies regarding other aspects of Shark Bay dolphins’ behavioral ecology. For example, individuals that specialize in habitat specific foraging behaviors also tend to show strong, temporally stable, habitat specialization, and most are female. Males, on the other hand, show increased habitat use diversity and larger home ranges, which links our understanding of male foraging strategies and the mating system in Shark Bay.
Beyond the basics, diving deeper (literally) into one foraging specialization, sponge foraging, revealed a possible case of problem solving in the wild, strengthened our understanding
154 of the conditions that favor tool use and innovation, and uncovered the development of expertise in a non-human animal, all of which ultimately improve our understanding of the factors influencing individual foraging specialization. Like many tool users, dolphins only resort to tools when their physical adaptations are not enough, and the characteristics of their targeted prey clarify many aspects of sponging behavior previously inexplicable. These prey are small, but reliable, meaning dolphins must spend considerable time sponge foraging. Add in that sponging proficiency appears to take over 20 years to reach, and it becomes more evident why sponging requires life-long individual specialization, is vertically transmitted, and primarily restricted to females.
While many aspects of the individual differences characteristic of Shark Bay bottlenose dolphins warrant further investigation, this study provided much needed data on individual level habitat use and ranging, and revealed a great deal on the inner workings of the only well- documented case of wild cetacean tool use. But as in any good study should, more questions were raised than answers given, so alas, there is more work to be done!
155
APPENDIX A
Chapter 2 was published in the journal PloS ONE. This journal does not require that authors receive permission prior to publishing the authors’ articles in their own theses
(http://www.plosone.org/static/license.action)
156