A Scientific Foundation for Informed Management Decisions: Quantifying the Abundance, Important Habitat and Cumulative Exposure

A scientific foundation for informed management decisions:

Quantifying the abundance, important habitat and cumulative exposure of the Hawaii Island spinner dolphin (Stenella longirostris)

stock to human activities

This thesis is presented for the degree of

Doctor of Philosophy of Murdoch University

2015

Submitted by

Julian Ashwell Tyne

BSc (Hons.) (Murdoch University, Western Australia)

I declare that this thesis is my own account of my research and contains its main content work which has not previously been submitted for a degree at any tertiary educational institution.

Julian Ashwell Tyne

Statement on the contribution of others

Supervision

Professor Lars Bejder, Professor Neil Loneragan, Professor Ken Pollock, Associate

Professor David Johnston.

Project Funding

This research was made possible by the financial commitment of the National

Oceanic and Atmospheric Administration (NOAA), The Marine Mammal

Commission, Murdoch University and Dolphin Quest.

Stipend

Australian Post-graduate Award (APA)

Contribution to Chapters

Chapter 2 - The use of area-time closures as a tool to manage cetacean-watch tourism: Julian Tyne reviewed the literature and wrote manuscript which was critically reviewed by Professor Neil Loneragan and Professor Lars Bejder.

Chapter 3 - Abundance and survival rates of the Hawaii Island associated spinner dolphin (Stenella longirostris) stock: Julian Tyne collected and analysed the data and wrote the manuscript. Professor Kenneth Pollock advised on the analysis. Professor

Lars Bejder, Professor Kenneth Pollock and Associate Professor David Johnston critically reviewed the manuscript.

Chapter 4 - Informing monitoring programs to detect trends in abundance of cetacean populations: a case study focussing on the Hawaii Island spinner dolphin

(Stenella longirostris) stock. Julian Tyne collected and analysed the data and wrote the manuscript. Professor Kenneth Pollock advised on the analysis. Professor Lars

Bejder, Professor Neil Loneragan, Professor Kenneth Pollock and Associate

Professor David Johnston critically reviewed the manuscript.

Chapter 5 - The importance of spinner dolphin (Stenella longirostris) resting habitat: Implications for management. Julian Tyne collected and analysed the data and wrote the manuscript. Robert Rankin helped with the analysis. Professor Lars

Bejder, Professor Neil Loneragan and Associate Professor David Johnston critically reviewed the manuscript.

Chapter 6 - Cumulative exposure of Hawaiian spinner dolphins (Stenella longirostris) to human interactions: No rest for the wicked: Julian Tyne collected the data and analysed the photo id data and wrote the manuscript. Dr Fredrik

Christiansen developed the activity budget models, Heather Heenehan analysed the acoustic data. Professor Lars Bejder and Professor Neil Loneragan critically reviewed the manuscript.

Abstract

Coastal dolphin populations are exposed to non-consumptive human activities that can pose conservation challenges. Consequently, effective management strategies, using rigorous scientific assessments of exposed populations, are needed to mitigate any potential negative impacts of these activities. To inform management decisions for the conservation of the Hawaii Island spinner dolphin (Stenella longirostris) stock, I: (i) estimated abundance and survival rates; (ii) measured the effectiveness of various sampling scenarios to detect changes in abundance; (iii) identified important spinner dolphin resting habitats; and (iv) measured cumulative exposure to human activities. Between September 2010 and March 2013, boat- based and land-based sampling was undertaken to collect dolphin photo- identification, group behaviour and acoustic data from both inside and outside four important spinner dolphin resting bays on the Kona Coast of Hawaii Island.

Between years, independent survival rate estimates were similar (0.97 ± 0.05 SE), and abundance estimates of 631 (95% CI 524-761) and 668 (95% CI 556-801; CV

=0.09) were very consistent. At this precision, and with 95% power and a monitoring interval of three years, a 5% change in abundance would not be detected for 12 years. I documented that should resting spinner dolphins be displaced from resting bays, they are unlikely to engage in rest behaviour elsewhere. When resting inside bays, dolphins were most likely to rest between

10:00-14:00, and over sandy substrates. Individual spinner dolphins spent between

49.5% and 69.4% of daytime resting (mean = 61.7%). Dolphins were chronically and repeatedly exposed to human activities during daytime hours (> 82% of time), with a median duration of only ten min between interactions. The short interval

between interactions may prevent recovery from disturbance and deprive individuals of rest and change their sleep state from “deep” to “light”. Rest deprivation and the disruption of sleep can lead to impaired cognitive abilities and ultimately effect population viability. These data provide a firm baseline for urgent consideration by managers to evaluate the risks to the spinner dolphins of Hawaii

Island, potential pathways for mitigating human interactions and ways to measure the success of management interventions.

Table of Contents

1 Introduction ...... 1 1.1 Objectives of the thesis ...... 5

2 The use of area-time closures as a tool to manage cetacean-watch tourism ...... 7 2.1 Introduction ...... 7 2.2 Benefits of the cetacean watch industry ...... 10 2.3 Costs of the cetacean watch industry ...... 12 2.4 Strategies for managing the cetacean watching industry ...... 15 2.4.1 Unmanaged and unregulated tourism ...... 18 2.4.2 Guidelines / Codes of conduct ...... 20 2.4.3 General legislative framework ...... 22 2.4.4 Permitted / licensed legislation framework ...... 25 2.5 Important considerations for implementing time-area management strategies ...... 27 2.5.1 Understanding the cetacean population targeted by the cetacean-watch industry ...... 35 2.5.2 Socio - economic considerations ...... 40 2.5.3 Management considerations ...... 42 2.6 Summary ...... 43

3 Abundance and survival rates of the Hawaii Island associated spinner dolphin (Stenella longirostris) stock ...... 45 3.1 Abstract ...... 45 3.2 Introduction ...... 47 3.3 Materials and methods ...... 52 3.3.1 Fieldwork ...... 52 3.3.2 Sampling design ...... 54 3.3.3 Photographic-identification ...... 54 3.3.4 Grading and sorting of photo-identification images ...... 55 3.3.5 Analyses ...... 56 3.3.6 Estimating abundance and demographic parameters ...... 57 3.3.7 Estimation of mark rate and total stock size ...... 59 3.4 Results ...... 61 3.4.1 Effort and summary statistics ...... 61 3.4.2 Mark rate of the stock ...... 63 3.4.3 Apparent survival and total stock abundance ...... 64 3.4.4 Discussion...... 66 3.4.5 Management implications ...... 70

4 Informing monitoring programs to detect trends in abundance of cetacean populations: a case study focussing on the Hawaii Island spinner dolphin (Stenella longirostris) stock...... 72 4.1 Abstract ...... 72

4.2 Introduction ...... 73 4.3 Materials and methods ...... 78 4.3.1 Fieldwork ...... 78 4.3.2 Sampling design ...... 80 4.3.3 Sampling effort ...... 80 4.4 Analyses ...... 81 4.4.1 Capture-recapture ...... 81 4.4.2 Detecting change in abundance ...... 83 4.5 Results ...... 84 4.5.1 Effort and summary statistics ...... 84 4.5.2 Estimates of apparent survival rate and stock abundance...... 85 4.5.3 Detecting change in abundance ...... 86 4.6 Discussion...... 91 4.6.1 Estimates of survival rates and abundance ...... 91 4.6.2 Monitoring changes in dolphin abundance over time ...... 92 4.6.3 Applications for monitoring ...... 93

5 The importance of spinner dolphin (Stenella longirostris) resting habitat: Implications for management ...... 96 5.1 Abstract ...... 96 5.2 Introduction ...... 97 5.3 Materials and Methods ...... 100 5.3.1 Group focal follows ...... 102 5.3.2 Land-based group focal follows ...... 103 5.3.3 Boat-based group focal follows ...... 104 5.3.4 Bathymetric and benthic data ...... 104 5.3.5 Modelling approach ...... 105 5.3.6 Base-learners ...... 106 5.4 Results ...... 108 5.4.1 Effort and sample sizes ...... 108 5.4.2 Boosted GAMs ...... 111 5.5 Discussion...... 117 5.5.1 Management of marine mammals ...... 119 5.5.2 Management implications ...... 120

6 Cumulative exposure of Hawaiian spinner dolphins (Stenella longirostris) to human interactions: No rest for the weary...... 124 6.1 Abstract ...... 124 6.2 Introduction ...... 125 6.3 Materials and methods ...... 129 6.3.1 Fieldwork ...... 129 6.3.2 Systematic photo-identification sampling ...... 132 6.3.3 Group focal follows ...... 132 6.3.4 Passive acoustic recordings ...... 133 6.4 Data analyses ...... 134 6.4.1 Cumulative activity budgets ...... 134 6.4.2 Calculation of dolphin activity budgets for each bay ...... 137

6.4.3 Factors affecting dolphin activity states ...... 137 6.4.4 Duration of interactions between dolphins and humans ...... 139 6.4.5 Passive acoustic recordings ...... 139 6.5 Results ...... 140 6.5.1 Summary of photo-identification and behavioural sampling efforts ...... 140 6.5.2 Bay use by individual dolphins ...... 142 6.5.3 Passive acoustic monitoring efforts ...... 143 6.5.4 Dolphin activity budgets inside and outside resting bays ...... 143 6.5.5 Cumulative activity budget for the sampled population ...... 144 6.5.6 Factors affecting dolphin activity ...... 145 6.5.7 Duration of interactions ...... 148 6.6 Discussion ...... 149 6.6.1 Management implications ...... 155

7 Conclusions ...... 158 7.1 Concluding remarks ...... 162

Appendix I: Peer-reviewed publications during candidature ...... 164

Appendix II: Model selection tables ...... 166

References ...... 167

List of tables

Table 2-2 :Protected area categories, descriptions and primary objectives as determined by the International Union for the Conservation of Nature (IUCN) (Dudley, 2008) ...... 33

Table 2-3 : Protected areas with cetacean habitat across 18 marine regions, adapted from Hoyt (2011)...... 34

Table 3-2 Highly distinctive (D1) population abundance estimates calculated from open and closed mark recapture models...... 65

Table 3-3 Mark rates and abundance estimates from this study and previous studies...... 65

Table 4-1 Apparent survival, abundance, over-dispersion and coefficient of variation calculated for the different sampling scenarios and capture-recapture models. Scenario 1 = 12 days of sampling covering four bays (two days in Makako Bay, four days in Kealakekua Bay, two days in Honaunau Bay and four days in Kauhako Bay), Scenario 2 = six days randomly subsampled from the 12 days covering four bays (one day in Makako Bay, two days in Kealakekua Bay, one day in Honaunau Bay and two days in Kauhako Bay) and Scenario 3 = six days covering two bays (two days in Makako Bay and four days in Kealakekua Bay). SE = standard error, CI = 95% confidence interval, ĉ = over-dispersion. 1estimates from Chapter 3; Tyne et al. (2014) ...... 86

Table 5-1 Definitions of spinner dolphin group activities, adapted from Norris et al (1994)...... 102

Table 5-4 Proportion of substrate types available to spinner dolphins inside and outside of bays within the study area...... 109

Table 5-5 The variables selected during the boosted Generalised Additive Modelling process to examine the influence of spatial position, previous behaviour, time-of- day, substrate and inside or outside of bays on the resting state of spinner dolphins. Optimal m is the point at which the model is stopped during the model fitting process to avoid over-fitting. Stability selection shows the probability of variable selection at optimal m. Maximum iterations = 1,000 bootstrap iterations with a 50- fold cross validation...... 113

Table 6-1 Definitions of spinner dolphin group activities, adapted from Norris et al (1994)...... 133

Table 6-2 Number of days, hours and mean length of photo-identification surveys conducted within the four resting bays between September 2010 and August 2012 ...... 140

Table 6-3 Number and duration of dolphin focal follows collected from land-based and boat-based platforms inside bays and outside of resting bays along the Kona Coast, Hawaii Island...... 141

Table 6-5 Studies that have quantified exposure rates of dolphins to human activities and whether authors noted or inferred an impact. MV = Motorised Vessels, K = Kayaks, SUP = Stand-Up Paddleboard, S = Swimmers ...... 151

List of figures

Figure 2.1: Tour boats and swimmers interact with Hawaiian spinner dolphins in their resting bays off the Kona Coast of the Island of Hawaii. Image taken under permit number GA LOC 15409...... 25

Figure 3.1 Map of the study area illustrating the locations of the four spinner dolphin resting bays, Kauhako Bay, Honaunau Bay, Kealakekua Bay and Makako Bay, along the Kona Coast of Hawaii Island in relation to the other island regions in the Main Hawaiian Islands (inset)...... 53

Figure 3.2 Frequency of individual spinner dolphin sightings from September 2010 to August 2011...... 62

Figure 3.3 Cumulative discovery curve of highly distinctive (D1) and distinctive (D2) Hawaiian spinner dolphins during 132 photographic identification surveys in the study area (all four bays combined) from September 2010 to August 2011...... 63

Figure 3.4 The combination of bays in which individual spinner dolphins have been sighted from September 2010 to August 2011. A = Kauhako Bay, B = Honaunau Bay, C = Kealakekua Bay and D = Makako Bay...... 63

Figure 4.1 Map of the study area illustrating the four spinner dolphin resting bays, Makako Bay, Kealakekua Bay, Honaunau Bay and Kauhako Bay, along the Kona Coast of Hawaii Island...... 79

Figure 4.2 Cumulative discovery curve of highly distinctive (D1) spinner dolphins during 276 photographic identification sampling days from September 2010 to August 2012. Short vertical lines indicate when 90% and 95% of the highly distinctive individuals had been identified. Long vertical dashed line indicates 12 months of sampling...... 85

Figure 4.3 Number of annual abundance estimates required to detect various rates of change in stock size at varying levels of precision (coefficient of variation, CV) from three sampling scenarios. Scenario 1 = 12 days of sampling covering four bays (two days in Makako Bay, four days in Kealakekua Bay, two days in Honaunau Bay and four days in Kauhako Bay), Scenario 2 = six days randomly subsampled from the 12 days covering four bays (one day in Makako Bay, two days in Kealakekua Bay, one day in Honaunau Bay and two days in Kauhako Bay) and Scenario 3 = six days covering two bays (two days in Makako Bay and four days in Kealakekua Bay). Type I error (α) probabilities were set at 0.05 and Type II error (β) probabilities were set at power = 1 – β = 0.95 (dark) and 1 – β = 0.80 (grey)...... 87

and Type II error (β) probabilities were set at power = 1 – β = 0.80 (grey) and power = 1 – β = 0.95 (dark)...... 88

Figure 5.1 The location of the spinner dolphin study area on the Kona Coast showing the four sheltered bays: Kauhako Bay, Honaunau Bay, Kealakekua Bay and Makako Bay, Hawaii Island and the behavioural observations of spinner dolphin groups (black circles) recorded during boat-based (n=28) and land-based (n=47) group focal follows. Each black circle (n=2,856) corresponds to the location where each ten minute scan sample was obtained...... 101

Figure 5.2 The proportion of time spinner dolphins were observed resting, socialising and traveling over available known substrate (aggregate reef, rock/boulder and sand) inside of sheltered bays along the Kona Coast, Hawaii Island. Error bars indicate 95% confidence intervals...... 110

Figure 5.3 The proportion of time spinner dolphins were observed resting, socialising and traveling over available known substrate (aggregate reef, rock/boulder and sand) outside of sheltered bays along the Kona Coast, Hawaii Island. Error bars indicate 95% confidence intervals...... 111

Figure 5.4 Marginal function estimate showing the probability of spinner dolphins resting inside and outside of the four sheltered bays (Kauhako Bay, Honaunau Bay, Kealakekua Bay and Makako Bay). Error bars indicate 95% confidence intervals. . 114

Figure 5.5 Time of day (TOD) base learner marginal function estimates showing the predicted probability of spinner dolphins resting (± 95% confidence intervals) during the morning (6am-10am), mid-morning (10am-2pm) and afternoon (2pm-6pm) for Kauhako Bay and Kealakekua Bay. Resting behaviour based on land-based observations...... 115

Figure 5.6 Predicted percentages for resting spinner dolphins modelled from Boosted GAMs in (A) Kealakekua Bay (n=1,526); and (A) Kauhako Bay (n=200). Grid cells are 50 m2 based on the resolution of available bathymetric and habitat maps...... 116

Figure 6.1 Map of the study area illustrating the four spinner dolphin resting bays, Makako Bay, Kealakekua Bay, Honaunau Bay and Kauhako Bay, along the Kona Coast of Hawaii Island...... 131

Figure 6.2 Integration of three discreet datasets to model the cumulative activity budget of spinner dolphins inside and outside of resting bays along the Kona Coast of Hawaii Island. A) Individual spinner dolphin presence/absence in resting bays from systematic photo id sampling, B) Dolphin group presence/absence in resting bays from 24 hours per day of acoustic monitoring in resting bays and C) Dolphin group behavioural time series from group focal follows inside and outside of resting bays...... 136

Figure 6.3 Sighting frequencies of 235 photographically-identified spinner dolphins in Makako Bay, Kealakekua Bay, Honaunau Bay and Kauhako Bay between September 2010 and August 2012...... 141

Figure 6.4 Proportion of the 235 identified spinner dolphins were documented in each of the four resting bays. Data from Kealakekua Bay and Kauhako Bay were standardised and error bars are shown...... 142

Figure 6.5 Proportion of time spent by spinner dolphins in different activity states inside and outside of four resting bays along the Kona Coast, Hawaii. Error bars represent 95% highest posterior density intervals, analogous to 95% confidence intervals...... 144

Figure 6.6 Density distribution of simulated dolphin cumulative activity budget, showing the relative proportion of each activity state (see legend), throughout the study period (Jan 8th 2011 – Aug 30th 2012) for the sample population...... 145

Figure 6.7 The probability of spinner dolphins (A) resting, (B) socialising and (C) travelling as a function of time-of-day, estimated from the 428 hours of behavioural focal follow observations...... 147

Figure 6.8 Frequency histograms of bout durations (continuous time spent in impact or control situations) of control (left column) and impact (right column) situations for dolphins, recorded from boat (top row) and land (bottom row)...... 149

Acknowledgements

What a ride! This thesis would not have been possible without the guidance, assistance and input of many people.

I am greatly indebted to my supervisors Professor Lars Bejder, Professor Neil Loneragan, Professor Kenneth Pollock and Associate Professor Dave Johnston for their guidance, patience, perseverance and friendship during my PhD candidature.

Professor Lars Bejder your support, encouragement and friendship during this journey has motivated me along the way and helped me to get the ‘story line’ right for my manuscripts. The games of dice and the gin & tonics during your visits to the field were also an extremely motivating factor. I thank you for giving me this incredible opportunity.

Professor Neil Loneragan I thank you for the whiteboard sessions that helped iron out the creases in developing the manuscripts and chapters for my thesis. Face to face is always the best way to answer questions and formulate ideas. I hope The West Coast Eagles have a good season this year.

Professor Kenneth Pollock it was a privilege to work with you, your guidance on the population models and enthusiasm was truly inspiring. I am, however, hoping England win The Ashes!

Associate Professor Dave Johnston it has been a pleasure working with you and your knowledge of the spinner dolphin plight in Hawaii has been instrumental during my PhD.

Thanks to Rob Rankin for help with the boosted GAMs and Fredrik Christiansen for help with the cumulative activity budget models, and of course the figures.

I would like to thank Damien Kenison, Jimmy Medeiros, the communities of Kealakekua, Ho’okena, Honaunau and Kailua Kona, Lisa van Atta, Lance Smith, Jayne LeFors, Jean Higgins and Laura McCue for dialogue and support for my project. Justin Vizbicke, Susan Rickards, Chris Gabriele, Suzanne Yin and Doug Perrine for help and equipment loan in the field, and Keith Unger and McCandless Land and Cattle Co for the use of their land overlooking Kauhako Bay. I would also like to thank the West Hawaii Marine Mammal Response Network for including us in their monthly meetings.

A BIG MAHALO NUI LOA to the many research assistants who made the fieldwork so much fun during the long days on the water and at the theodolite stations of Kealakekua and Kauhako Bay. The ‘A’ Team, I am indebted to your dedication in collecting and processing data: Kim New and John Symons for managing a field season each while I was back in Australia, Alexandrea Abe, Destiny Archambault, Audrey Archer, Isabel Baker, Bret Banka, Miriam Battye, Rosie Blackburn, Tasha Boerst, Laura Bray, Amy Brossard, Laura Ceyrac, Sharon Chan, Marie Chapla-Hill,

Manon Chautard, Laura Cunningham, Alicia Day, Bianca Dekker, Sarah Deventer, Martin Durham, Sophia Gelibter, Bob Gladden, Stacia Goecke, Alison Greggor, Daniella Hanf, Dan Hazel, Heather Heenehan, Merra Howe, Lonneke Ijsseldijk, Rebecca Ingram, Tory Johnson, Shannon Jones, Carissa King, Crystal Kulcsar, Kelsey Lane, Jacqueline Loevenich, Brigid McKenna, Krista Nicholson, Katrina Nikolich, Megan O’Toole, Nicole Penkowski, Stephanie Petrus, Rita Reis, Emily Richardson, Robyn Smith, Kate Sprogis, Ashley Steinkraus, Mariel ten-Doeschate, Megan White, Kaja Wierucka, Virginie Wyss.

An especially BIG MAHALO NUI NOA to Bob and Donna Gladden for everything you did for me and my assistants during our time in Kona: for the loan of your boat MALOLO, welcoming us into your home, and keeping us all very well fed with the many meals you cooked for us, for taking us out for dive and snorkel trips during the few days off from collecting data and for all the maintenance help you provided Bob, it was invaluable.

To the MUCRU coffee gang Simon Allen, Delphine Chabanne, Krista Nicholson, Alex Brown, Kate Sprogis, Rob Rankin and more recently Fredrik Christiansen. It has been a blast. Our regular, sometimes ridiculous, sometimes ridiculously long, morning coffees have been a great way to start the day.

Love to my Mum and Dad, Maureen and John and my Sister Vanessa who have encouraged me from the other side of the world.

Finally, but by no means least, to Anita, when I said I wanted to change the direction in my life, you didn’t flinch. You encouraged me to take those first nerve wracking and tentative steps back to school. See what happened! I will be forever grateful for your love and support while following my dream. I cannot put into words how much you mean to me, but I do know that I love you with all my heart and cherish every moment with you.

Publications

The publications listed below form the basis for the parts of this thesis:

1. Tyne, J.A., Loneragan, N.R. and Bejder, L. (2014). The use of area-time closures as a tool to manage cetacean-watch tourism. In Whale-watching, sustainable tourism and ecological management (eds J. Higham, L.Bejder and R. Williams), pp. 242-260. Cambridge University Press, Cambridge. (Chapter 2)

2. Tyne, J.A., Pollock, K.H., Johnston, D.W. and Bejder, L. (2014). Abundance and Survival Rates of the Hawaii Island Associated Spinner Dolphin (Stenella longirostris) Stock. PLoS ONE 9, e86132. (Chapter 3)

3. Tyne, J.A., Johnston, D.W., Loneragan, N.L., Pollock, K.H. and Bejder, L. (in prep). Informing monitoring programs to detect trends in abundance of cetacean populations: a case study focussing on the Hawaii Island spinner dolphin (Stenella longirostris) stock. (Chapter 4)

4. Tyne, J.A., Johnston, D.W., Rankin, R., Loneragan, N.R. and Bejder, L. (2015). The importance of spinner dolphin (Stenella longirostris) resting habitat: Implications for management. The Journal of Applied Ecology (Chapter 5)

5. Tyne, J.A., Christiansen, F., Heenehan, H., Johnston, D.W. and Bejder, L (in prep). Cumulative exposure of Hawaiian spinner dolphins (Stenella longirostris) to human activities: No rest for the wicked. (Chapter 6).

1 Introduction

In the early 1980s, cetacean-based tourism experienced unprecedented growth worldwide (Hoyt, 2001). By the late 1990s, it was a US$1 billion industry, and a decade later the industry engaged more than 13 million participants in 119 countries, generating over US$2 billion (O'Connor et al., 2009). Despite the undoubted economic benefits that cetacean-based tourism brings, the sustainability of the industry has been questioned (e.g. Orams, 1999, Corkeron,

2004, Higham et al., 2014). These concerns were recognised by the International

Whaling Commission, stating that “there is new compelling evidence that the fitness of individual odontocetes repeatedly exposed to whale watching vessel traffic can be compromised and that this can lead to population level effects” (IWC, 2006).

Specific concerns over the potential impacts of repeated exposure to tour vessels relate to: changes to behavioural budgets (Lusseau, 2003a), energetic deficits

(Williams et al., 2006, Christiansen et al., 2014), leading to reduced vigilance for predators (Johnston, 2014), alteration of social interactions with conspecifics

(Lusseau and Newman, 2004), inadequate recovery from day-time disturbances

(Lusseau, 2004), and displacement from preferred habitat (Bejder et al., 2009) with an increase in predation risk (Heithaus and Dill, 2002). These types of disturbance to the behaviour and physiology of cetaceans may affect their vital rates or reproduction (National Research Council, 2005, Christiansen and Lusseau, 2015) and lead to long-term consequences for the viability and fitness of individuals and populations (e.g. Bejder et al., 2006b, Lusseau et al., 2006).

As a consequence of the cetacean-based tourism boom and the increasing evidence of its negative impacts, there is a growing need to develop and implement effective management strategies that ensure tourism activities do not endanger the viability of cetacean populations (Corkeron, 2006, Higham et al., 2014). However, measuring the long-term effects of cetacean-based tourism on long-lived animals with low fecundity in order to provide a sound basis for improved management presents a significant challenge. Long-term research and monitoring are required in order to demonstrate causal relationships between exposure to human activities, behavioural change and biological significant impacts (Bejder and Samuels, 2003).

Legislation to protect marine mammals from human activities was first introduced in the United States under the Marine Mammal Protection Act 1972 (MMPA, 1972).

However, the MMPA’s fundamental objective is to maintain marine mammal stocks at their optimum sustainable populations, focusing primarily on preventing unsustainable takes, harassment and disturbance associated with the commercial fishing industry (Roman et al., 2013). Despite marine mammals being afforded a high level of protection from anthropogenic activities under this Act, it remains unclear how the MMPA applies to cetacean-based tourism. Consequently, there is a need to introduce or amend legislation to consider policies that effectively protect cetaceans from takes, harassment and disturbance associated with tourism.

Wildlife management agencies face significant challenges in implementing and evaluating suitable regimes for cetacean-based tourism, as the growth of the industry far exceeded our ability to collect appropriate scientific information upon

2 which to base management decisions (Higham et al., 2008). The management of cetacean-based tourism ranges broadly from voluntary codes of conduct and guidelines, through to strict government legislation (Garrod and Fennell, 2004,

Whitt and Read, 2006, Allen et al., 2007; Chapter 2). Effective, long-term management strategies are those that monitor tourism operations, establish thresholds of human-cetacean interactions and respond adaptively to the impacts of these operations and natural phenomena (Higham et al., 2008). Spatial management, including the use of protected areas or closures to commercial operations, can be an effective approach in protecting both terrestrial and marine ecosystems (Pauly et al., 2002, Hoyt, 2011, Edgar et al., 2014). Protected areas have been implemented as measures to reduce the risks of overexploitation in marine ecosystems, especially when scientific knowledge about the ecosystem is lacking

(Hoyt, 2011). Areas have been used and proposed to protect cetacean activities that are especially vulnerable to human activities, e.g. feeding in southern resident killer whales; and beach rubbing in northern resident killer whales (Williams et al.,

2009). For protected areas with limited scientific knowledge, the spatial area for protection might be increased as a precaution to account for the uncertainty of available information. Spatial management has been used to conserve biodiversity, and delineate areas for specific use to mitigate anthropogenic threats, enhance productivity and provide public focus for marine conservation (Chapter 2). There is also emerging evidence that protected areas can be effective for cetaceans

(Gormley et al., 2012). In addition to spatial management, temporal closures can be introduced to limit human access to cetaceans during specific times that are critical

3 to targeted animals/populations (Notarbartolo di Sciara et al., 2008, Constantine et al., 2004).

Many cetacean populations exhibit specialized behavioural routines that are temporally partitioned. Some baleen whales migrate vast distances on an annual basis for the purposes of feeding at high latitudes and breeding at low latitudes

(e.g. Clapham, 2000). Hawaiian spinner dolphins (Stenella longirostris), on the other hand, have evolved a daily behavioural pattern involving foraging cooperatively in deep water during the night and gathering in sheltered bays to socialize and rest during the day (Norris et al., 1994, Benoit-Bird and Au, 2009). They form island- associated stocks with unique behavioural features (Norris et al., 1994, Benoit-Bird and Au, 2009). This specialized behavioural ecology may render Hawaiian spinner dolphins particularly vulnerable to anthropogenic disturbance. The bays in which they rest are also the locations of on-water human activities, exposing spinner dolphins to a range of human operations: recreational and commercial boating, swimmers and kayakers (Courbis and Timmel, 2009, Delfour, 2007). Furthermore, genetic analyses have revealed that Hawaiian spinner dolphins are distinct from populations found elsewhere (Andrews, 2009), and that sub-populations within the

Hawaiian archipelago are genetically distinct from each other (Andrews et al.,

2010).

In recognition of the sub-population structuring of Hawaiian spinner dolphins, the

National Oceanic and Atmospheric Administration’s (NOAA) National Marine

Fisheries Service (NMFS) divided them into five different island/island-group

4 management units under two broad geographical regions: three units in the Main

Hawaiian Islands (Hawaii Island, Oahu/4-Islands (Maui, Moloka’i, Lana’i and

Kaho’olawe) area, Kauai/Niihau); and two units in the North-Western Hawaiian

Islands (Pearl & Hermes Reef and Kure/Midway Atolls). The MMPA mandates that each management unit be assessed regularly by NMFS. In 2005, NOAA published an

Advanced Notice of Proposed Rulemaking about the concerns surrounding human- spinner dolphin interactions and to solicit feedback on potential options for future regulations under the MMPA (NOAA, 2005). In addition, NOAA initiated a research program to investigate the scale of the effects and possible ways to mitigate potential negative impacts of human interactions. This thesis forms part of an initiative to assess the abundance and survival rates of the spinner dolphin stock and to help inform an effective management strategy for spinner dolphin-based tourism.

1.1 Objectives of the thesis An assessment of the baseline characteristics of the dolphin stock is integral to the development of an effective management strategy for ensuring its long-term viability, particularly in the face of intensive anthropogenic activities, including nature-based tourism. This assessment should include an estimation of the dolphin stock abundance and survival rates in order to provide a benchmark from which alternative management strategies can be implemented and evaluated. The paucity of knowledge on the abundance, demographic parameters, habitat use and cumulative exposure rates of the Hawaii Island spinner dolphin stock to human activities was used to frame the specific objectives of my thesis, which are to:

1. Review the use of time-area closures as a tool to manage cetacean-watch

tourism (Chapter 2);

2. Develop a rigorous and systematic photo-identification sampling regime used in

combination with capture-recapture models to estimate the survival,

abundance and movement rates of the Hawaii Island spinner dolphin stock

(Chapter 3);

3. Estimate the power of various sampling regimes to detect trends in spinner

dolphin abundance, ultimately to inform management about options for

developing an effective monitoring program (Chapter 4);

4. Identify habitat features that contribute to the occurrence of resting behaviour

in spinner dolphins by linking behaviour and habitat features using gradient

boosted Generalised Additive Models (Chapter 5);

5. Determine the cumulative exposure of spinner dolphins to human activities by

integrating a suite of sampling regimes, including boat-based photo-

identification, boat- and land-based focal group follows and passive acoustic

monitoring (Chapter 6); and

6. Develop recommendations for mitigating possible impacts of human-spinner

dolphin interactions and for monitoring the efficacy of mitigation strategies

proposed by NOAA (Chapter 7).

2 The use of area-time closures as a tool to manage cetacean-watch tourism

2.1 Introduction The world’s oceans have been exploited for their resources for generations. In some cases, this has led to the removal of top predators from ecosystems resulting in a cascading effect through trophic levels altering the ecosystem and restructuring the food web, referred to as fishing down the food web (Pauly et al., 2002, Myers and Worm, 2003, Ferretti et al., 2010). Cetaceans (whales, dolphins and porpoises) have been targeted for their oils and meats, and with advances in technology and methods of hunting, some populations were driven perilously close to extinction.

Fortunately, attitudes towards the harvesting of cetaceans have changed (Bearzi et al., 2010), and rather than harvest cetaceans for their products, it is now more desirable to observe them in their natural environment.

Today, cetaceans are the embodiment of marine conservation efforts. The USA was the first country to introduce legislation to protect marine mammals through the

Marine Mammal Protection Act 1972 (MMPA). The MMPA was designed to minimise the capture or ‘take’, harassment and disturbance of marine mammals, primarily from fishing operations as by-catch and from cetacean hunting. The

MMPA defines the term ‘take’ as “hunting, killing, capture and harassment of a marine mammal or the attempt thereof”. Since the declaration of the MMPA, other countries have adopted their own legislation similar to the MMPA, e.g., The Marine

Mammals Protection Act 1978 in New Zealand and the Environment Protection and

Biodiversity Conservation Act 1999 in Australia.

Although the protection of cetaceans is supported enthusiastically in many countries around the world, only 1% of the worlds oceans are protected (Hoyt,

2005), and cetacean populations and their habitats are still vulnerable to a multitude of anthropogenic impacts. These threats can be categorised into two broad groups: direct threats (i.e., those that are immediate and readily observable) and cumulative impacts (i.e., those that are not readily observable and likely to cause effects through repeated exposure). Direct threats are those that cause the death of individuals immediately, such as whaling (Gales et al., 2005), ship strikes

(Panigada et al., 2006) and bycatch (Mangel et al., 2010). Although the deaths of individual cetaceans are readily detected, quantifying the effects of direct impacts on cetacean populations and their future viability is difficult as it requires information on the population size and the connectivity of populations. For example, the question of whether the annual loss of 20 dolphins through bycatch from a trawl fishery is likely to cause a significant decline in the population is not easily answered without information on population size and its connectivity to other populations. Cumulative effects, or indirect impacts, include sources of disturbance that are likely to affect the behaviour and/or physiology of cetaceans and as a consequence, are more difficult to identify and quantify. Indirect effects include noise pollution (Nowacek et al., 2007, Tyack, 2008), chemical pollution

(Reijnders et al., 2009), tourism (Lusseau and Higham, 2004, Lusseau et al., 2006,

Bejder et al., 2006b), coastal development (Chilvers et al., 2005, Jefferson et al.,

2009), prey exploitation (Bearzi et al., 2006), oil and gas exploration (Harwood and

Wilson, 2001), shipping (Clark et al., 2009), aquaculture (Watson-Capps and Mann,

2005) and climate change (Alter et al., 2010).

Ironically, cetacean-watch tourism operations, most of which are promoted as beneficial, can cause significant impacts to cetaceans if not managed appropriately.

Specifically, dolphin-watching can cause biologically significant impacts on exposed communities by causing habitat displacement and reducing the reproductive success of individuals exposed to this form of “eco-tourism” (Lusseau, 2005, Bejder et al., 2006b, Bejder et al., 2006a). The Whale Watching sub-committee of the

International Whaling Commission has noted that cetacean populations targeted by tourism operations can be divided into four broad categories based: 1) resident populations where breeding, nursing, and feeding occur in the same area; 2) cetaceans on their breeding grounds; 3) cetaceans on their feeding grounds; and 4) cetaceans on their migratory corridors (IWC, 2006). It is important to note that each category is likely to require different levels and types of protection, e.g. potentially, it is more important to protect cetaceans on their breeding grounds than on their migratory corridor. In addition, cetacean-watch tourism around the world exists within varying social, cultural, economic, political and environments (Higham et al.,

2008). These aspects require careful consideration when developing an effective management framework, as each situation is unique, i.e. there is no “one size fits all” or “magic bullet” solution. As a consequence, management frameworks for cetacean-watch operations must be designed based on the overall context in which the cetacean-watch operation takes place in relation to the target cetacean population. This raises the question of which approach(es) is/are the most appropriate to protect cetacean populations against impact(s) from tourism operations?

In this chapter I discuss area-time closures as a management option to mitigate threats to cetaceans from commercial cetacean-watch tourism. I begin by evaluating the benefits and potential threats of the cetacean-watch industry; and discuss the variety of legislation available and its effectiveness. Subsequently, I discuss the development of area-time closures as part of management frameworks to help mitigate threats to cetacean populations, and lastly I discuss important issues that need to be considered when developing area-time closures.

2.2 Benefits of the cetacean watch industry Cetacean-watch tourism has the potential to contribute to economic growth, education, conservation and the collection of scientific data. Cetacean-watch tourism is a rapidly-growing industry. In 1998, it was a US $1 billion industry (Hoyt,

2001), which by 2009, it had doubled to approximately 13 million tourists paying to observe cetaceans in their natural environment, generating an annual turnover of more than US $2 billion and estimated to employ 13,000 people in coastal communities around the world (Cisneros-Montemayor et al., 2010, O'Connor et al.,

2009). If cetacean-watch operations start in countries where currently they are not operating, employment is predicted to increase to more than 19,000 people, and that the total annual revenue will increase to US $2.5 billion (Cisneros-Montemayor et al., 2010). As a consequence of this rapid growth, coastal communities where cetacean-watch tourism is practiced are highly dependent on the income generated by local cetacean-watch operations. For example, in Kaikoura, New Zealand, the local community resurrected their ailing economy by developing commercial whale-

10 watch operations (Hoyt, 2007). Before whale-watching, approximately 3,400 tourists visited Kaikoura annually, which increased to 80,000, seven years after the development of commercial whale-watching in 1986 and after 12 years, had increased more than ten-fold, to an estimated 873,000 visitors annually (Hoyt,

2007). Cetacean-watch operations provide an ideal platform from which to educate and raise awareness of the biology and environment of the target cetacean populations, the threats to their population and efforts to conserve the population.

Many cetacean-watch operations include an educational and interpretive component during their tour (Lück, 2003). Provided with accurate scientific information, trained tour guides can play a critical role in informing the public about local cetacean populations, and raising their awareness of the environmental and the conservation measures being implemented to conserve the population.

Properly developed education programs can be effective in managing tourist interactions with free-ranging animals in their natural environment (Orams, 1997).

Furthermore, some argue that cetacean-watching experiences can lead to behavioural changes in the tourists by encouraging a more environmentally aware behaviour (Orams, 1997, Ballantyne et al., 2010).

Commercial cetacean-watch operations also provide a platform for scientific research (Bejder and Samuels, 2003). Using a tour-vessel for research may, however, restrict the appropriate behavioural sampling methods and the type of abundance, distribution and behavioural data that can be reliably collected and interpreted reliably. However, vessels used in the cetacean-watch industry provide frequent and relatively inexpensive access to cetaceans, and the ability to obtain

11 many observations of cetacean-tourist interactions. For example, commercial tour vessels were used to good effect to study commercial swim-with-dolphin operations (Constantine, 2001), and in controlled approach experiments, combined with observations made from a land-based theodolite, to record the behaviours of the target cetaceans when approached by the cetacean-watch vessel (Williams and

Ashe, 2007). In addition, the presence of researchers on a cetacean-watch vessel allows accurate and up-to-date knowledge to be communicated to tourists, and social science researchers can also obtain data on the cetacean-watching experience and tourist expectations.

2.3 Costs of the cetacean watch industry Cetacean-watch tourism, by its very nature, repeatedly seeks out prolonged close- up encounters with specific communities of wild cetaceans. The cumulative impacts of repeated encounters have the potential to cause significant biological effects on the population. Possible effects must be evaluated with any changes in the cetacean populations due to natural phenomena (Lusseau and Higham, 2004), and as a consequence, evaluating whether cetacean-watch operations are having a biologically significant effect on target cetacean populations is challenging.

The National Research Council (2005), developed the Potential Consequences of

Acoustic Disturbance (PCAD) model for cetaceans. This model incorporates five groups of variables with observable features: sound, behaviour change, life functions, vital rates, and population effect. The conceptual model identifies sound characteristics that may cause a disturbance with a resulting behavioural change

(e.g. the sound may cause a change in dive behaviour or movement). It then links how behavioural changes can cause alterations to life functions (e.g. feeding, breeding) which can then cause changes to vital rates (e.g. survival, reproduction) and finally, how changes to vital rates can translate into population effects (e.g. population growth rate). Although this model was developed specifically for acoustic disturbance, the framework can be incorporated in the evaluation of any potential impact on cetacean populations. For example, repeated disruption to resting spinner dolphins by cetacean-watch operations (i.e. tourism replacing the disturbance by sound) could result in a behavioural change from resting to socialising, which reduces the resting behavioural budget (life function effected), which in turn reduces the energy for reproduction (vital rate), which could reduce the population growth rate (population effects). The manifestation of repeated behavioural disruptions, mediated through cetacean-watch vessel disturbance, may cause long term biologically significant effects on cetacean populations.

The behavioural responses of cetaceans to vessels vary greatly, ranging from approaching and bow-riding to complete avoidance. A number of studies have demonstrated behavioural changes in cetacean populations as a consequence of vessel-cetacean interactions. For example, northern resident killer whales (Orcinus orca) on the Pacific coast of Canada change their swimming path from one of a convoluted pattern to that of a more direct path, with an increase in the number of approaching whale-watching vessels (Williams and Ashe, 2007). In New Zealand, the resting behaviour of bottlenose dolphins decreased as the number of boats increased in the Bay of Islands (Constantine et al., 2004) and in Milford Sound

(Lusseau et al., 2006). In Hauraki Gulf, New Zealand the presence of tour boats reduced the likelihood that common dolphins (Delphinus sp.), would continue foraging (Stockin et al., 2008). In Shark Bay, Western Australia, long-term exposure to dolphin-watch vessels has caused long-term, biologically significant effects on the reproductive success and declines in relative abundance of bottlenose dolphins in an area where boat-based tourism occurred (Bejder et al., 2006b).

Anthropogenic noise in the marine environment also affects the quality of cetacean habitat (Tyack, 2008). Noise pollution has the potential to impact cetaceans as they rely on their auditory capabilities as a primary means of communication, foraging and sensing their marine environment. Man-made noise, such as that generated by motorboat engines or sonar, can interfere with cetacean acoustic systems and impair their communication, thereby diminishing their ability to detect natural sounds, including sounds generated by conspecifics (Tyack, 2008, Nowacek et al.,

2007). This acoustic interference, referred to as acoustic masking (Clark et al., 2009,

Jensen et al., 2009), may make cetaceans vulnerable to predation, limit their ability to catch prey, and affect their navigation and communication, which may, in turn, have long term biologically significant effects on the population. Noise levels from small vessels mask the acoustic communications in bottlenose dolphins (Tursiops sp.) and short-finned pilot whales (Globicephala macroryhnchus) (Jensen et al.,

2009). Furthermore, avoiding sonar frequencies leads to disruption of swim path, navigation and usual response to detection of shallow waters (Zimmer and Tyack,

2007). However, at present, insufficient data are available to evaluate the long term effects of anthropogenic acoustic pollution on cetacean populations (National

Research Council, 2005). Thus, the long term monitoring of cetacean behaviour in response to acoustic pollution is important for predicting the effects of this type of impact on the population.

Some tour operators offer swim-with cetacean activities (Kessler and Harcourt,

2010, Courbis, 2007, Constantine, 2001, Samuels and Spradlin, 1995, Bejder et al.,

1999), an increasingly popular activity for tourists (Hoyt, 2007). The methods used to place swim-with customers in the path of a group of wild cetaceans alters their long-term behaviour (Constantine, 2001); e.g. the magnitude of avoidance response of dolphins to swimmers in the Bay of Islands, New Zealand, increased over time and the tour operator’s success with swim-with attempts decreased over an three year period (Constantine, 2001). It has not been possible to evaluate whether these behavioural changes have had long-term biologically significant impacts on the population. This highlights the importance of long-term studies on cetacean populations exposed to commercial tourism activities, to evaluate their effects on the population and developing appropriate management strategies for conserving the population.

2.4 Strategies for managing the cetacean watching industry The approaches used to manage commercial cetacean-watch operations vary around the world. Due to the rapid growth of the industry (Garrod and Fennell,

2004, Hoyt, 2001), planning and management agencies face significant challenges in developing and evaluating management priorities to match this growth (Higham et al., 2008). The fast growth of the industry has also limited the amount of scientific

15 information available to support management decisions, and to evaluate the potential consequences of alternative management options (Higham et al., 2008).

However, a number of cetacean-watch management frameworks have been implemented to mitigate the impacts of commercial cetacean-watch tour operations. These frameworks range from voluntary codes of conduct and guidelines to government legislation (Table 2-1). Carlson (2009) reviewed the cetacean-watching regulations of 47 jurisdictions around the world. Although all these jurisdictions had some form of guidelines or codes of conduct, only six countries had legislation for the general protection of cetaceans, but not specifically directed to the cetacean-watch industry, and 13 countries had legislation that required operators to have permits or licenses to carry out cetacean-watch activities in some region(s) (Table 2-1). The effectiveness of different frameworks such as unmanaged and unregulated cetacean-watch operations, codes of conduct, guidelines, general legislation (where cetacean protection is in place but not specific for the cetacean-watch industry) and permitting strategies, for mitigating any effects on cetaceans from cetacean-watch operations varies considerably.

Table 2-1 The use of permits/licensing, general legislation and guidelines for cetacean-watch tourism in 47 jurisdictions around the world adapted from Carlson (2009). ACCOBAMS = (Agreement on the Conservation of Cetaceans of the Black Sea, Mediterranean Sea and contiguous Atlantic area. Jurisdiction Permit / General Guidelines Licensed regulations for the Legislation protection of cetaceans ACCOBAMS x Antarctica x Argentina x x Australia x x Azores x x Bahamas x Brazil x x British Virgin Islands x Canada x x Canary Islands x x Chile x x Colombia x Dominica x x Dominican Republic x Ecuador x x France x Galapagos x Guadeloupe x Hong Kong x Iceland x Indonesia x Ireland x Japan x Madagascar x Mauritius x Mexico x x Mozambique x New Caledonia x x Newfoundland and x Labrador New Zealand x x Niue x x Norway x Oman x Pacific Islands Region x x Philippines x Puerto Rico x x South Africa x x St Lucia x x

Tanzania x Tonga x Turks and Cacos x United Kingdom x x United States x x Uruguay x Number 47 13 6 47

2.4.1 Unmanaged and unregulated tourism Recently, the growth in nature-based tourism is to shifting from the industrialized nations (Pergams and Zaradic, 2008) to developing countries (Balmford et al.,

2009). This shift has implications for both the wildlife tourism industry and nature conservation in developing countries, where typically, little or no legislation is in place to manage these operations. As ecologically sensitive areas and vulnerable wildlife populations are targeted by tourism operations, these populations may be disturbed or even become extinct. For example, in India, new facilities to house the increasing number of tourists have been constructed in ecologically sensitive areas and migration corridors for wildlife (Karanth and DeFries, 2011). Although nature- based tourism is increasing, if the focal species that draws the tourists to a local area becomes extinct, then tourist numbers will also decrease in that area (Karanth and DeFries, 2011). In 2005, tigers (Panthera tigris) became locally extinct in

Sariska, India, and the annual visitor growth rate decreased by 3% (Karanth and

DeFries, 2011). Could the same scenarios exist for cetaceans?

Possibly, as a consequence of the rapid increase in commercial cetacean-watch operations, and the inability of management agencies to keep pace with this expansion, most operations are still unregulated. These unregulated, commercial

18 operations raise concerns for the populations of cetaceans they target. For example, in New Caledonia in the South Pacific, one of the smallest populations of humpback whales (Megaptera novaeangliae) (< 500 individuals) are known to migrate from Antarctica to breed in this region (Schaffar et al., 2010). During the breeding season, this population is the focus of local daily cetacean-watch operations, and is exposed to more tour vessel activity, for prolonged periods of time, than populations in areas where legislation exists (Schaffar et al., 2010). Self- imposed voluntary guidelines introduced to manage cetacean-watch operations proved ineffective (Schaffar et al., 2010), and in 2009, New Caledonia introduced new legislation to prevent the intentional disturbance of the local cetacean populations (Carlson, 2009).

In the 1990s, unmanaged and unregulated cetacean-watch tourism was initiated in two areas of habitat for the Irrawaddy dolphin (Orcaella brevirostris) in the Mekong

River, Cambodia (Beasley et al., 2010). This population is classified by the IUCN as critically endangered (Beasley et al., 2010). A co-operative project was developed in the region, so that the local stakeholders would share revenue, manage the local industry and provide visitor satisfaction. The success of this project was short-lived, however, as it was de-railed by government interference to instigate a significant increase in the number of cetacean-watch vessels conducting tours (Beasley et al.,

2010). In this case, political priorities disrupted a program aimed at developing a sustainable industry and lead to a decline in the population and loss of tourism. This example highlights the conflict between conservation and development priorities in many countries, and that typically, conservation priorities are low on the political

19 agenda. Cetacean-watch tourism in Lovina, North Bali, Indonesia, developed from local artisanal fishers in the 1980s. Between 35 to 100 tour boats operate daily from

Lovina, targeting mainly the local population of dwarf spinner dolphins (Stenella longirostris roseiventris) (Mustika et al., 2011). Where the practices of the tourism industry is unregulated, up to 83 vessels have been observed surrounding a group of dolphins, vessels were driven erratically, and vessel captains attempted to encroach within the recommended minimum approach distance of 50 m (Mustika et al., 2011). Similarly, other charismatic megafauna such as the whale shark,

(Rhincodon typus), have been targeted by tour operators, and in areas with no legislation, boat strikes and boat crowding have been observed on and around whale sharks (Quiros, 2007). These examples provide strong support for the need, to develop legislation, education programs and incentives to mitigate the negative impact, and potential biologically significant effects to focal wildlife from nature- based tourism.

2.4.2 Guidelines / Codes of conduct Few countries around the world have implemented legislation to protect cetaceans from the effects of human disturbance and fewer countries have legislation that specifically addresses commercial cetacean-watch tourism. However, in an attempt to offer some protection to cetacean populations, numerous self-imposed voluntary codes of conduct and guidelines have been developed for commercial cetacean-watch operators (Garrod and Fennell, 2004). In the absence of specific government legislation focussing on cetacean-watching, codes of conduct and guidelines consisting of agreements between parties who undertake cetacean-

20 watch activities, have been developed to mitigate potential impacts on the target cetacean population. However, guidelines and codes of conduct lack legislative power and legally binding rules. In general, adherence to codes of conduct are based on ethical obligation and peer pressure (Garrod and Fennell, 2004), which are often ineffective in reducing impacts on cetaceans. For example, in Hawaii, commercial cetacean-watch operators have been observed to flout voluntary guidelines, by steering bow riding spinner dolphins (Stenella longirostris) directly to clients in the water (Wiener et al., 2009). In Zanzibar, Tanzania, the guidelines for dolphin-watching are violated as the numbers of cetacean-watch tourist vessels have increased and this is having a detrimental effect on a local population of bottlenose dolphins (Christiansen et al., 2010).

In Port Stephens New South Wales, Australia, a variety of legislative measures were adopted after the voluntary code of conduct failed to adequately reduce the impacts of tourism activities on the local cetacean population (see Allen et al.,

2007). However, the New South Wales government introduced an amendment to their National Parks and Wildlife Regulation to include marine mammals, and that adopted all the aspects of the national guidelines as part of the regulations. Port

Stephens was also declared a Marine Protected Area (MPA), the Great Lakes

Marine Park, to protect the diversity of the region. The inclusion of different zoning areas within this MPA may have a positive effect on the local dolphin population, as it may help protect their natural habitat, and any commercial operation that wishes to undertake dolphin-watching tours in the MPA must obtain a license from the management agency. These amendments to the regulations have provided a

21 mechanism that allows most stipulations within the formally voluntary code of conduct, to be enforced (Allen et al., 2007). In addition dolphins, in the MPA could be further protected from cetacean-watch operations by the implementation of spatial and temporal dolphin watching zones (Allen et al., 2007). It is all well and good implementing different zoning areas within the MPA, but it is vitally important that the efficacy of these zones in mitigating disturbance to the local population of cetaceans is determined by continuous monitoring. For example, in the Great Lakes

Marine Park, speed restrictions zones were introduced as a mitigation measure to minimise boat impacts on the local cetacean population. However, research has identified these speed restriction zones to be ineffective in minimising boat impacts on the local cetacean groups that included calves, and a revision on the placement of these zones was recommended (Steckenreuter et al., 2012). This measure highlights the importance of continuously monitoring the performance of a mitigation approach that attempts to minimise impacts on cetacean populations from tourism. As a consequence, the management agency must be able to react quickly and adapt the existing management framework accordingly, an option that would have been unavailable without a legislative framework in place.

2.4.3 General legislative framework Under general legislative frameworks, legislation is often developed such that any activity that has the potential to be detrimental to a cetacean, can be prosecuted by law. However, these general legislative frameworks aren’t specific to the cetacean-watch industry. The US Marine Mammal Protection Act (MMPA) was designed to minimise harassment and disturbance to marine mammals, primarily

22 for takes from commercial fishing operations as by-catch and from cetacean hunting. The MMPA defines the term ‘take’ as hunting, killing, capture and harassment of a marine mammal or the attempt thereof. However, the interpretation of ‘harassment’ in the MMPA has become a grey area in this legislative framework and it is not clear how activities on behalf of the cetacean- watchers fall within this Act. Under this Act, harassment is defined as follows “The term ‘harassment’ means any act of pursuit, torment, or annoyance which (i) has the potential to injure a marine mammal or marine mammal stock in the wild; or (ii) has the potential to disturb a marine mammal or marine mammal stock in the wild by causing disruption of behavioral patterns, including, but not limited to, migration, breathing, nursing, breeding, feeding, or sheltering.” This leads to confusion and difficulty in determining when cetacean-watch activities are deemed to be harassing a cetacean(s). Under these circumstances, voluntary codes of conduct and voluntary guidelines are often agreed upon and implemented to help mitigate impacts alongside the general legislative framework. For example, these instruments have been developed for the local Hawaiian spinner dolphin population on the island of Hawaii that is protected under the MMPA. In addition to the general protection legislation and codes and guidelines, specific legislation is being considered to further limit impacts.

Due to growing concerns about the potential impact of cetacean-watch operations on the Hawaiian spinner dolphin (Figure 2.1), the National Oceanic and

Atmospheric Administration’s Fisheries Service (NOAA Fisheries) Pacific Islands

Regional Office (PIRO), in conjunction with the Pacific Islands Fisheries Science

Centre (PIFSC), is developing a legislative framework to reduce the exposure of resting spinner dolphins to human activity in Hawaiian waters. Hawaiian spinner dolphins display a rigid diurnal behavioural pattern. At night they venture off shore to feed on shrimp, squid and fish that migrate towards the surface from the mesopelagic zone (Beniot-Bird and Au, 2003). During the day the spinner dolphins move into coastal areas to socialise and rest. These resting areas are usually sheltered, shallow (< 50 m) and have sandy bottoms (Norris and Dohl, 1980, Norris et al., 1994). The rigid behavioural and movement pattern of Hawaiian spinner dolphins and their day-time reliance on sheltered bays to rest and socialize, coupled with the ease of human accessibility to dolphins in these same habitats, is likely to render them more exposed and more susceptible to human disturbance than other dolphin species. At present, it is unclear how these interactions may affect spinner dolphins. Recent studies suggest that the resting periods for the Hawaiian spinner dolphin may be interrupted or truncated by exposure to human activity, although no clear conclusions regarding possible biological significant effects have been researched (Courbis and Timmel, 2009, Danil et al., 2005, Delfour, 2007, Courbis,

2007). Furthermore, research has documented that the genetic diversity of the

Hawaiian spinner dolphin is low (Andrews et al., 2010). Studies of the genetic structure of spinner dolphins along the Kona Coast of the island of Hawaii, which is the target for large scale cetacean-watch operations, show that this population is genetically distinct from all other spinner dolphin populations in the Hawaiian

Archipelago (Andrews et al., 2010). As a consequence, this population may be one of the most vulnerable Hawaiian spinner dolphin populations to anthropogenic disturbance. Thus, developing a legislative framework specifically for the protection

24 of Hawaiian spinner dolphins from cetacean-watch operations is a necessity for the long-term sustainability of both the local spinner dolphin population and the local cetacean-watch tourism industry that relies on this dolphin population.

Figure 2.1: Tour boats and swimmers interact with Hawaiian spinner dolphins in their resting bays off the Kona Coast of the Island of Hawaii. Image taken under permit number GA LOC 15409.

2.4.4 Permitted / licensed legislation framework Management frameworks exist where a legislative system requires commercial cetacean-watch operators to obtain a permit/license to engage in cetacean-watch activities. This legislative framework provides the management agency with the opportunity to regulate the level of exposure of a cetacean population to tourism operations. In this case, should the cetacean-watch industry be shown to have a detrimental effect, the management agency has the power to revoke licenses.

Clearly, reversing or revoking a license creates short-term economic problems for the operators and local communities. However, such a legislative structure allows for such an action. For example, an unprecedented management decision was made on the dolphin-watch operations in Shark Bay, Western Australia, where the number of cetacean-watch operators was reduced from two to one within a

“tourism zone”. The removal of one commercial license operating within a specific area was made after it was shown that the existing number of cetacean-watch operators had reduced the relative abundance and reproductive success of the target bottlenose dolphin population (Bejder et al., 2006b, Higham and Bejder,

2008).

In every circumstance, management agencies lack the scientific basis for deciding on the number of permits to be allocated at the onset of a cetacean-watch industry at a given location. Thus, the allocation of a conservative number of initial licenses is important because of the difficulty of removing licenses once already issued

(Higham et al., 2008). The appropriate number of licenses for allocation is likely to vary between regions and populations as each cetacean population is exposed to cetacean-watch tourism under differing circumstances. For example, some areas are more important than others (critical habitats) and individual susceptibility of animals to impacts vary with age (Müllner et al., 2004, Stalmaster and Newman,

1978, Constantine, 2001), sex, (Lusseau, 2003b, Williams et al., 2002b) previous experience (Bejder et al., 2009, Bejder et al., 2006a) and reproductive condition

(Nellemann et al., 2000, Culik and Wilson, 1995, Beale and Monaghan, 2004, Parent and Weatherhead, 2000). Clearly, the interactions between the tourist operation

26 and cetacean populations need to be investigated scientifically, the potential effects of the interactions determined and the levels of acceptable change determined. For example, cetaceans on their breeding grounds are potentially more susceptible to the effects of cetacean-watch operations than those on their migration corridor.

Legislation does not guarantee operator compliance, particularly when the laws are not communicated adequately to operators (Keane et al., 2011) or are not enforced

(Lundquist and Granek, 2005). For example, in Port Phillip Bay, Victoria, Australia, commercial cetacean-watch operations were routinely documented in breach of four permit conditions: approach type, swim time, time in proximity of dolphins and interaction with newborn calves (Scarpaci et al., 2003). As a consequence, the management framework may ultimately fail as its goals may not be achieved

(Kelleher, 1999). Improvements in operator compliance with regulations may require operator education, tourist education, regulation enforcement or a combination thereof (Scarpaci et al., 2003). Thus, it is important that legislation be explicit for the protection of cetacean populations from cetacean-watch activities and that it is supported by programs to ensure public awareness of the program and that the legislation is enforced to ensure compliance.

2.5 Important considerations for implementing time-area management strategies Effective long-term management strategies are those that monitor cetacean-watch operations, establish thresholds of human-cetacean interactions and respond adaptively to the impacts of cetacean-watch operations and natural phenomena 27

(Higham et al., 2008). Spatial management, including the use of protected areas or closures to commercial operations or no-take/no-watch areas, at the appropriate scale, is an effective approach in protecting both terrestrial and marine ecosystems

(Hoyt, 2005, Pauly et al., 2002). Protected areas have been implemented as precautionary measures when managing marine ecosystems to reduce the risks of overexploitation, especially, when scientific knowledge about the ecosystem is lacking (Hoyt, 2005). This precautionary approach incorporates uncertainty in the decision-making process (Hoyt, 2005, Lauck et al., 1998). For protected areas where scientific knowledge is limited, the spatial area of the protected area might be increased as a precaution to take into account the uncertainty in the available information. Spatial management has been used to conserve biodiversity, protect cetaceans, and delineate areas for specific use to mitigate anthropogenic threats, enhance productivity and provide public focus for marine conservation. It has also been a major part of fisheries management to protect spawning aggregations, immature individuals and critical habitats. In addition to spatial closures, limiting access to cetacean-watching in time, i.e. temporal closures, can be introduced to prohibit human access to cetaceans during specific times that are critical to targeted animals/populations and therefore restrict human-cetacean interaction at these times (Notarbartolo di Sciara et al., 2008, Constantine et al., 2004).

The IUCN (1994) definition of a protected area is “[a]n area of land and/or sea especially dedicated to the protection and maintenance of biological diversity, and of natural and associated cultural resources, and managed through legal or other effective means” and six categories of protected area have been defined based on

28 the on the main management purpose and the primary objective of the protected area (Table 2-2). Several of these categories are significant for the conservation of cetaceans, particularly the Nature Conservation Reserves, that are established to maintain, conserve and restore species and habitats and the Resource Reserves that are designed to protect natural ecosystems and use natural resources sustainably, when conservation and sustainable use can be mutually beneficial

(Table 2, Dudley, 2008).

Protected areas for cetaceans are growing in number around the world (Hoyt,

2011) (Table 2-3) and are contributing to their conservation (Williams et al., 2009,

Notarbartolo di Sciara et al., 2008, Hinch and De Santo, 2010). Currently, the greatest number of these protected areas are found in Australia and New Zealand

(75), the Wider Caribbean (65) and the South Atlantic (56), while the number of reserves is likely to almost double in the Mediterranean and Black Seas should the proposed protected areas be approved (Table 2-3). International boundary agreements between a number of countries have been established to protect cetaceans. For example, a ‘sister-sanctuary’ relationship between the US Stellwagen

Bank National Marine Sanctuary (SBNMS), which is located between Cape Ann and

Cape Cod in the south west of the Gulf of Maine in the north, and Santaurio de

Mamiferos Marinos de la República Dominicana (SMMRD), 3000 miles to the south, was established in 2006 to protect the North Atlantic humpback whale, Megaptera novaeangliae, at both ends of its range (Table 2-3). The SBNMS (2,181 km2) protects the feeding and nursery areas of this population while the SMMRD (2500 km2) protects the mating and calving areas (Ward and MacDonald, 2009).

Area-time management frameworks have been developed to intervene when unregulated and unmanaged cetacean-watch tourism has been identified as having the potential to threaten the viability of the target cetacean population. For example, Samadai Reef, on the coast of the Red Sea, Egypt, was an area well known by locals as an important for area for spinner dolphins (Notarbartolo di Sciara et al.,

2008). Tour operators in the region began to offer unregulated swim-with dolphin tours and the number of tourists visiting the area increased dramatically, with more than 800 swimmers reported to be interacting with spinner dolphins in the small

1.5 km2 lagoon in a single day (Notarbartolo di Sciara et al., 2008). Spinner dolphins presence started to decrease dramatically and concerns were raised about the effects of human activities on the local dolphin population (Notarbartolo di Sciara et al., 2008). As a consequence, authorities took the extreme measure of suspending all tourist visits to Samdai Reef until a suitable management plan was developed. In 2004, an area-time management regime was introduced that included zoning the reef into four different use areas: no tourist zone; diving and snorkelling zone; boat mooring zone; and dive sites zone, and interactions with people were confined to a limited time each day (10 am – 2 pm). In addition, the number of visitors was restricted to 100 divers and 100 snorkelers per day, visitor entrance fees were introduced and visits were allowed only under the supervision of trained and certified guides. Monitoring and enforcement programs were also introduced (Notarbartolo di Sciara et al., 2008).

To achieve the successful implementation of an area-time management plan for cetacean watching, a number of developmental steps are required, these include

30 consultation with the commercial tour operations, the social science community and natural scientists (Higham et al., 2008). It is the responsibility of the management agency to establish and coordinate the development of the legislative framework, to which cetacean-watch tourism operators are required to adhere during cetacean encounters. The legislative framework should also be developed prior to the onset of a commercial tourism program (Higham et al., 2008).

Considering area-time closures as a management framework, sizes, locations of areas and access restrictions to these areas during times when the target cetacean population is deemed most vulnerable to disturbance, should be identified In addition, regulations for the allocation/revocation of permits and legislation to control tourism operations such as restrictions on engine noise, speed/approach, time with dolphins, and the implementation of a visitor interpretation program, should be developed. The likely response of tour operators to restrictions of access to cetaceans also needs to be considered, as this may lead to increased interactions in other areas which may have unforeseen detrimental effects on the population.

For example, if an area-time management framework were introduced in the resting bays of Hawaiian spinner dolphins would this result in more interactions with this population outside the resting bays? This issue of shifts in tour operations in response to restrictions or closures to access, is also a challenge that faces fisheries managers who have the issue of displaced fishing effort following the implementation of sanctuary or no-take zones for fishing (Hilborn et al., 2004).

)

ity, free of ity,

Dudley, 2008 Dudley,

(

tainable use can can tainable be use

will have formed have will been

ortant and landscapes/seascapes

human be human degraded will and forces

supporting environmental processes, supportingprocesses, environmental

of Nature (IUCN) (IUCN) ofNature

term ecological integrity of natural areas of ecological natural term integrity

Primary objective Primary nationally outstanding regionally, or globally Toconserve and or / (occurrences species or aggregations) ecosystems, these attributes features: geodiversity non by or mostly entirely very human light but when to subjected all ordestroyed impact. long Tothe protect activ significant undisturbed human by are that and forces and where infrastructure processes natural modern and the current have so that generations future predominate, areas. to such experience opportunity its along underlying with biodiversity Tonatural protect and ecologicalstructure and recreation. to education promote and and their features natural outstanding Tospecific protect biodiversity and habitats. associated and restore habitats. and species conserve Tomaintain, imp sustain Toand protect created values conservation nature associated other and the humans management traditional with through interactions by practices. and resources ecosystems natural use Tonatural protect and sus when conservation sustainably, beneficial. mutually

e e

a a

scale

nd management reflects nd management

ermanent or significant ermanent

f people and nature over time produced over time nature an has and people f

industrial use of natural resources compatible resources of use natural industrial

level non level

systems characteristic of the area, which also provide a foundation a of provide also characteristic the which area, systems

ociated nature conservation and other values. conservation and nature other ociated

oportion is under sustainable natural resource management and under management resource sustainable natural is oportion

Description also aside areas and protected set biodiversity toStrictly protect where human geomorphological / features, geological possibly to and ensur limited are and use controlled strictly impacts visitation, areas serve can values. conservation protected Such of the protection and monitoring. research reference areas forscientific indispensable as their areas, unmodified large retaining or modified slightly Usually influence, character p and without natural are and which so to as protected preserve habitation, managed human condition. natural their aside large to set protect or natural areas large natural near Are with and along of the complement species ecologicalprocesses, eco culturally scientific, spiritual, and compatible forenvironmentally and recreational visitor opportunities. educational, can which monument, benatural specific a to set protect aside Are as a such submarine feature mount, geological sea cavern, landform, are They feature grove. ancient such an as living a caveor even visitor often areas value. protected and have quite high small generally a species orto particular habitats protect Aim active will regular, IV category areas need priority. protected Many this or of the species requirements particular toto address interventions is this category. not of but the habitats, maintain requirement a o the interaction Where cultural significant biological, ecological, character with of area distinct the integrity of and this safeguarding where scenic value: and its area and is and sustaining to protecting vital the interaction ass cultural with and associated ecosystems habitats, together Conserve They resource systems. and natural are traditional management values where condition, natural a most a of large, in with the area generally pr low where of aims the is area. main of seen one conservation the as nature with

stem conservation conservation stem

Managed for Managed reserve nature Strict area Wilderness Ecosy protection and of natural Conservation features through Conservation management active Landscape/seascape and conservation recreation of use Sustainable resources natural

:Protected area categories, descriptions and primary objectives as determined by the International Union for the Conservation for Union the the International by determined as objectives and categories, primary descriptions area :Protected

al park al

Table Table

Category Ia)Scientific reserve Scientific Ib) reserve II)Nation III)National / monument national landmark IV)Nature conservation reserve V)Protected landscape VI)Resource reserve

Table 2-3 : Protected areas with cetacean habitat across 18 marine regions, adapted from Hoyt (2011). MPA or PA (marine protected High seas MPA National EEZ area or protected area for river (marine protected Sanctuary (cetacean dolphins on land) area outside no hunting zone national waters of within a countries EEZ) national waters or EEZ) Existing with proposed Marine region Existing Proposed expansion Existing Proposed Existing Proposed Total 1 Antarctica 4 0 0 3 1 0 0 8 2 Arctic 29 8 4 0 4 0 0 45 3 Mediterranean 45 38 11 1 17 0 0 and Black Seas 112 4 North West 9 2 1 0 1 1 0 Atlantic 14 5 North East 14 27 8 0 1 2 0 Atlantic 52 6 Baltic 6 6 0 0 0 0 0 12 7 Wider 65 4 2 0 0 4 1 Carribean1 76 8 West Africa 40 4 1 0 0 0 1 46 9 South Atlantic 56 7 3 0 1 3 0 70 10 Central Indian 17 7 8 0 0 1 0 Ocean 33 11 Arabian Seas 20 4 1 0 0 0 0 25 12 East Africa 22 2 2 0 1 1 0 28 13 East Asian 23 8 1 1 0 0 1 Seas 34 14 North and 21 3 4 0 0 12 0 South Pacific 40 15 North East 20 3 5 0 0 1 0 Pacific 29 16 North West 21 3 0 0 0 0 0 Pacific 24 17 South East 30 5 2 0 1 4 0 Pacific 42 18 Australia – 75 7 0 0 0 2 0 New Zealand 84

Total 517 138 53 5 27 31 3 774

Ideally, the management agency would develop a draft management plan based on the collection and analysis of baseline data on the proposed target population, and consultations with key stakeholders, before any cetacean-watch tourism operations commence. However, this is rarely the case (Bejder and Samuels, 2003), and management frameworks are generally implemented after cetacean-watch tourism has begun and when concerns have been identified about the viability of the target cetacean population.

2.5.1 Understanding the cetacean population targeted by the cetacean-watch industry Critical habitats for cetaceans are areas in which the cetacean species carries out behaviours that are vital to the viability of the underlying population. These areas include foraging, breading, nursing, socialising or resting habitats (Hoyt, 2011).

Repeated disturbance of cetaceans within these critical habitats has been implicated as a factor in reducing the viability of the target population (Bejder et al.,

2006b, Lusseau et al., 2006). Prior to developing a management plan, it is important to understand the behaviour and biology of the focal population, particularly to identify critical areas, and the time of their use (Lusseau, 2003a, Lusseau et al.,

2009). A monitoring program should consist of a rigorous sampling design that allows an understanding of the target cetacean population to be developed. The sampling design should incorporate methodologies that allow reliable and detailed data on the population, its habitat, and potential threats to be collected. The sampling program should aim to estimate population size, population structure

(e.g. age structure, sex ratio), reproductive rates and behavioural budgets (Lusseau,

2004) and provide data for developing models to identify critical areas and habitat

35 use. For example, NOAA initiated the development of a research program to understand the spinner dolphin population and its interactions with cetacean watching activities along the west coast of the island of Hawaii. A suite of modern visual and acoustic techniques, including boat-based photographic identification/focal follows, land-based theodolite tracking, and bottom-mounted acoustic loggers, are being used in this study. A systematic photographic identification sampling regime, based on Pollock’s Robust Design mark-recapture model (Pollock et al., 1990), was developed to determine spinner dolphin population parameters in four resting bays along the west coast of the island of

Hawaii. In addition, group focal follows are carried out, to observe behaviours and human interactions outside the resting bays. Where possible land-based theodolite tracking of spinner dolphins, their behaviours and human interactions within the resting bays are carried out. In addition to understanding the characteristics of the cetacean population and its interactions with tourism operations, abiotic and biotic habitat characteristics should be mapped. The potential threats from cetacean- watch operations to the habitat should be understood, and the potential for these threats to effect the target cetacean population assessed.

Cetaceans transition between behavioural states over time. For example, cetaceans can display foraging behaviour followed by socialising behaviour which may be followed by resting behaviour. When cetaceans are disturbed, however, the probability of transitioning from one behavioural state to another is altered

(Lusseau, 2003a), resulting in a change in their overall behavioural budget. Thus, if the probability of transition from one behavioural state to another, in the absence of cetacean-watch tourism, can be identified for a species, then an change in this

36 transition probability could provide an early warning indicator of a deleterious effect in the presence of cetacean watching tourism (Lusseau, 2004). Furthermore,

‘show stoppers’, such as a decline in reproductive success, should be identified that provide immediate evidence of significant impact and would result in management intervention (Bejder et al., 2006b).

Baseline data should be used to develop monitoring programs and management plans (Higham et al., 2008) so that changes in the population can be monitored during the onset and growth of tourism activities (Bejder and Samuels, 2003).

Reference points can be developed from the indicators and the baseline data to provide points target reference points for tourism and limit reference points, at which management actions are initiated, i.e. when pre-determined acceptable thresholds have been exceeded. Target and limit reference points are used commonly in fisheries to monitor and manage the health of fish stocks and their ecosystems, e.g. spawning biomass at Maximum Sustainable Yield, Percent of Virgin

Biomass, and Spawning Potential Ratio. Similarly, Potential Biological Removal (PBR)

(Wade, 1998) assesses the allowable limits of mortality from anthropogenic disturbance (Williams et al., 2009). The PBR is calculated as the product of a minimum population estimate (Nmin, one half of the maximum theoretical net productivity rate (Rmax) and recovery factor (Fr) (PBR = Nmin x 0.5Rmax x Fr)

(Wade, 1998).

Area-time management systems require an understanding of why and when specific habitats/areas are critical to the focal population. Critical habitats

37 encompass areas of high animal density, and those areas essential to the viability of the population e.g. a nursing area where only mothers and calves are present.

Critical habitats include foraging, breeding, nursing, socialising and resting areas

(Hoyt, 2005, Lusseau and Higham, 2004, Williams et al., 2009, Ross et al., 2011).

Quantifying the importance of an area to a cetacean population by assessing habitat preference, and the behaviours in these habitats, is becoming an important tool in the development of area-time management systems to mitigate anthropogenic threats (Higham and Lusseau, 2007). For example, a high proportion of the population of northern resident killer whales (Orcinus Orca), in Johnstone

Strait, Canada, uses a small proportion of their habitat for a rare behaviour called beach-rubbing, where individuals rub their bodies on the smooth pebble beaches, thought to remove parasites or have some social significance (Williams et al., 2009).

This does render the population vulnerable due to the high proportion of the population that uses this small area, coupled with the heavy human use of

Johnstone Strait by large ships nearby (Williams et al., 2009). Bottlenose dolphins in

Fiordland, New Zealand, are particularly sensitive to boat interactions while resting, and to a lesser extent, while socialising (Lusseau, 2004). Using behavioural state observations, Lusseau and Higham (2004) identified critical areas for dolphin resting and socialising in Doubtful Sound. However, a voluntary code of conduct was introduced which includes some elements of zoning (unregulated), but not in the areas identified by Lusseau and Higham (2004). Similarly, the commercial dolphin- swim/watch industry in the Bay of Islands, New Zealand, alters the behaviour of bottlenose dolphins (Constantine et al., 2004). Resting behaviour was the most sensitive behavioural state to commercial boat numbers, and decreased

38 significantly with increasing number of vessels. Constantine et al. (2004), suggested that the local legislation was ineffective in protecting the bottlenose dolphin population from commercial operations, and, as a consequence, recommended measures to minimise tour-boat impacts by restricting the number of boat trips, trip durations and limiting the amount of time dolphins are exposed to tour-boats.

Identifying critical habitat should also consider both the abiotic and biotic characteristics of the habitat, and link these characteristics to the focal cetacean population’s behaviour while using the area. These characteristics may include abundance of prey, and characteristics of the bathymetry, substrate, temperature, salinity, turbidity, tide and currents. For example, Hawaiian spinner dolphins prefer sheltered sandy bays to rest in during the day (Norris and Dohl, 1980). Any changes to this habitat, may have significant biological consequences for the cetacean population and the efficacy of the protected area. For example, it was suggested that a decline in prey resources inside a protected area in the Moray Firth, Scotland, caused a population of bottlenose dolphins to extend their range outside the boundaries of the protected area (Wilson et al., 2004).

Oceanographic features may also be used to identify critical cetacean habitats.

Johnston and Read (2007) highlighted an ecological link between a predictable oceanographic feature in time and space that attracts cetaceans to the Bay of

Fundy, Canada. In the summer and during flood tides, the Grand Manan Island wake attracts foraging fin whales (Balaenoptera physalus), minke whales,

(Balaenoptera acutorostrata) (Johnston et al., 2005a) and harbour porpoise

(Phocoena phocoena) (Johnston et al., 2005b). Oceanographic observations provided an understanding of the spatial and temporal variability in the physical forces controlling the island wake (Johnston and Read, 2007). Secondary flows in the wake aggregate prey to predictable locations where the cetaceans focus their foraging efforts. These ecological links between the foraging habitats and foraging behaviour over space and time are therefore important factors when considering the spatial boundaries of a marine protected area in this region and Johnston and

Read (2007) recommended that a proposed protected area at the Grand Manan

Island encompass the island wake. The predictable nature of this oceanographic feature also allows human activities to be controlled within the protected area during the predictable tidal flows that generate the island wake and while being exploited by foraging mega fauna.

2.5.2 Socio - economic considerations Cetacean-watch operations are important for the economic sustainability of many coastal communities around the world (Hoyt, 2007). Access restrictions to protect cetacean populations from tourism operations may have implications for the economic status of local coastal communities. Therefore, it is important to highlight the benefits of the sustainable management of cetacean watching operations and the significance of area-time management systems to local businesses and the wider community, with the aim of maximizing the economic viability of the local cetacean-watch industry, while sustainably managing the target cetacean population.

Raising awareness of the conservation efforts among visitors and the local community is an important component in the development of any long-term management framework for cetacean watching. Many commercial cetacean-watch operations provide an educational component that provides information on the target cetacean population, its environment, the associated conservation efforts, and to some degree the plight of cetaceans worldwide (Orams, 1997, Higham and

Carr, 2002, Christensen et al., 2007). In some jurisdictions, e.g. New Zealand, tourism operations are required to provide an educational component as part of the cetacean-watch experience, in order to obtain a cetacean-watch permit

(Carlson, 2009). As a consequence, in general, these cetacean-watch operations are perceived to be beneficial and as ecotourism (Bejder and Samuels, 2003). However, unless the cetacean-watch operation is environmentally responsible and contributes to conservation, it should be considered as nature-based tourism, not ecotourism (Bejder and Samuels, 2003). The debate is continuing on the efficacy of these education programs in achieving the goal of raising environmental and conservational awareness and ultimately changing human behaviour (Orams, 1996,

Ballantyne et al., 2010, Higham and Carr, 2002). Studies are showing that carefully designed educational nature-based programs that incorporate strategies to facilitate behavioural change, can instil greater environmental and conservational awareness in tourists and lead to changes in their behaviour and attitude to interactions with marine fauna (Orams, 1997, Higham and Carr, 2002, Ballantyne et al., 2010). It is important, however, that social science research continues to collect data on visitors and their perceptions on the cetacean watching operation to

41 enable the effectiveness of the education program to be assessed and adapted if necessary.

2.5.3 Management considerations Reliable and detailed scientific data on the critical habitat of cetaceans should be used to provide management agencies with the information to establish protected areas. The spatial scale of the protected area should be large enough to be biologically relevant and small enough that cetacean-watch operations can be effectively managed within its boundaries (Ashe et al., 2010, Ross et al., 2011). In addition, the temporal scale of the behaviour identified as critical, should inform management agencies on when restrict human access to the protected area.

Population estimates, reproductive rates and changes in behavioural budget

(Lusseau, 2004, Bejder et al., 2006b), should inform management agencies in establishing quantifiable Limits of Acceptable Change (LAC) in the target cetacean population (Higham et al., 2008). Spatial and temporal scales and LAC criteria should then be used to establish clearly defined legislation for operators and enforcement within the protected area boundaries.

Rules and regulations are only part of the necessary tools to help ensure that commercial cetacean-watch operations minimise impacts on the focal cetacean population - these rules and regulations must be supplemented with education and enforcement programs to ensure the success of a management plan (Keane et al.,

2008). Without these programs, compliance with the management plan may be low, which makes the rules and regulations irrelevant, as no distinction would exist

42 between the protections offered within the management area to those outside its boundaries (Guidetti et al., 2008, Vanzella-Khouri, 2009). A lack of compliance to rules and regulations relating to the reduction of impacts on cetaceans from cetacean-watch operations, has been demonstrated where legislation has been invoked and enforcement is not present or is ineffective (Scarpaci et al., 2003, Allen et al., 2007, Wiener et al., 2009, Christiansen et al., 2010). Legislation, clearly defined, should provide management agencies with the authority to revoke the operator licenses (Bejder et al., 2006b, Higham and Bejder, 2008). Therefore, it seems prudent that management agencies develop a licensing system that provides clear instruction to the commercial cetacean-watch operation on permit conditions, including the circumstances under which a permit could be revoked. Without enforcement or legislation, conservation management plans may fail to meet their goals and, ultimately, fail to protect the focal cetacean population and as a consequence the long-term viability of the cetacean-watch tour operations.

2.6 Summary The increase in cetacean-watch tourism, particularly since the 1970s, and its rapid expansion into developing countries, has in general, been perceived as beneficial and a benign activity. However, concerns have been raised about the impacts of cetacean-watch tourism on targeted populations around the world, and whether these impacts are likely to cause biologically significant effects. In this chapter I argue that area-time management systems should be considered as an important tool to manage cetacean-watch tourism.

The important challenge for scientists is to develop rigorous and reliable techniques that quantify the behavioural dynamics of cetacean populations. This knowledge can then be used to quantify the effects of both natural phenomena and anthropogenic disturbance on these behavioural dynamics, to identify biologically significant changes in the underlying population. Providing estimates on population size, identifying habitat use, behaviours in different habitats, and the behavioural changes in cetacean populations in response to cetacean watching operations, scientists will then be able to identify when and where cetacean populations are at their most vulnerable. Management agencies can use this information as a basis to develop the appropriate legislative management framework to protect cetacean populations from disturbance when they are most vulnerable. In addition, the efficacy of the management framework, as a mitigation approach, must be continuously monitored and should adapt quickly and accordingly when necessary

The management framework must also provide management agencies with sufficient legislation to prevent tour operators from operating when rules have been violated. I argue that a licensing system should be mandatory, so that management agencies have the authority to revoke a licence should it be deemed necessary. Furthermore, studies have shown that effective education programs and enforcement are essential to ensure that operators comply with legislation; poor- compliance is likely to result in detrimental effects on the cetacean population, the cetacean-watch operators and the socio-economic state of the community that relies on the tourism industry.

3 Abundance and survival rates of the Hawaii Island associated spinner dolphin (Stenella longirostris) stock

3.1 Abstract Reliable population estimates are critical to implement effective management strategies. The Hawaii Island spinner dolphin (Stenella longirostris) is a genetically distinct stock that displays a rigid daily behavioural pattern, foraging offshore at night and resting in sheltered bays during the day. Consequently, they are exposed to frequent human interactions and disturbance. I estimated population parameters of this spinner dolphin stock using a systematic sampling design and capture-recapture models. From September 2010 to August 2011, boat-based photo-identification surveys were undertaken monthly over 132 days (>1,150 hours of effort; >100,000 dorsal fin images) in the four main resting bays along the Kona

Coast, Hawaii Island. All images were graded according to photographic quality and distinctiveness. Over 32,000 images were included in the analyses, from which 607 distinctive individuals were catalogued and 214 were highly distinctive. Two independent estimates of the proportion of highly distinctive individuals in the population were not significantly different (p=0.68). Individual heterogeneity and time variation in capture probabilities were strongly indicated for these data; therefore capture-recapture models allowing for these variations were used. The estimated annual apparent survival rate (product of true survival and permanent emigration) was 0.97 SE±0.05. Open and closed capture-recapture models for the highly distinctive individuals photographed at least once each month produced similar abundance estimates. An estimate of 221±4.3 SE highly distinctive spinner dolphins, resulted in a total abundance of 631±60.1 SE, (95% CI 524-761) spinner dolphins in the Hawaii Island stock which is lower than previous estimates. When 45 this abundance estimate is considered alongside the rigid daily behavioural pattern, genetic distinctiveness and the ease of human access to spinner dolphins in their preferred resting habitats, this Hawaii Island stock is likely more vulnerable to negative impacts from human disturbance than previously believed.

3.2 Introduction Many islands in tropical and sub-tropical regions represent isolated oases of marine life, exhibiting higher levels of primary productivity, secondary productivity and enhanced communities of top predators than the oligotrophic pelagic background around the islands (Wolanski and Hamner, 1988). In many situations, the cetacean top predators that have evolved to exploit island-associated productivity in these regions represent resident, isolated populations, often with high site fidelity and restricted gene flow amongst nearby island regions (Aschettino et al., 2012, Baird et al., 2008, Martien et al., 2012). Furthermore, many island associated small cetacean populations exhibit specialized behaviours and social dynamics that have evolved to facilitate their survival. However, due to their specialized demography and behavioural ecology, it is becoming increasingly clear that island-associated, populations of small odontocetes may be particularly vulnerable to anthropogenic effects (e.g. false killer whales in the Hawaiian Archipelago). Hawaiian spinner dolphins represent one such species – they exist as small isolated populations with restricted ranges and exhibit a specialized behavioural ecology (Norris and Dohl,

1980, Norris et al., 1994) that renders them vulnerable to human activities in coastal environments. Spinner dolphins occur in sub-tropical and tropical oceans worldwide and are named because of their aerial behaviours (Perrin, 1990). Gray's spinner dolphin, (Stenella longirostris), is the most widely distributed subspecies

(Perrin et al., 1999) and occurs throughout the entire Hawaiian archipelago.

The Hawaiian archipelago consists of the mainly uninhabited North West Hawaiian

Islands, from Kure Atoll in the north to the eight inhabited main Hawaiian Islands,

47 with Hawaii Island in the south. Recent genetic analyses revealed that Hawaiian spinner dolphins are distinct from populations found elsewhere (Andrews, 2009), and moreover, subpopulations within the Hawaiian archipelago were also found to be genetically distinct (Andrews et al., 2010). As a consequence, Hawaiian spinner dolphins have been divided into five different island/island-group management units under the US Marine Mammal Protection Act (MMPA) by the National

Oceanic and Atmospheric Administration’s National Marine Fisheries Service

(NMFS) that correspond with two broad geographical regions: 1.) three in the Main

Hawaiian Islands: Hawaii Island, Oahu/4-Islands area, Kauai/Niihau, and 2.) two in the Northwest Hawaiian Islands: Pearl & Hermes Reef and Kure/Midway. The NMFS is mandated by the MMPA to assess the population status and threats for all identified stocks of marine mammals in U.S. waters.

At present, reliable abundance estimates are not available for any stock of

Hawaiian spinner dolphins, a significant impediment to developing appropriate management plans for any spinner dolphin management unit in Hawaii. Previously, a line transect survey of the entire Hawaiian Exclusive Economic Zone resulted in an abundance estimate of 3,351 (Barlow, 2006) spinner dolphins throughout the entire

Hawaiian archipelago which assumed a single Hawaiian stock (Carretta et al., 2013).

Considering the ship’s track and the coastal daytime reliance of this species, the large ship-based estimate provided by (Barlow, 2006) is not appropriate for estimating the abundance of inshore spinner dolphins. Other studies which estimated the abundance of spinner dolphins along the Kona Coast were based on opportunistic photo-identification sightings and were not specifically designed to

48 estimate abundance (Norris et al., 1994, Ostman, 1994, Ostman-Lind et al., 2004,

Mobley et al., 2000). As a consequence, a collaborative project, ‘spinner dolphin acoustics population parameters and human interaction research’ (SAPPHIRE) was developed in 2010 to assess the abundance, distribution and behaviour of spinner dolphins along the Kona Coast. The SAPPHIRE project combines boat-based photo- identification and group focal follows and land-based theodolite observations, along with passive acoustic monitoring, to evaluate the effects of human interactions on spinner dolphins in the region.

Hawaiian spinner dolphins exhibit a rigid, diel behavioural pattern. At night, they forage cooperatively offshore in deeper water (Benoit-Bird and Au, 2009). During the day, they move into shallow, coastal habitats to rest and socialise (Norris and

Dohl, 1980), preferring sandy-substrate locations that are sheltered from the wind, typically less than 50 m deep (possibly to aid in predator detection) and within close proximity to their deep-water foraging areas (Thorne et al., 2012, Norris and Dohl,

1980, Norris et al., 1994). This rigid, behavioural pattern is unlike the less predictable patterns observed in other coastal dolphin species, such as the bottlenose dolphin (Tursiops sp.), a species known to readily switch between behavioural states, e.g. from foraging to resting to socialising (Bearzi et al., 1999).

To maintain this rigid behavioural pattern, spinner dolphins are dependent on these sheltered bays to rest (Norris and Dohl, 1980, Norris et al., 1994). However, within these same habitats, dolphins are easily accessible and thereby exposed to human interactions and disturbance (Courbis and Timmel, 2009, Ostman-Lind, 2008,

Timmel et al., 2008, Wiener et al., 2009). When anthropogenic impacts are

49 considered in combination with their genetic distinctiveness and low gene flow,

Hawaiian spinner dolphin’s susceptibility to human disturbance is of serious and increasing concern for the survival of the stock.

During periods of activity, animals usually exhibit enhanced brain function, which is often referred to as vigilance. Vigilance is required for many activities including foraging, socializing and predator avoidance. As animals undertake these cognitively challenging activities they tire, and accrue what is often referred to as a vigilance decrement (Dukas and Clark, 1995). In higher vertebrates, vigilance decrements can manifest in a decreased ability to detect camouflaged predators or cryptic prey (Dukas and Clark, 1995). They may also manifest in more abstract ways such as reduced decision-making capabilities (Dukas and Clark, 1995). To recover from a vigilance decrement, animals must rest (Cirelli and Tononi, 2008). The derived behavior of spinner dolphins renders them especially vulnerable to interrupted resting bouts during the day, as they have a limited ability to recover before embarking on another foraging bout the following evening.

Dolphin-watch tourism can cause biologically significant effects on exposed communities by causing habitat displacement (Lusseau, 2005, Bejder et al., 2006b).

Short-term studies reveal that an increase in vessel, kayak and swimmer traffic both inside and outside of known resting bays in Hawaii have resulted in spinner dolphins spending less time in important habitats (Ostman-Lind et al., 2004).

Consequently, their rest periods are truncated and interrupted (Courbis, 2007,

Courbis and Timmel, 2009, Delfour, 2007). This type of anthropogenic disturbance

50 of spinner dolphins in their resting habitat may have negative, long-term impacts that will likely reduce their distribution and abundance over the long term (Ostman-

Lind et al., 2004, Lammers, 2004). Unfortunately, current scientific literature is lacking accurate information to inform how long-term disturbance may impact

Hawaiian spinner dolphins, specifically in response to the cumulative exposure of human disturbance in important resting habitats.

For small cetaceans, capture-recapture studies, based on photo-identification, have proven to be a reliable method for estimating population parameters, such as abundance, survival and recruitment rates (Pollock et al., 1990, Nicholson et al.,

2012, Cantor et al., 2012, Silva et al., 2009, Gormley et al., 2005, Smith et al., 2013).

However, the characteristics of individual cetaceans and the methods used to photograph them can introduce heterogeneity in the capture probabilities and misidentification of individuals (Hammond et al., 1990). Careful attention to the study design can help improve the adherence of the sampling methodology, to the assumptions of capture-recapture models and mitigate biases due to heterogeneity and misidentification of dolphins. Two types of population models are generally considered for capture-recapture sampling designs: closed and open population models (Wilson et al., 1999, Larsen and Hammond, 2004, Gormley et al., 2005,

Cantor et al., 2012, Nicholson et al., 2012, Smith et al., 2013). During long-term studies, it is not always possible to assume that the population being studied is closed, and therefore, open population models should be used (Pollock et al., 1990,

Nicholson et al., 2012, Smith et al., 2013).

As part of the SAPPHIRE research project, the objectives of this study were to estimate the population abundance and survival rate of the Hawaii Island spinner dolphin stock using a systematic sampling design, and both open and closed capture-recapture population models. The resulting scientific data are the first to provide accurate and reliable baseline population estimates for this stock. This information will be useful for management agencies for both stock assessment purposes, and to assess the effectiveness of planned management actions that are aimed at mitigating negative impacts of human-dolphin interactions.

3.3 Materials and methods 3.3.1 Fieldwork The Hawaiian archipelago is located in the Pacific Ocean, approximately 3,200 km southwest of mainland United States. Hawaii Island is the largest, youngest and most southerly of the main Hawaiian Islands. On the leeward side of the island is the Kona Coast, where the four main spinner dolphin resting bays are located

(Figure 3.1): Kauhako Bay, Honaunau Bay, Kealakekua Bay and Makako Bay (Norris and Dohl, 1980, Norris et al., 1994, Courbis and Timmel, 2009). In addition, these bays are consistently used by boats, kayaks, stand-up paddle boards and swimmers for recreational purposes, thus providing opportunities for people to interact with the resting dolphins (Ostman-Lind, 2008, Courbis and Timmel, 2009, Timmel et al.,

2008, Norris et al., 1994)

3.3.2 Sampling design A systematic sampling design was developed to study the Hawaii Island spinner dolphin stock. From September 2010 to August 2011 (excluding May 2011) boat- based photographic-identification surveys were carried out during 12 days of each month in the four resting bays in a sequential order: Kauhako Bay for four days;

Honaunau Bay for two days; Kealakekua Bay for four days; and Makako Bay for two days. I would arrive at a bay (only one bay each day) at 0700h. If the dolphins weren’t present I would wait until 1600h to see if they would arrive. I carried out boat-based photo-identification (see below) if dolphins were present (or arrived during the day). Each bay was systematically surveyed on the same dates each month, regardless of whether dolphins were present or absent. This sampling regime provided consistent and even effort throughout the study period and area.

3.3.3 Photographic-identification The boat-based photo-identification team consisted of three to five observers with two digital SLR cameras: a Nikon D300s and a Nikon D300, both with Nikon 80mm to 400mm AF VR Zoom lenses. I used a ‘100 m chain rule’ (Smolker et al., 1992,

Gero et al., 2005), to determine members of each group of spinner dolphins, where any animals within 100 m of each other were considered to be members of the same group. When a dolphin group was sighted the dolphins were approached for surveying and group size was determined. With dolphin groups <=20 I had a greater probability of obtaining good photographs of all individual group members which, in turn optimised the chance that the more distinctly marked individuals were not more likely to be photographed than the less distinctly marked individuals.

Photographs were taken when dolphins surfaced within 25 m of the research vessel. A dolphin survey would last a minimum of 30 minutes and maximum an hour with a minimum of a 30 minute break between surveys. Breaks between dolphin group surveys were to limit the disturbance to the focal group from the research vessel. Repeated dolphin group surveys optimised the probability of capturing all animals in the group. Field observations also noted if groups from outside the bays joined the focal group. Dolphin surveys would continue throughout the day until either: the whole group was photographed, the dolphins left the bay, or when environmental conditions deteriorated, i.e. sea state >

Beaufort 2.

3.3.4 Grading and sorting of photo-identification images All photographs were graded according to photographic quality and distinctiveness in order to minimise the introduction of bias and to reduce misidentification (Urian et al., 1999, Friday et al., 2000, Gowans and Whitehead, 2001, Nicholson et al.,

2012). Following (Urian et al., 1999), all photographs were assigned absolute values based on clarity and focus (2, 4 or 9), degree of contrast (1 or 3), angle of dorsal fin to the camera (1, 2 or 8), dorsal fin visibility and the proportion of the frame filled by the dorsal fin (1 or 5). These values were then summed to produce an overall image quality score. Excellent quality images received scores of 6-7, good quality images had scores from 8-11, and poor quality images had scores > 11 (Nicholson et al., 2012, Urian et al., 1999).

Dorsal fin distinctiveness varies between individual dolphins; thus not all fins were distinctively marked enough to be included in capture-recapture analyses (Read et al., 2003, Wilson et al., 1999, Nicholson et al., 2012). As a consequence, photographs were analysed for individual distinctiveness based on patterns of nicks and notches on the leading and trailing edges of the dorsal fin that were visible from both sides (Urian et al., 1999). Overall distinctiveness was based on a scale of

D1 (highly distinctive, features evident in distant and poor quality photographs), D2

(smaller less distinctive nicks and notches) and D3 (not distinctive) (Urian et al.,

1999, Nicholson et al., 2012). Individuals with a distinctiveness rating of D1 or D2 were integrated into the photographic-identification catalogue and highly distinctive individuals (D1) were used to calculate the mark rate of the stock which in turn, was used to scale up to estimate total stock size. Every individual was compared to all others in the catalogue before being assigned a unique identification code and added separately to the catalogue. Individuals with a distinctiveness rating of D3 were given a generic identification code but not included in the catalogue.

3.3.5 Analyses A capture was defined as a photograph of sufficient quality of an individual dolphin’s distinctly marked dorsal fin. Only highly distinctive (D1) fins in photographs of excellent and good quality were included in the capture-recapture analyses to reduce misidentification errors. Capture histories corresponded to whether or not an individual was “captured” or “recaptured” during a sampling occasion. This information was compiled for each individual (excluding calves), after

56 the photo grading process. The program MARK (White and Burnham, 1999) contains a suite of capture-recapture models and goodness-of-fit tests. Using MARK open and closed capture-recapture models were then applied to these data.

All capture-recapture models make the following assumptions (Williams et al.,

2002a): 1) marks are not lost during the study; 2) marks are correctly recognised on recapture; 3) individuals are instantly released after being marked; 4) intervals between sampling occasions are longer than the duration of a sample; 5) all individuals observed during a given sampling occasion have the same probability of surviving until the next one; 6) study area does not vary; and 7) homogeneity of capture probabilities, i.e. that all animals in a sampling occasion have equal probability of being captured. This assumption is relaxed for certain models which do allow heterogeneity of capture probabilities.

3.3.6 Estimating abundance and demographic parameters A variety of closed and open capture-recapture models were fitted using the program MARK (White and Burnham, 1999). They used the capture histories of all highly distinct individuals captured on at least one occasion during each month in any of the four bays. Therefore, the population abundance estimate refers to the highly distinct individuals.

POPAN (Schwartz and Arnason, 1996) is an integrated combined likelihood formulation of the original Jolly-Seber open capture-recapture model (Jolly, 1965,

Seber, 1965). POPAN estimates a super-population size (N), entry probabilities,

57 apparent survival rates and capture probabilities. Maximum likelihood was used to estimate the following parameters: N, the super population size, which is all the animals that existed in the population (stock) at any point during the study period;

φt is the apparent survival probability from sampling period t to sampling period t +

1 and is the product of true survival times the probability the animal does not emigrate; pt is the probability that an individual available for capture in sampling period t would be captured in sampling period t; and βt is the probability of entry of an individual into the population between sample t and sample t + 1. Derived estimates of the stock sizes at each sampling time (Nt) can also be estimated if necessary.

A suite of POPAN candidate models were developed to allow for fixed or time- varying effects on the entry probabilities, apparent survival rates and capture probabilities. For model selection, Akaike's Information Criterion (AIC) was applied, which provides a measure of the model fit but is penalized when there is an increase in the number of parameters (Akaike, 1974). RELEASE, a goodness-of-fit program in MARK (White and Burnham, 1999), was used to determine goodness-of- fit for the POPAN models (Lebreton et al., 1992). Over-dispersion in the models was accounted for, by estimating the over-dispersion measure ĉ using the chi-square statistic divided by its degrees of freedom. QAIC values were used for model selection (Anderson et al., 1994) with the lowest QAIC value an indication of the most parsimonious model.

MARK was also used to obtain closed population model estimates for models that allow heterogeneity and time variation of capture probabilities (M0, Mh, Mt and Mth)

[38], because spinner dolphins are long lived. The advantage of using the closed models, if appropriate, is that they provide estimates with higher precision than open models and allow for heterogeneity of capture probabilities among individuals, which is very common in most capture-recapture studies [22, 38].

3.3.7 Estimation of mark rate and total stock size Estimates of the stock size from the capture–recapture models relate only to the identifiable animals in the study. Therefore, to estimate the total stock size, estimates need to be scaled based on the proportion of individuals that are identifiable. Here, I estimated the proportion of highly distinctive individuals (D1) in the Hawaii Island spinner dolphin stock using two independent measures of mark

ˆ ˆ rates: 1 and  2 .

Mark Rate 1 ( ): For groups consisting of > 20 dolphins, a mark rate was calculated from the proportion of randomly taken photographs that contained identifiable dolphins that were obtained from the two photo-identification cameras that were working simultaneously (Williams, 1993, Wilson et al., 1999). To be included in the analyses, photographs had to be of sufficient quality to identify a dolphin if it had been identifiable.

number of high quality photographs with highly distinctive fins ˆ  1 total number of high quality photographs with distinctive and non distinctive fins

ˆ Mark Rate 2 ( 2 ): A second independent mark rate was calculated using only the photo-identification data collected for group sizes that were ≤ 20 dolphins. Unlike with large groups, this scenario assumed that all individuals in the group were photographed to a quality that would allow dolphins to be identified if they were identifiable. Thus, for each group that consisted of ≤ 20 dolphins, was calculated based on the knowledge of group size, together with the number of highly distinctive individuals in each group:

ˆ total number of highly distinctiv e individuals in each group 2  total dolphin group sizes

The standard errors (SE) for both mark rate estimates are:

ˆ(1ˆ) SE(ˆ)  n where n is the sample size in each equation.

Both of these methods assumed that the proportion of identifiable individuals in the sample was equivalent to the proportion of identifiable individuals in the entire stock (Hammond, 1986). The numbers of highly distinctive and non-distinctive individuals were summed over all surveys and used to estimate the total number of individuals in the stock:

ˆ ˆ N dist Ntotal  ˆ

Where is the estimated abundance of all individuals (distinctive and non- distinctive) identified during the study period, is the abundance estimate of the highly distinctive individuals, and ˆ is the estimated proportion of distinctive individuals (Burnham et al., 1987).

The variance for the total stock size estimate was derived as follows (Williams et al.,

2002a):

2  SE Nˆ 1ˆ  SE(Nˆ )  Nˆ 2   dist    total total  ˆ 2 ˆ  Ndist n  

Log-normal 95% confidence intervals were calculated with a lower limit of

ˆ L ˆ ˆ L ˆ ̂ L ̂ N total  Ntotal / C and upper limit of N total  Ntotal x CNtotal = NtotalxC,

(Burnham et al., 1987) where:

  ˆ 2      SE(Ntotal )    C  exp1.96 ln 1      Nˆ      total     

3.4 Results 3.4.1 Effort and summary statistics From September 2010 to August 2011, photo-identification surveys were carried out for a total of 132 days (> 1,150 hours of effort; > 100,000 dorsal fin images) in the four bays. More than 32,000 images were of sufficient quality to be added to the catalogue, from which 607, D1 and D2 individuals were identified and contained

214 highly distinctive, D1, individuals.

Seventy-six percent of individuals were photographed on more than one occasion, with one individual photographed as many as 18 times (Figure 3.2). On average, individual spinner dolphins were photographed on four (SE±0.14) occasions during the study period (Table 3-1). A cumulative discovery curve (Figure 3.3) indicated that the identification of new individuals was reaching a plateau before the end of the study period, with few new dolphins identified after 120 days of effort. Resting 61

bay usage of individual spinner dolphins varied, in that some individuals were only

photographed in one resting bay, while others were observed in all four resting

bays (Figure 3.2).

160 145

140

120 96 100 82 80 55 60 49 37 37 40 27 21 19 19 20 9 3 4 1 1 1 1 Number of individuals of Number 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Number of sightings

Figure 3.2 Frequency of individual spinner dolphin sightings from September 2010 to August 2011.

Table 3-1 Number of photographic identification surveys, hours in each bay, spinner dolphin encounters in four resting bays along the Kona Coast of Hawaii Island from September 2010 to August 2011. Mean Total Hours Hours Group group Min Max survey dolphins dolphins Photo-id encounter size group group Location Surveys hours absent present hours rate % (±SE) size size Kauhako Bay 44 407 257 150 28 39 29±3 8 50 Honaunau Bay 22 195 116 79 16 41 24±4 6 40 Kealakekua Bay 44 397 171 226 52 52 41±6 5 110 Makako bay 22 198 56 142 32 73 102±17 25 250 Study Area 132 1197 600 597 128 49 49±6 5 250

D1 D1 & D2 700 90% 95% 600

500

400

300 90% 95%

200 Individuals 100

0 0 12 24 36 48 60 72 84 96 108 120 132 Surveys

300

243 250

200

150

Individuals 100 85

47 35 35 37 50 25 22 15 16 18 8 7 5 9 0 A B C D AB AC AD BC BD CD ABC ABD ACD BCD ABCD Bay combinations

Figure 3.4 The combination of bays in which individual spinner dolphins have been sighted from September 2010 to August 2011. A = Kauhako Bay, B = Honaunau Bay, C = Kealakekua Bay and D = Makako Bay.

3.4.2 Mark rate of the stock To calculate the proportion of highly distinct individuals using the first independent measure, over 100,000 photographic images were randomly taken of spinner

63 dolphins encountered in groups that comprised > 20 individuals. Of these, 40,715 high-quality photographs contained distinctive and non-distinctive spinner dolphin dorsal fins. From the 40,715 high-quality photographs, 32,519 photographs contained distinctive individuals graded as D1 or D2, and 14,405 were of highly distinctive individuals graded as D1. Therefore, the first independent measure

ˆ estimating the proportion of identifiable individuals ( 1 ) in the stock produced a mark rate of 35%:

ˆ 14,405 1   0.35  0.02 SE 40,715

Of all the 65 groups encountered, a total of 14 groups comprised ≤ 20 dolphins.

There were a total of 168 individual spinner dolphins encountered within these groups. Of these, 132 were distinctive individuals and 60 were highly distinctive D1.

Thus, the second independent measure estimating the proportion of identifiable

ˆ individuals ( 2 ) in the stock produced a mark rate of 36%:

ˆ 60 2   0.36  0.03 SE 168

A Z-test showed that the two estimates were not significantly different (p=0.68), the first value was used in all subsequent adjustments.

3.4.3 Apparent survival and total stock abundance The goodness of fit test to the open model did not suggest the presence of over- dispersion χ2 =27, df=26, p=0.4275, and ĉ=χ2/df=1.04. The abundance estimate of distinct individuals and a range of closed and open models are presented in Table

3-2. In all cases the estimates are very close to 214, the number of distinct animals

64 seen, because the capture probabilities were very high (approx. 0.40 per period), consequently almost all animals were captured by the end of the study. This can also be seen by the flatness of the discovery curve (Figure 3.4). The closed and open models were very similar because I found that the annual apparent survival rate was 0.97±0.05 SE which is not significantly less than 1. As heterogeneity and time variation are strongly indicated for these data I used the estimate based on Mth

221±4.3 SE resulting in a total estimate of 631±60.1 SE (95% CI 524-761) spinner dolphins in the Hawaii Island stock (Table 3-3).

Table 3-2 Highly distinctive (D1) population abundance estimates calculated from open and closed mark recapture models. Open Model Estimate of highly Closed Models Estimate of highly distinctive distinctive individuals (D1) individuals (D1) POPAN φ(t) ρ(t) β(t) 219±2.9 SE M0 214±0 SE Mt 214±0 SE Mh Jacknife 226±5.69 SE Mh Chao 221±4.32 SE Mth Chao 221±4.32 SE

Closed models: M0 = equal capture probability, Mt= variation in capture probability over time, Mh=individual heterogeneity of capture probability and Mth= variation in capture probability over time with individual heterogeneity of capture probability.

Table 3-3 Mark rates and abundance estimates from this study and previous studies. Marked Study Mark rate individuals Total stock estimate This study ˆ ˆ 214 631± 60.1 SE (95% CI 524-761) 1 =35% and 2 =36% (Norris et al., 1994) 20% 192 960 (Ostman, 1994) 29% 677 2,334 (Ostman-Lind et 21.7% and 25.4% 217 1,001 and 855 al., 2004)

3.4.4 Discussion This present study is the first concerted effort to estimate abundance and apparent survival rate estimates for the Hawaii Island spinner dolphin stock. Two key conclusions can be drawn from this study that have implications for the management of this stock. Firstly, my systematic sampling approach and capture/recapture analyses produced an apparent yearly survival estimate of

0.97±0.05 SE for this stock of spinner dolphins. Apparent survival represents the product of true survival and permanent emigration. Therefore, if permanent emigration approaches zero, apparent survival can be representative of true survival. The Hawaii Island spinner dolphin stock is the most genetically distinct of the five island associated stocks (Andrews et al., 2010, Carretta et al., 2013).

Therefore, permanent emigration of the Hawaii Island stock could be assumed to be zero and consequently apparent survival is representative of true survival for the stock. Secondly, my total abundance estimate for this stock 631±60.1 SE (95% CI

524-761) is lower than any previous published estimates, 960 (Norris et al., 1994) ,

2,334 (Ostman, 1994) and 855 – 1,001 (Ostman-Lind et al., 2004) (Table 3-3).

The approaches employed by previous studies to collect photographic identification data to estimate abundance of Hawaiian spinner dolphins along the Kona Coast were not designed for specific capture-recapture models (Ostman-Lind et al., 2004,

Norris et al., 1994, Ostman, 1994). These previous studies used opportunistic photographic identification data retrospectively to estimate the population size. As a consequence, effects due to the inherent characteristics of individual spinner dolphins, and the variation in photographic identification effort throughout the

66 study period weren’t allowed for. The proportion of distinctive individuals in these previous abundance estimates was determined by dividing the ‘total number of identified individuals’ by the ‘mean percentage of individuals identified per group.’

(Norris et al., 1994, Ostman-Lind et al., 2004, Ostman, 1994). Therefore, the resulting abundance estimates did not take into account uncertainty or heterogeneity of individual capture probabilities and the estimates of the proportion of distinct animals were likely biased.

The systematic approach employed by this study was designed specifically to determine spinner dolphin abundance estimates using capture-recapture models.

The consistent data collection effort throughout the study area and period (same bays on the same dates each month) helped to eliminate biases associated with the heterogeneity in capture probabilities due to the variation in individual characteristics. The use of only the highly distinct (D1) individuals helped to eliminate heterogeneity in capture probabilities due to variation in individual distinctiveness, and furthermore reduced misidentification errors of individuals during the identification process. Two independent methods used to determine the proportion of distinctive individuals produced similar results (~36%). These proportions are higher than reported in previous studies in the region (Ostman,

1994, Ostman-Lind et al., 2004, Norris et al., 1994). Advances in digital imaging technology allowed for a greater number of spinner dolphin images to be taken, compared to previous studies that relied on film photography and processing

(Norris and Dohl, 1980, Norris et al., 1994, Ostman, 1994, Ostman-Lind et al., 2004).

Furthermore, advances in technology allowed for high resolution dolphin images,

67 which, in turn, allowed less-distinctive individuals to be identified and included in the catalogue, more so than in previous studies that may have categorised the same quality of photographs as non-distinctive, and as a consequence may have contributed to the low proportion of distinctive individuals identified (Table 3-3).

As my study lasted for one year I expected that I would need to use an open capture-recapture model. However, I found that closed population models gave almost identical population estimates to the open models. I think the population is approximately closed for two reasons. First, the Hawaii Island spinner dolphin stock is genetically distinct from the other island associated spinner dolphin stocks in the

Hawaiian archipelago (Andrews et al., 2010), which is strong evidence for there being little movement in or out of this area. In fact, evidence from recent genetic work indicates that spinner dolphins inhabiting the Kona Coast of Hawaii Island exhibit a greater degree of philopatry than any other spinner dolphin stock in the

Main Hawaiian Islands (Andrews et al., 2010). Second, spinner dolphins are long- lived animals with an estimated annual survival rate of 0.97±0.05 SE and even though a shift in spinner dolphin sighting distribution from the leeward side to the windward side of Hawaii Island has been documented (Norris et al., 1994), this shift was for only one month (a lot shorter than my sampling period), after which it shifted back to the leeward side (Norris et al., 1994) and would be included in our closed population estimate. Therefore no significant new recruits would be expected to enter the population in just one year. As heterogeneity and time variation of capture probabilities are strongly suggested I decided to use the

68 abundance estimate based on the closed model Mth. However, in this case the estimates of all the models are almost identical (Table 3-2).

Bias can also be introduced into abundance estimates from misidentification of individuals. This can occur in two ways: one individual being identified as two individuals (positive bias) and two individuals being identified as one individual

(negative bias). In this study only highly distinctive spinner dolphins were used which helps to mitigate the introduction of bias from individual misidentification.

The photographic identification of individuals for survival rates and abundance estimates was undertaken across the four major resting bays along the Kona Coast but did not survey the entire coastline of Hawaii Island. It is possible that the abundance estimate of this stock underestimates the whole Kona Coast spinner dolphin population. However, I suspect that any potential underestimation is insignificant and that almost all members of the population use these four main resting bays. Earlier studies documented spinner dolphins on the windward side of

Hawaii Island (Norris et al., 1994), however, it is unlikely that this represents prime resting habitat for them given results on habitat preference studies (Thorne et al.,

2012). Earlier studies also observed individual spinner dolphins moving from the north to the south of the Kona Coast encompassing the four main resting bays of this study (Norris et al., 1994, Ostman, 1994, Ostman-Lind et al., 2004). In addition, radio tagged individual spinner dolphins have been observed travelling 20 – 70 km along the Kona Coast (Norris et al., 1994). This study is part of a larger project

(SAPPHIRE) in which spinner dolphin group focal follows were also undertaken

69 outside, and to the north and south of the four resting bays. Individual spinner dolphins observed during these focal follows were also observed in at least one of the four main resting bays during our photographic identification (Tyne, J.A, unpublished data). This suggests our work sampled the entire population of the

Hawaii Island stock (Andrews et al., 2010, Carretta et al., 2013).

3.4.5 Management implications The dolphin-watch tourism industry in Kona has increased over the past 20 years

(Hu et al., 2009), paralleling the dramatic increase in the industry worldwide

(O'Connor et al., 2009). Recent short-term research has suggested that an increase in human traffic inside and outside of the dolphin resting habitats (Ostman-Lind et al., 2004, Courbis and Timmel, 2009, Timmel et al., 2008) resulted in dolphins spending less time in these resting habitats (e.g. Ostman-Lind et al., 2004) and that their resting behaviour was interrupted as a consequence. It has been suggested that spinner dolphins may leave the bays in direct response to human interactions

(Courbis and Timmel, 2009, Timmel et al., 2008, Delfour, 2007, Ostman-Lind, 2008).

However, it was not possible to identify population level effects from these short- term studies. Elsewhere, dolphin-human interactions have had detrimental effects on the focal population. In New Zealand, the resting behaviour of bottlenose dolphins decreased as the number of boats increased in the Bay of Islands

(Constantine et al., 2004) and in Milford Sound (Lusseau et al., 2006). In Shark Bay,

Western Australia, long-term exposure to dolphin-watch vessels caused declines in relative abundance of bottlenose dolphins in an area where boat-based tourism occurred (Bejder et al., 2006b).

Due to growing concerns, the United States National Oceanic and Atmospheric

Administration’s Fisheries Service Pacific Islands Regional Office, in conjunction with the Pacific Islands Fisheries Science Centre, published a Notice of Intent (NOAA,

2005) to prepare an Environmental Impact Statement assessing potential impacts of a proposed rulemaking on human activity. The proposed rule seeks to implement time-area closures in specific spinner dolphin resting habitat to reduce the cumulative exposure to human activity along the Kona Coast of Hawaii Island

(NOAA, 2005).

The rigorous systematic sampling during this study produced the first baseline estimates of abundance and apparent survival rates for the Hawaii Island spinner dolphin stock. These estimates can provide valuable assistance to management agencies, for comparison with historical estimates and to assess the effectiveness of future management actions seeking to mitigate negative human-dolphin interactions. The current estimate of 631 (95% CI 524-761) is substantially lower than previous abundance estimates (Table 3-3). When this estimate is combined with the rigid daily behavioural pattern of spinner dolphins, the genetic distinctiveness of the stock and the ease of human access to the spinner dolphins in their preferred resting habitats, this stock is likely more vulnerable to negative impacts from human disturbance than previously believed.

4 Informing monitoring programs to detect trends in abundance of cetacean populations: a case study focussing on the Hawaii Island spinner dolphin (Stenella longirostris) stock.

4.1 Abstract Cetaceans are long-lived, and their slow growth rates and low fecundity present challenges for developing effective monitoring programs. Rigorous sampling designs are required to produce precise and unbiased population estimates needed to detect changes in abundance. Currently, data are not available on the trends in abundance for any spinner dolphin stock in the Hawaiian archipelago, which is hampering conservation and management efforts to identify potential impacts. Two consecutive estimates were made for the stock size and survival of the Hawaii

Island spinner dolphin (Stenella longirostris). These data were used to determine the ability of three different sampling scenarios to detect a change in abundance by varying the effort, precision, power and interval between abundance estimates for each scenario. The first scenario represented the same sampling effort used to obtain annual abundance estimates for two consecutive years and consisted of monthly surveys, each of 12 days within four spinner dolphin resting bays. Sampling effort was reduced by 50% in Scenarios 2 and 3. Specifically, Scenario 2 consisted of six sampling days per month randomly subsampled from Scenario 1 across all four bays; and Scenario 3 consisted of six sampling days per month in two specific bays.

Scenario 1 produced the most precise annual abundance estimate of 668 ± 62 SE

(95% CI = 556-801; CV=0.09) individuals, consistent with my previously published estimate (Chapter 3; Tyne et al., 2014a). The precision of annual abundance estimates under Scenarios 2 and 3 were slightly less than Scenario 1 with

72 coefficients of variation of 0.11 and 0.10, respectively. With power set at 80% and

95%, and at a precision of 0.09, it would take nine and 12 years, respectively, to detect a 5% change in abundance under Scenario 1. If the change in abundance was a decline, then the population would have decreased by 37% and 46% respectively.

In light of the time duration to detect change and the potential for a large population decline prior to detection, the precautionary principle suggests that the lower power level should be used as an indicator for management intervention.

These results provide management with guidelines for evaluating sampling programs of different intensity and estimate the time required to detect a change

(decline/increase) in the stock. Furthermore, results also provide information for designing sampling programs to evaluate the efficacy of management interventions

(time-area closures) that aim to reduce the number and intensity of human-dolphin interactions.

4.2 Introduction Management decisions for the conservation of marine fauna should be based on sound scientific investigations and rigorous monitoring regimes, particularly for those populations whose viability is threatened (Jaramillo-Legorreta et al., 2007,

Turvey et al., 2007). However, the allocation of scarce funding resources is a perennial problem in conservation biology (Williams and Thomas, 2009, Williams et al., 2011). As a compromise, managers must cut costs of research and lower the burden of proof that a population is in decline before introducing a mitigation approach (Williams and Thomas, 2009, Williams et al., 2011). Under such circumstances long-lived animals are a management challenge, as their slow growth

73 rate and low fecundity can lead to slow recovery from disturbance (Congdon et al.,

1993). These species also require long-term monitoring strategies (Clutton-Brock and Sheldon, 2010, Magurran et al., 2010) to detect small rates of change which are hard to detect which means that populations may be reduced to low levels before management can intervene (Taylor and Gerrodette, 1993, Taylor et al., 2007,

Gerrodette, 1987, Thompson et al., 2000, Wilson et al., 1999).

Population monitoring programs designed to detect change and determine management actions require development to provide precise and unbiased estimates of population parameters (Taylor and Gerrodette, 1993, Taylor et al.,

2007). Therefore, they must be designed to satisfy assumptions of the estimation methods being used to be unbiased and have sufficient sampling effort to produce precise population estimates (Wilson et al., 1999, Thompson et al., 2000). The correct identification of individuals within a population is also important for ensuring that capture-recapture estimates are reliable. Sampling programs that estimate population abundance at different time intervals can inform on trends in abundance (Gerrodette, 1987, Taylor and Gerrodette, 1993, Thompson et al., 2000,

Wilson et al., 1999). The power to detect trends in abundance through time depends on the relationship between the rate of change in the population, the precision of the population estimate (e.g., coefficient of variation) and the acceptable levels of making errors to detect change (Type I (α) and Type II (β) errors). Variations in these parameters can then determine the power to detect trends in abundance of proposed monitoring programs and provide a scientific basis for the level of precaution required to address management issues.

The U.S National Oceanic and Atmospheric Administration (NOAA) has the mandate under the Marine Mammal Protection Act 1972 (MMPA) to protect all cetaceans, seals and sea lions in US waters and the National Marine Fisheries Service (NMFS) and the U.S. Fish and Wildlife Service (USFWS) have the responsibility for assessing the stocks of cetaceans and pinnipeds. The frequency of stock assessments depends on the classification of the stock: strategic stocks require annual stock reviews, while non-strategic stocks require reviews every three years or when new information becomes available (Carretta et al., 2013). A strategic stock is defined under the MMPA as a marine mammal stock “(A) for which the level of direct human-caused mortality exceeds the potential biological removal level; (B) which, based on the best available scientific information, is declining and is likely to be listed as a threatened species under the Endangered Species Act (ESA) within the foreseeable future; or (C) which is listed as threatened or endangered under the

ESA, or is designated as depleted under the MMPA.” Currently, Hawaiian spinner dolphins (Stenella longirostris) are not listed as threatened, endangered or depleted. Furthermore, the levels of serious injury and mortality due to anthropogenic causes does not exceed the Potential Biological Removal (PBR) level for the stock. Therefore are classified as a non-strategic stock (MMPA, 1972).

In Hawaii, spinner dolphins live in small (Chapter 3; Tyne et al., 2014a), isolated stocks with restricted ranges (Andrews et al., 2010) and have evolved a specialised behavioural ecology (Norris and Dohl, 1980). They forage cooperatively offshore at night, and return to sheltered bays to socialise and rest during the day (Norris and

Dohl, 1980, Norris et al., 1994, Benoit-Bird and Au, 2009, Chapter 5; Tyne et al.,

2015). This temporal partitioning of behaviours allows spinner dolphins to maximize their foraging efficiency, while avoiding predation during periods of recovery from the exertions of their night time foraging bouts (Johnston, 2014). However, the resting time of spinner dolphins in sheltered bays overlaps almost completely with the times when humans use the bays for tourism, recreational and subsistence purposes (Heenehan et al., 2015). Some of these activities, in particular nature- based tourism, engage in repeated, close-up encounters with dolphins on a daily basis (Chapter 6; Heenehan et al., 2015).

These close-up encounters may have negative consequences for spinner dolphins, as recent research shows that dolphins are less likely to rest when swimmers approach within 150 m (Symons et al. University of Aberdeen unpublished data). It has been suggested that repeated approaches to dolphins interrupt their resting behaviour and, as in response, dolphins spend less time in resting bays (Courbis and

Timmel, 2009, Timmel et al., 2008, Delfour, 2007). Thus, human activities may adversely impact on the overall resting budget of dolphins (Chapter 5: Tyne et al.,

2015), which, in turn, may cause rest deprivation and affect energetic budgets

(Williams et al., 2006) and ultimately impact their reproductive success (National

Research Council, 2005). As such, concerns have been raised regarding the possible effects that human-dolphin interactions are having on the spinner dolphins (Danil et al., 2005, Courbis and Timmel, 2009). Consequently, NOAA are looking to implement a management strategy to reduce the number and intensity of human- dolphin interactions in Hawaii (NOAA, 2005). One strategy under consideration is

76 the implementation of time-area closures in the preferred resting habitats of the dolphins during their resting periods (NOAA, 2005). Currently, data are not available on the trends in abundance for any spinner dolphin stock in the Hawaiian archipelago (Carretta et al., 2013). This lack of scientific data is hampering conservation and management efforts to identify potential impacts on Hawaiian spinner dolphin stocks.

In the current study, I first provide a second consecutive annual abundance and survival estimate of the Hawaii Island spinner dolphin stock to examine the variation between years. Next, three scenarios with different levels of sampling effort were designed based on the systematic approach employed in Chapter 3

(also Tyne et al. (2014a)). Finally, the ability of sampling scenarios to detect trends in abundance was investigated by varying sampling effort, rate of change in abundance, precision, power and interval between annual abundance estimates.

The results provide management with guidelines for evaluating sampling programs of different intensity and estimate the time required to detect a trend

(decline/increase) in the population. The results also provide fundamental information for designing sampling programs to evaluate the efficacy of management interventions (time-area closures) that are designed to reduce the number and intensity of human-dolphin interactions.

4.3 Materials and methods 4.3.1 Fieldwork The Hawaiian archipelago is located approximately 3,200 km southwest of mainland United States. Hawaii Island is the largest, youngest and most southerly of the main Hawaiian Islands. On the leeward (west) side of the island is the Kona

Coast, where four important dolphin resting bays are located: Makako Bay,

Kealakekua Bay, Honaunau Bay and Kauhako Bay (Figure 4.1); (Norris et al., 1994,

Thorne et al., 2012, Chapter 3; Tyne et al., 2014a, Chapter 5; Tyne et al., 2015).

Figure 4.1 Map of the study area illustrating the four spinner dolphin resting bays, Makako Bay, Kealakekua Bay, Honaunau Bay and Kauhako Bay, along the Kona Coast of Hawaii Island.

4.3.2 Sampling design From September 2010–August 2012, boat-based photographic-identification was carried out in four resting bays of the Hawaii Island spinner dolphin stock using the systematic sampling design presented in Chapter 3; Tyne et al. (2014a). Each bay was sampled on the same dates each month, regardless of whether dolphins were present or absent, thus providing consistent and even effort throughout the study period and area. This design (Scenario 1) consisted of 12 consecutive sampling days each month for each of the two study years. Two additional sampling regimes of reduced intensity (Scenarios 2 and 3) were evaluated against Scenario 1.

Abundance was estimated for each year and was based on twelve months of sampling effort for each scenario.

4.3.3 Sampling effort Scenario 1 consisted of monthly sampling over 12 consecutive days of photo- identification covering four bays: two days in Makako Bay, four days in Kealakekua

Bay, two days in Honaunau Bay and four days in Kauhako Bay. Scenarios 2 and 3 each had a 50% reduction in sampling effort (i.e. six days of sampling each month).

The sampling effort in each of the three scenarios was:

 Scenario 1 – 12 sampling days per month across four bays.

 Scenario 2 – six sampling days per month, spread across the four bays, with

the days chosen by randomly selecting half the number of days from each

bay in Scenario 1.

 Scenario 3 – six sampling days per month, across two bays where dolphins

were encountered most frequently (two days in Makako Bay and four days

in Kealakekua Bay).

4.4 Analyses 4.4.1 Capture-recapture All photographs were graded according to photographic quality and distinctiveness to minimise the introduction of bias and to reduce misidentification (Urian et al.,

2015). Only highly distinctive (D1) fins in photographs of excellent and good quality were included in the capture-recapture analyses (Urian et al., 2015, Gowans and

Whitehead, 2001, Nicholson et al., 2012). A capture was defined as a photograph of sufficient quality of an individual dolphin’s distinctly marked dorsal fin. Capture histories corresponded to whether or not an individual dolphin was “captured” or

“recaptured” during a sampling occasion. This information was compiled for each individual (calves excluded) after a photo-grading process. See Chapter 3; Tyne et al. (2014a) for more detail on the photo-grading process.

For both years, open and closed capture-recapture models in the program MARK

(White and Burnham, 1999) were applied to the photo-identification data to estimate stock size, variability and evaluate the goodness-of-fit of the models. See

Chapter 3; Tyne et al. (2014a) for full details on modelling approach. The POPAN approach is able to estimate probabilities of entry (immigration) and probabilities of exit (emigration and mortality), to and from the study area between sampling occasions (Schwartz and Arnason, 1996). Under Scenario 2, capture histories of

81 individual dolphins were created based on six days subsampled 100 times from

Scenario 1. Capture-recapture modelling was then applied to each of the 100 spinner dolphin capture histories. Annual abundance estimates, apparent survival and overdispersion were each calculated from the mean of the 100 abundance estimates, survival rates and overdispersion (ĉ=χ2/df) for each year. Standard errors

(SE) for both the annual abundance estimates and survival rates were then calculated from the standard deviation of the empirical sampling distributions of the estimates.

All capture-recapture models make the following assumptions (Williams et al.,

2002a): 1) marks are not lost during the study; 2) marks are correctly recognised on recapture; 3) individuals are instantly released after being marked; 4) intervals between sampling occasions are longer than the duration of a sample; 5) all individuals observed during a given sampling occasion have the same probability of surviving until the next one; 6) study area does not vary; and 7) homogeneity of capture probabilities, i.e. that all animals in a sampling occasion have equal probability of being captured. These assumptions are relaxed for certain models that allow heterogeneity in the capture probabilities. See Chapter 3; Tyne et al.

(2014a) for more detail on the methods used to estimate abundance, mark rate and total stock size. To determine whether data were overdispersed (when the variance is greater than the mean (Cox, 1983)), the inflation factor (ĉ) was calculated for the abundance estimates (Anderson et al., 1994) and Quasi-likelihood adjustments were applied to models take over-dispersion into account.

4.4.2 Detecting change in abundance Detecting significant change in abundance over time requires that the null hypothesis (H0) of no change in abundance is rejected. The probability of detecting a significant change in abundance when one doesn’t exist, i.e., the Type I error, is generally set at α = 0.05, which is policy in the United States (Taylor et al., 2007).

However, even when H0 is not rejected, it is possible that the abundance has changed, i.e., a Type II error is present. Power analysis can be used to identify the ability of sampling regimes to adequately detect trends in abundance and to minimise the probability of Type II errors occurring (Gerrodette, 1987). The ability of three scenarios to detect change in abundance was investigated using

Gerrodette’s (1987) inequality model:

2 3 2 2 푟 푛 ≥ 12퐶푉 (푍훼/2 + 푍훽)

Where 푟 = the rate of population change, 푛 = the number of estimates, 퐶푉 = the coefficient of variation of the abundance estimate (a measure of precision), 푍훼 = normal deviate corresponding to the probability of making a Type I error, 푍훽= normal deviate corresponding to the probability of making a Type II error, α = the one-tailed probability of making a Type I error and β = the probability of making a Type II error. The probability of making a Type I error (α) was set at 0.05, and the r probability of making a Type II error (β) was set at 0.05 (i.e., power = 1 – β = 0.95) and 0.20 (power = 0.80).

The mean CVs obtained from the two annual abundance estimates from each sampling scenario were used to investigate the number of years required to detect

83 varying rates of change (1 to 20%) in abundance at 80% and 95% power. A range of

CVs (5% to 20%) were then used to determine the number of years required to detect 5% and 10% change in abundance at 80% and 95% power. Finally, I examined the number of years it would take to detect a 5% change in abundance under the three scenarios.

4.5 Results 4.5.1 Effort and summary statistics A total of 276 days (> 2,350 hours of on-water effort) of photo-identification was carried out in the four bays between September 2010 and August 2012.

Approximately 4,000 h of effort was required to identify and grade the individual spinner dolphins from the more than 200,000 images. More than 64,500 of these images were of sufficient quality to be added to a photo-identification catalogue in which 235 highly distinctive individuals (D1) were identified. The identification of new individuals reached a plateau before the end of the two-year study period

(August 2012, on sampling day 276), with 211 dolphins (90%) identified after 114 sampling days (July 2011) and 223 (95%) after 187 sampling days (February 2012,

Figure 4.2).

250 Year 1 Year 2

200

90% 95%

150

100 Individuals

0 0 12 24 36 48 60 72 84 96 108 120 132 144 156 168 180 192 204 216 228 240 252 264 276 Sampling day

4.5.2 Estimates of apparent survival rate and stock abundance Apparent survival is the probability of surviving and staying in the study area and represents the product of true survival and permanent emigration. The two independent yearly apparent survival rates estimated under Scenario 1 were identical (0.97, Table 4-1). With the 50% reduction in sampling, the estimated survival rates for Scenario 2 and 3 were higher in 2011 than 2012 (0.2 and 0.08 higher for Scenarios 2 and 3, respectively). The abundance estimates were higher in

2012 than 2011 for all three scenarios. Although the abundance estimates were more precise from Scenario 1 (CV = 0.09), there was very little difference in precision between the three scenarios (Scenarios 2 and 3, CV = 0.10 and 0.11). The goodness-of-fit measure (ĉ=χ2/df) suggested that the data were over-dispersed for two of the six estimates (Scenarios 1 and 2 in 2012) (Table 4-1).

2: 6 d, 4 bays 2011 0.88 ± 0.10SE 552 ± 57 (450-678) 1.4 0.11 2012 0.68 ± 0.09SE 632 ± 62 (522-766) 1.7 0.11

3: 6 d, 2 bays 2011 0.92 ± 0.07SE 557 ± 56 (458-678) 1.1 0.10 2012 0.84 ± 0.08SE 659 ± 64 (545-796) 1.2 0.10

4.5.3 Detecting change in abundance From the three sampling scenarios, the number of abundance estimates required to detect a change in the dolphin stock decreased as the rate of change increased

(Figure 4.3). For example, at a CV of 0.10 and a 5% rate of change at 95% power, nine abundance estimates are needed to detect change, compared with five abundance estimates to detect a 10% change (Figure 4.3). Furthermore, as the precision decreased (i.e., CV increase), the time necessary to detect a change increased (Figure 4.4).

1 – β = 0.95 S1 = 0.09

10 S2 = 0.10 S3 = 0.11

8 1 – β = 0.80 S1 = 0.09 S2 = 0.10 6 S3 = 0.11

2 Annual abundance estimates abundance Annual

0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Rate of population change (%)

Yearly at 5% Yearly at 10% 3 Yearly at 5% 3 Yearly at 10% Yearly at 5% Yearly at 10% 3 Yearly at 5% 3 Yearly at 10% 25

10 Years to detection to Years 5

0 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Level of precision (coefficient of variation)

Figure 4.4 Predicted time it would take to detect an annual change of 5% and 10% with varying levels of precision (coefficient of variation, CV) using monitoring intervals of one year and three years. Type I error (α) probabilities were set at 0.05 and Type II error (β) probabilities were set at power = 1 – β = 0.80 (grey) and power = 1 – β = 0.95 (dark).

Of the three scenarios, annual abundance estimates from the most intensive sampling scenario (Scenario 1) were the most precise (CV = 0.09; Table 4-2). Under

Scenario 1, it would take seven annual abundance estimates over six years to detect a 5 % annual change (decline/increase) with 80% power. Under the same scenario, it would take eight annual abundance estimates over seven years to detect a 5 % change with 95% power (Table 4-2). As the time interval between abundance estimates increased from one to three years, the number of abundance estimates required to detect a change decreased, but the time taken to detect a change increased (Table 4-2). This is due to the increase in the effective percentage change in abundance per interval (Gerrodette, 1987, Wilson et al., 1999). To detect an 88 annual 5% change at 80% and 95% power, it would take four and five abundance estimates (at three year intervals), over nine and 12 years respectively. If the change was a continuous decline, the abundance would have declined by 37% and

46% by the time of detection, equivalent to a decline from 668 ± 62 SE (95% CI 556-

801) to 433 and 372. If the change in abundance was an increase, the abundance estimate would have increased by 55% (1,035) and 80% (1,202) at the time of detection.

Table 4-2 Number of annual abundance estimates, effective percentage change, years to detection, total percentage change at detection, at varying degrees of precision, to detect an annual 5% change (decline/increase) in abundance between one, two and three year monitoring intervals based on Scenario 1 = 12 days of sampling covering four bays (two days in Makako Bay, four days in Kealakekua Bay, two days in Honaunau Bay and four days in Kauhako Bay), Scenario 2 = six days randomly subsampled from the 12 days covering four bays (one day in Makako Bay, two days in Kealakekua Bay, one day in Honaunau Bay and two days in Kauhako Bay) and Scenario 3 = six days covering two bays (two days in Makako Bay and four days in Kealakekua Bay). Probability of a Type I Error (α = 0.05) and a Type II Error (1 – β = 0.95 and 1 – β = 0.80). CV = coefficient of variation. Power CV Monitoring Annual Effective % Effective % Total % Total % interval abundance decline per increase per Years to decline at increase at (years) estimates interval t interval t detection detection detection (t) (n) (0.95t – 1) (1.05t – 1) (t(n-1)) (0.95t(n-1) – 1) (1.05t(n-1) – 1) 0.80 0.09 1 7 -5 5 6 -26 34 0.09 2 5 -9.8 10.3 8 -34 48 0.09 3 4 -14.3 15.8 9 -37 55 0.95 0.09 1 8 -5 5 7 -30 41 0.09 2 6 -9.8 10.3 10 -40 63 0.09 3 5 -14.3 15.8 12 -46 80 0.80 0.10 1 7 -5 5 6 -26 34 0.10 2 5 -9.8 10.3 8 -34 48 0.10 3 4 -14.3 15.8 9 -37 55 0.95 0.10 1 9 -5 5 8 -34 48 0.10 2 7 -9.8 10.3 12 -46 80 0.10 3 6 -14.3 15.8 15 -54 110 0.80 0.11 1 8 -5 5 7 -30 41 0.11 2 6 -9.8 10.3 10 -40 63 0.11 3 5 -14.3 15.8 12 -46 80 0.95 0.11 1 9 -5 5 8 -34 48 0.11 2 7 -9.8 10.3 12 -46 80 0.11 3 6 -14.3 15.8 15 -54 110

4.6 Discussion

In this chapter I aimed to estimate the abundance and survival of Hawaii Island spinner dolphins in consecutive years and model the ability of different sampling scenarios to detect change in abundance over time. Two main findings emerged from this research. Firstly, the additional survival rate and abundance estimate of the Hawaii Island spinner dolphin stock are virtually identical to those from the first year (Chapter 3; Tyne et al., 2014a), suggesting that the sampling design in combination with capture-recapture models is rigorous and that the estimates from the first year are reliable. Secondly, although there was little difference in the precision between sampling scenarios, the sampling effort affects the ability of the sampling regime to detect a trend in abundance over time. These results provide a scientific basis for the level of precaution required to address management issues.

4.6.1 Estimates of survival rates and abundance The systematic sampling approach developed in Chapter 3; Tyne et al. (2014a) was designed specifically to estimate the abundance of the Hawaii Island spinner dolphin stock. In the present study, it was used as the basis for investigating the ability of three different sampling scenarios to detect a change in abundance over time. The most intensive sampling effort (Scenario 1 with 12 days each month in four bays) produced the most precise annual abundance estimates. Both of the other sampling scenarios, where sampling effort was reduced by 50%, gave lower estimates of apparent survival rates and less precise abundance estimates.

However, the standard errors of Scenarios 2 and 3 were still similar to those of

Scenario 1. This is partly as a consequence of the mean number of days dolphin 91 groups were encountered from Scenario 1 being the same number of days as half the sampling effort reflected in Scenarios 2 and 3. The annual abundance estimates in this study and in Chapter 3; Tyne et al. (2014a) are lower than any previous estimates (Ostman, 1994, Ostman-Lind et al., 2004, Norris et al., 1994). However, caution should be had when making comparisons as previous research efforts were not specifically designed to estimate abundance. Consequently, it is not possible to assess the current trend in population size of the Hawaiian spinner dolphins, except to acknowledge that the stock is smaller than previously estimated.

4.6.2 Monitoring changes in dolphin abundance over time The ability to confidently detect trends in abundance over time is critical when making conservation decisions (Taylor and Gerrodette, 1993, Thompson et al.,

2000, Wilson et al., 1999). The current trend in abundance of the Hawaii Island spinner dolphins is uncertain. Degrees of precision, power, sampling effort and interval between abundance estimates were varied to evaluate the ability of three sampling scenarios to detect change in abundance. As the sampling effort increased, so did the precision of the abundance estimates, and thus changes in abundance could be detected earlier. However, even at the most intensive sampling effort (Scenario 1), with annual surveys and abundance estimates assessed every three years at 95% power, the population of spinner dolphins may have declined by 46% at the time when a significant trend is detected (i.e. from 688 to 372 individuals). This rate of decline is approximately 50% over 15 years, a rate that has been defined as precipitous (Taylor et al., 2007) and could lead to the stock being classed as ‘depleted’ under the MMPA (Taylor et al., 2007).

4.6.3 Applications for monitoring Hawaiian spinner dolphins are classed as a non-strategic stock under the MMPA and under the current legislation, their abundance is assessed once every three years (Carretta et al., 2013). NOAA are considering a management approach to reduce the number and intensity of human-dolphin interactions in preferred resting habitat of spinner dolphins, including the introduction of time-area closures of the four spinner dolphin resting bays from this study (NOAA, 2005). If time-area closures were introduced, a monitoring program to detect trends in dolphin abundance would help evaluate the effectiveness of this management strategy.

If the rate of change in abundance is small, then precision will have a large effect on the time needed to detect a change (Figure 3; see also Thompson et al., 2000,

Wilson et al., 1999, Taylor et al., 2007). The sampling effort for the most precise sampling scenario in this study required a significant investment of time and field personnel and for the processing of the dolphin photo-identification images. These resources were only made possible through the presence of a dedicated PhD student and large numbers of research assistants and significant financial and logistical support. The chronic underfunding of monitoring programs (Williams and

Thomas, 2009, Williams et al., 2011), requires the careful consideration of resources required to conduct an adequate follow-up photographic-identification study of the spinner dolphin stock. Other considerations in the design of the program include the rate of change in abundance and the confidence of detecting significant change. By increasing power (confidence) to detect a change, both the number of annual abundance estimates and study duration required will increase

(Taylor et al., 2007, Gerrodette, 1987). The time taken to detect a decline can be critical for small, genetically isolated stocks, such as those of the Hawaii Island spinner dolphins (Thompson et al., 2000, Wilson et al., 1999). A precipitous decline in abundance may have significant, negative biological consequences for this spinner dolphin stock. Consideration of these factors is paramount, especially in determining the level of precaution required to address management issues.

Most coastal cetacean populations are exposed to a range of human activities

(Chapter 6; Jefferson et al., 2009, Pirotta et al., 2015, Christiansen et al., 2010,

Lusseau et al., 2006, Lusseau et al., 2009, Bejder et al., 2006b, Constantine et al.,

2004), and each population has a diversity of issues to be considered when developing and implementing an effective management strategy. This study demonstrated that power analyses should be an integral part of a population management strategy. The approach presented here is widely applicable and can inform monitoring programs for populations (cetaceans and other fauna), where individuals can be identified and estimates of population size can be made through mark-recapture analyses. For example, in north Western Australia, oil and gas resources are in great demand, infrastructure is expanding rapidly and shipping activity is increasing dramatically (Allen et al., 2012, Bejder et al., 2012). Despite this rapid development, little information is available on the characteristics of cetacean populations in the region, which is hampering the assessment of their conservation status, e.g. the Australian snubfin dolphin (Orcaella heinsohni) and

Australian humpback dolphins (Sousa sahelensis) (Brown et al., 2014, Allen et al.,

2012, Bejder et al., 2012). One of the biggest problems with a mark-recapture

94 approach to estimate abundance is assigning the abundance estimate to a particular geographic location. A number of state marine parks have been proposed to protect important cetacean habitats in the region, e.g. the Camden

Sound Marine Park and Roebuck Bay Marine Park (Department of Parks and

Wildlife, 2013). The efficacy of these parks could be evaluated based on the approach presented here, using spatially explicit capture-recapture methods

(Williams et al., 2014, Borchers and Efford, 2008) to assist in assigning abundance estimates to a geographical area to be protected. More broadly, there are currently 553 globally-established, marine mammal protected areas with an additional 168 being considered for declaration (Hoyt, 2011, Chapter 2; Tyne et al.,

2014b). This study provides a template for developing sampling regimes and for testing efficacy of marine protected areas that aim to increase cetacean abundance.

5 The importance of spinner dolphin (Stenella longirostris) resting habitat: Implications for management

5.1 Abstract Linking key ecological characteristics with animal behaviour is essential for identifying and protecting important habitats that support life functions. Spinner dolphins display a predictable diurnal behavioural pattern where they forage offshore at night and return to sheltered bays during daytime to rest. These bays, which are also subject to considerable use by humans, have long been recognised as key habitats for this species, however the extent to which dolphins rely on specific characteristics of these habitats for rest has not been quantified. A novel integration of boat-based and land-based group focal follow sampling regimes and three gradient boosting Generalised Additive Models (GAMs) were developed to identify habitat features that contribute to the occurrence of resting spinner dolphins in coastal waters off Hawaii Island. Two ‘in-bay’ models used data collected within-in bays and a third ‘coastal’ model (near-shore, outside of bays) used data collected both inside and outside of bays.The coastal model identified that spinner dolphins were unlikely to rest outside sheltered bays. In-bay models showed that dolphins rested throughout daylight hours within bays with a peak resting period between 10 am to 2 pm. The models also identified bottom- substrate-type as an important predictor of rest. Pseudo R2 values of 0.61 and 0.70 for the in-bay models and 0.66 for the coastal model showed that these models provided a good fit to the behavioural data for the occurrence of resting spinner dolphins. To date, studies evaluating spinner dolphin resting habitat have focussed on areas inside bays only. Here, I combined data collected inside and outside bays, and illustrate that should resting spinner dolphins be displaced from resting bays, 96 they are unlikely to engage in rest behaviour elsewhere. Results provide further information on the importance of bays as important habitat for resting spinner dolphins. To mitigate the disturbance from human interactions during important rest periods, I recommend that management keep the spinner dolphin resting areas free from human activities. My quantitative approach where models explicitly link behaviour with habitat characteristics is applicable to identify important habitats for protection of other taxa.

5.2 Introduction Animals choose between behavioural activities across time and space (habitat) to optimally exploit resources such as prey (Heithaus and Dill, 2002) and shelter (Lima,

1998) and to avoid predators (Heithaus et al., 2008). The costs and benefits associated with choosing one behaviour over another shapes the evolution of behavioural strategies which, in turn, influence individual fitness (Lima and Dill,

1990, Dill, 1987). Identifying relationships between behaviour and ecology is challenging as they vary over space and time (Dill, 1987). Spatially, these relationships exist over distances varying from a few metres to thousands of kilometres and, temporally, over hours to months (Corkeron et al., 2001, Armstrong et al., 2013).

Key habitats may function as critical for population viability by providing optimal resources (e.g. shelter, prey)(Dill, 1987). In addition to coping with environmental variations and resource availability within key habitats, many animals must also cope with the consequences of human disturbance, including climate change

(Johnston et al., 2012), de-forestation (Johnson et al., 2004), development (Holdo et al., 2011), overfishing (Worm et al., 2013), fisheries by-catch (Allen et al., 2014) and tourism (Bejder et al., 2006b, Lusseau et al., 2006, Constantine et al., 2004). To quantity potential negative impacts of human disturbance on animal populations, important areas for population viability can be identified by linking habitat characteristics to either animal presence (Goetz et al., 2012) and/or important life functional behaviours (Lusseau and Higham, 2004). Critical habitats can be defined as areas where animals exhibit important behaviours such as foraging, breeding, nursing, socialising and resting (Hoyt, 2011, Lusseau and Higham, 2004).

Spinner dolphins (Stenella longirostris) in Hawaii exploit sheltered bays to socialise and rest during the day, following a night of cooperative foraging in open-water foraging grounds (Norris et al., 1994, Benoit-Bird and Au, 2009). This temporal partitioning of behaviours allows spinner dolphins to maximize their foraging efficiency while minimizing predation risk during periods of rest (Norris et al., 1994,

Benoit-Bird and Au, 2009). This predictable behavioural pattern makes spinner dolphins vulnerable to perturbation during rest periods, especially if they are unable to compensate for disrupted resting periods (Johnston, 2014). The Hawaii

Island associated spinner dolphin population may be especially vulnerable to human disturbance because their resting habitats are subject to considerable human activity (Heenehan et al., 2015), the population is small (Tyne et al., 2014a) and genetically-isolated from other populations (Andrews et al., 2010).

Specifically, sheltered bays used by spinner dolphins to rest are also used by people for recreational and commercial purposes (Heenehan et al., 2015). Spinner dolphin resting periods are interrupted or truncated by exposure to human activity (Courbis and Timmel, 2009), and they are less likely to rest when swimmers are within 150 m

(J. Symons et al. University of Aberdeen unpublished data).

The US National Oceanic and Atmospheric Administration (NOAA) is mandated to protect all cetaceans, seals and sea lions in US waters, including the protection of

“essential habitat, including rookeries, mating grounds, and areas of similar significance for each species of marine mammal from the adverse effect of man’s actions.” (MMPA, 1972). Evidence suggests that protected areas can be effective for marine mammal conservation if of appropriate size (Gormley et al., 2012, Edgar et al., 2014). NOAA is considering several management strategies to mitigate the negative effects of human-spinner dolphin interactions (NOAA, 2005), including the use of area closures to reduce the number and intensity of interactions during dolphin resting periods. This strategy proposes to identify specific areas that are important to the population’s survival and restricting human access (Tyne et al.,

2014b).

I combined boat-based and land-based group focal follow data to determine the resting behaviour of spinner dolphins across a range of available habitats, inside four bays and along open coastline adjacent to the bays. My specific objectives were to 1) identify key habitat factors that contribute to the likelihood of spinner

99 dolphin rest, and 2) determine time periods that the spinner dolphins are most likely to rest within these habitats.

5.3 Materials and Methods This study took place in the waters off the Kona Coast (between 19 55° 37’N, 155

53° 45’W and 19 21° 40’N, 155 53° 31’W) on the leeward side of Hawaii Island

(Figure 5.1). Here, spinner dolphins are often observed within four bays during daylight hours (Makako Bay, Kealakekua Bay, Honaunau Bay and Kauhako Bay,

Figure 5.1) (Norris et al., 1994, Thorne et al., 2012, Chapter 3; Tyne et al., 2014a).

Land-based and boat-based group focal follows were undertaken to collect behavioural data on dolphin groups both inside and outside (within 1km of the coastline) these four sheltered bays.

100

5.3.1 Group focal follows Established group focal follow protocols were employed to collect positional and behavioural information on spinner dolphins during daylight hours from both boat- based and land-based platforms (Table 5-1). Group focal follows often consist of a combination of continuous and scan sampling procedures (Altmann, 1974, Mann,

1999). Continuous sampling was used to obtain all occurrences of specific dolphin behavioural events. Instantaneous scan sampling was used to record predominant group behavioural activity at regular intervals, e.g. resting and socialising (Altmann,

1974, Mann, 1999, Bejder et al., 2006a).

Table 5-1 Definitions of spinner dolphin group activities, adapted from Norris et al (1994). Predominant group activity: Rest: Characterised by tight group, slow speed moving back forth generally over sandy substrate or meandering movement. Individuals typically take multiple breaths; synchronous group diving; changing direction whilst underwater; spend long periods of time submerged (1.5-3 minutes). Social: Characterised by regular, consistent, aerial behaviours within the group; little time is spent below the surface; dives are brief. Travel: Characterised by regular and consistent spatial progress with respect to the bottom (in practice surface and shoreline features), i.e. directed swimming that is roughly straight. Travel speed is typically >3.2km/hr. Forage: Spinner dolphins do not forage during the day, but only at night offshore (Norris et al., 1994, Beniot-Bird and Au, 2003), thus this behavioural state was not observed.

A spinner dolphin group was defined using a 100-m chain rule: when A is within 100 m of B and B is within 100 m of C but A and C are more than 100 m apart, A and C are considered to be in the same group (modified from Smolker et al., 1992).

Continuous sampling was employed to record all occurrences of fission-fusion

102 events by individuals of the focal group. A fission event was defined as when an individual, or part of the group, moved beyond the 100 m chain and a fusion event was defined as when they joined the focal group by moving within the 100 m chain.

Instantaneous scan sampling protocols were employed at 5 (boat-based) and 10 min (land-based) intervals to record the predominant group activity of the majority

(> 50%) of individuals in the focal group, group size (minimum, best and max group size estimates) and dolphin group location. A minimum of four people continuously tracked spinner dolphins during a group focal follow. An observation period was terminated when the behaviour of the dolphins could no longer be reliably determined because of events, such as poor visibility, dolphins moved out of range or dolphins split into too many groups.

5.3.2 Land-based group focal follows Land-based group focal follows were undertaken from high vantage points overlooking Kealakekua Bay (139 m, 19° 28' 59.7", 155° 55' 51") and Kauhako Bay

(57 m, 19° 22' 44.5", 155° 53' 47.5"). A SOKKIA DT5-10 digital theodolite equipped with a 30 x lens was connected to a laptop computer running the computer program PYTHAGORAS (Gailey and Ortega-Ortiz, 2002). PYTHAGORAS used data on the theodolite’s position, height above sea level (including tidal fluctuations) and a reference point used for zeroing, to convert theodolite positional fixes of target objects (dolphin groups, boats, swimmers, kayaks) into latitudinal and longitudinal coordinates (Würsig et al., 1991). At the start of tracking (usually between 06:30 am and 07:00 am) and at hourly intervals, a scan was carried out to fix the position of all vessels, swimmers and kayakers in the bay. A positional fix was taken at the

103 centre of the focal dolphin group every 5 min and the predominant group behaviour was recorded every 10 min. Theodolite observations were not carried out at Honaunau and Makako bays because the elevation was too low to reliably track dolphins from land.

5.3.3 Boat-based group focal follows A 7m research vessel equipped with a 90 hp four stroke outboard, left dock at sunrise, with a minimum of four researchers on board to look for spinner dolphin groups moving inshore from their night time foraging grounds. The vessel travelled within 1 km of the coast until spinner dolphins were located. Continuous and instantaneous scan sampling were then initiated to document group behavioural information. The predominant group activity was recorded every 5 min. Using

Logger software (IFAW, 2000), GPS coordinates of the focal group were recorded at

30 s intervals. Fission-fusion events were recorded continuously during the sampling period. To minimise the impact of the presence of the research vessel on the spinner dolphins during group focal follows, the vessel was maintained at a distance of approximately 100 m from the focal group and the vessel was positioned behind and to the side of the group. All care was taken to minimise disturbance and changes in the dolphin group behaviour induced by the presence of the vessel.

5.3.4 Bathymetric and benthic data Bathymetric and benthic habitat data were produced using high-resolution satellite, LiDAR (light detection and ranging) and acoustic SONAR (sound navigation

104 and ranging). These data were downloaded from the Centre for Coastal Monitoring and Assessment website (http://coastalscience.noaa.gov/) with a resolution of 50 x

50 m2. Focal follow data were converted from latitude and longitude projection to the Universal Transverse Mercator (UTM) coordinate system and overlayed upon the bathymetric and habitat maps using ArcGIS 10.1. Maps were overlayed on a grid divided into 50 m2 cells. Thereafter, the corresponding depth, distance from shore, habitat type, position, time of day for each 10 min dolphin group behavioural sample in each cell, was extracted and exported to an Microsoft Access database.

The land-based focal follow protocol collected data every 5 min, however, alternate data points were removed so that these data matched the boat-based protocol. For some behavioural observations, habitat type and depth could not be determined from the remotely sensed bathymetric and benthic habitat data. These data were removed from the modelling process.

5.3.5 Modelling approach I employed a method for addressing the complexities of non-linear and auto- correlated ecological data commonly referred to as component-wise gradient boosting (Friedman et al., 2000). Component-wise gradient boosting is a machine learning method for obtaining statistical model estimates via gradient descent techniques (Friedman et al., 2000, Hastie et al., 2009). Binomial component-wise gradient-descent Boosted GAMs, hereafter referred to as Boosted GAMs, were used to explore the relationship between resting spinner dolphins and a number of environmental, spatial and temporal factors inside and out of the four sheltered bays along the Kona coast of Hawaii Island. Boosting allows an integrated method

105 for fitting constrained models with multiple sources of variation, including smooth spatial interdependence by using spatial splines, as well as other non-linear functions of environmental covariates. I fitted a spatial spline in order to fit variance that is purely spatial and not related to the other covariates which is considered by

Hothorn et al. (2010b) as a way to partially address spatial autocorrelation. Spinner dolphin behaviours were collapsed into a binominal response for behaviour states: i.e., resting and non-resting (1 and 0, respectively). The R software (R Core Team,

2014), the boosting package ‘mboost’ (Hothorn et al., 2010a) and ‘rodbc’ (Ripley,

2013) package for database access were used to develop the Boosted GAMs.

5.3.6 Base-learners Each predictor was added to the models via effect functions known as base- learners (for details see Hofner, 2011). Bootstrapping and cross-validation was used to determine the optimal number of boosting iterations to provide maximum prediction accuracy and, in combination with automatic predictor selection, to prevent over-fitting (Hofner et al., 2014). Stability selection was used to determine the probability of predictor selection during the model fitting process (Meinshausen and Bühlmann, 2010).

Three models were developed to investigate the relationship between resting spinner dolphin groups and environmental, spatial and temporal factors. Two models used data collected inside Kauhako Bay and Kealakekua Bay (‘in-bay models’), respectively, while the third model (‘coastal model’) used data collected both inside and outside of four sheltered bays: Kauhako Bay, Honaunau Bay,

106

Kealakekua Bay and Makako Bay. Models developed for data collected inside

Honaunau Bay and Makako Bay were unable to converge due to insufficient data from inside the bays and they were therefore not included in further analysis. The two in-bay models were implemented using six base-learners, while the coastal model included a seventh base-learner and an interaction (Table 5-2). A maximum number of 1,000 iterations was applied to each bootstrap. The optimal number of iterations was then determined and the base-learners that contributed to the model fit in order of importance were identified from their selection frequencies and probabilities of selection. Marginal function plots were used to illustrate the relationship between the response and the predictor variables after accounting for all other covariates (Maloney et al., 2012).

Table 5-2 List of base-learners used in the three Boosted Generalised Additive Models (GAMs) to explore relationships between resting spinner dolphins (resting or non-resting) and environmental, spatial and temporal factors. Single bay models = Kealakekua Bay and Kauhako Bay; coastal model includes all bays and coastal waters. Model Base learners All models (single Substrate (sand, aggregate coral, rock/boulders) bay and coastal Depth (m) – mean centred model) Distance from shore (m) – mean centred Spatial position (converted to UTM) Time-of-day (morning: 6am-10am; mid-morning: 10am-2pm; afternoon: 2pm-6pm) Behavioural state during previous scan observation (resting/not-resting) coastal model Inside or outside bays

For each model, 50 bootstrap samples from the full data set were used as a training data set to which gradient boosting was applied, from which the pseudo R2

107

(Nagelkerke, 1991) was estimated. Pseudo R2 estimates are a generalisation of the

R2 coefficient of determination, used to determine the proportion of variance in the dependent variable shown by the independent predictor variables in linear regression models (Nagelkerke, 1991). The higher the pseudo R2 value the better the model fit. The prediction accuracy of each GAM was tested using bootstrap cross validation (Hofner, 2011). Predicted responses were back-transformed to their original measurement scales and used to produce predicted probability maps of resting spinner dolphin groups in Kauhako Bay and Kealakekua Bay.

5.4 Results 5.4.1 Effort and sample sizes A total of 488 hours of group behavioural data were collected during boat-based

(n=121 h) and land-based (n=367 h) focal follows, with 402 h of observations

(82.4%) made inside bays and 86 h outside of bays (Table 5-3). This resulted in

2,856 observations of spinner dolphin behaviour (2,395 inside bays and 461 outside bays; Figure 5.1). The proportion of different substrate types available to spinner dolphins in the study area varied inside and outside bays. However, the highest proportion was rock/boulder, followed by sand and then aggregate reef both inside and outside bays (Table 5-4). Spinner dolphins predominately rested inside bays

(Figure 5.2) while predominantly travelled outside bays (Figure 5.3). Although the highest proportion of substrate available to spinner dolphins inside and outside bays was rock/boulder, spinner dolphins occurred disproportionately more over sand than any other substrate, 54% inside bays and 38% outside bays.

108

Table 5-3 Number of focal follows, focal follow hours and mean focal follow duration from land-based and boat-based group focal follows of spinner dolphins inside bays and outside of bays along the Kona Coast, Hawaii Island. Focal follow Number of Focal follow Mean focal focal hours follow duration follows (hh:mm) Land-based Kealakekua Bay 40 329 8:31 ± 0:19 SE Kauhako Bay 7 38 3:25 ± 1:17 SE Total 47 367 7:40 ± 0:22 SE

Boat-based Honaunau Bay 4 21 5:15 ± 1:24 SE Makako Bay 4 14 3:26 ± 0:27 SE Outside Bays 20 86 4:18 ± 0:33 SE Total 28 121 4:20 ± 0:27 SE

Overall total 75 488 6:30 ± 0:20 SE

Table 5-4 Proportion of substrate types available to spinner dolphins inside and outside of bays within the study area. Substrate Inside bays Outside bays Aggregate reef 0.20 0.15 Rock/boulder 0.46 0.63 Sand 0.34 0.22

109

0.60 Rest Inside bays 0.54 Social 0.50

Travel

0.40

0.29 0.30

0.20 Proportion of time of Proportion 0.10 0.06 0.07 0.02 0.03 0.00 0.00 0.00 0.00 Aggregate Reef Rock/Boulder Sand

Substrate type

110

0.60 Outside bays (Coastal) Rest

0.50 Social Travel 0.38 0.40

0.30

0.20 0.20 0.18

Proportion of time of Proportion 0.09 0.10 0.06 0.04 0.04 0.02 0.00 0.00 Aggregate Reef Rock/Boulder Sand

Substrate type

5.4.2 Boosted GAMs The Boosted GAM modelling technique was used to take into account autocorrelation in the focal follow data. Since depth and substrate data could not be determined from the remote sensed bathymetric and habitat maps for some of the dolphin group observations, 406 and 102 data points were removed from inside and outside bays, respectively. This resulted in 2,348 behavioural, depth and substrate data observations being included in the modelling procedure.

The prediction accuracy determined by the mean pseudo R2 values for the models to predict the resting state of spinner dolphins was greater than 0.6 for all models

(0.701, 0.655 to 0.734 and 0.613, 0.608 to 0.617 for the two in-bay models; 0.663,

111

0.660 to 0.664 for the coastal model). These models provided a good fit to the behavioural data as pseudo R2 values of between 0.55 and 0.60 have been described as moderate/good (Maloney et al., 2012). Locations, behaviour during previous scan observation and time of day were the most important variables for predicting resting behaviour of spinner dolphins in both the in-bay models (Table

5-5). For the coastal model, the inside/outside variable was the most influential variable in the model, with the time of day, behaviour during previous observation and the inside/outside bay x substrate interaction having a similar level of influence to each other (Table 5-5). No other predictor interactions were influential in predicting spinner dolphin group resting behaviour in the three models.

The coastal model indicated that spinner dolphin groups are unlikely to rest when outside sheltered bays (Figure 5.4). When inside the bays, substrate was influential in predicting resting behaviour (Table 5-5). Depth and distance from shore were never selected as main predictors for any of the three models. Time-of-day was fitted to all three models, which predicted that spinner dolphins had a higher probability of resting in the mid-morning than in the morning or afternoon (Figure

5.5). In addition, spinner dolphin groups in Kauhako Bay and Kealakekua Bay predominantly rested over a sandy substrate (Figure 5.6).

112

Stability selection Variables and selection probability at optimal Model Optimal m frequencies m a) In-bay 1 234 Spatial position 0.67 1 (n = 200) Previous behaviour 0.25 0.98 Time-of-day 0.06 0.91

b) In-bay 2 223 Spatial position 0.52 1 (n =1,596) Previous behaviour 0.31 1 Time-of-day 0.17 1

c) coastal 431 Inside/outside bays 0.30 1 (n =2,348) Inside/outside bays + substrate 0.26 1 Time-of-day 0.16 1 Previous behaviour 0.15 1 Spatial position 0.12 0.94

113

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2 Predicted probability of resting of probability Predicted 0.1

0 Inside Bays Outside Bays

Location

114

1.0 Morning Mid-Morning Afternoon 0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2 Predicted probability of resting of probability Predicted 0.1

0.0 Kauhako Bay Kealakekua Bay Location

115

116

5.5 Discussion The gradient boosted GAM analytical approach (Bühlmann and Hothorn, 2007) was used to identify factors that influence the resting behaviour of spinner dolphins.

This approach overcomes the complexities of non-linearity and autocorrelation that can be found in ecological data. Boosting automatically selects variables and reduces effect estimates towards zero, which in combination, avoids over-fitting

(Hothorn et al., 2010b). Model fitting is constrained by stopping the model fitting process at the optimal number of boosting iterations. This approach was used in preference to maximum likelihood estimates that can over fit models when there are many predictors and complex spatial and temporal processes. Spatial autocorrelation was addressed by fitting a spatial spline in order to fit variance that is purely spatial and not related to the other covariates (Hothorn et al., 2010b). The results confirm that four sheltered bays (Kauhako Bay, Honaunau Bay, Kealakekua

Bay and Makako Bay) along the Kona Coast are important resting habitat for spinner dolphins during daylight hours. Although dolphins spent significant proportions of time resting in bays throughout daylight hours, most rest occurred between 10am and 2pm. This study expands on previous research that highlighted the importance of sheltered bays to spinner dolphins (Norris et al., 1994, Thorne et al., 2012), by illustrating that dolphins are unlikely to rest outside of the key habitats that support this important life function. These results provide management agencies with valuable information to assist in implementing an effective protected area management approach, to reduce the exposure of dolphins to human disturbance during resting periods.

117

Previous studies have shown that spinner dolphins rest over substrate types of low complexity within these bays, i.e. areas of sandy bottom, and usually in less than

60 m of water (Thorne et al., 2012, Norris et al., 1994). To date, however, studies evaluating spinner dolphin resting habitat have focussed on areas inside bays only.

The results from this study show that some habitat variables (e.g. depth and distance from shore) were not important predictors of spinner dolphin rest. In fact, the most important factor contributing to the likelihood of rest was whether dolphins were within a bay or not. The interaction between substrate type and in- bay presence, suggests that substrate (sand) is partially influential in predicting resting behaviour. In coastal areas outside of bays, spinner dolphins spent disproportionately more time over sandy substrates than over other substrates available. However, in contrast to their behaviour in bays, dolphins seldom rested over these sandy substrates. Instead, they were observed mainly travelling over sand outside bays. This may be because the sandy substrate outside bays fails to provide spinner dolphins with as safe a place to rest than when inside bays (Norris et al., 1994) .

Cetacean-based tourism has increased dramatically in Hawaii over recent years

(O'Connor et al., 2009) which has led to increased human exposure to spinner dolphins (Delfour, 2007, Courbis and Timmel, 2009). The cumulative exposure of dolphin populations to human interactions has had detrimental effects on bottlenose dolphins in Doubtful Sound, New Zealand (Lusseau, 2005) and on bottlenose dolphins, Tursiops aduncus, in Shark Bay, Western Australia (Bejder et al., 2006b, Higham and Bejder, 2008). In New Zealand, resting has been identified

118 as the most sensitive behavioural state to disturbance of one population of bottlenose dolphins (Lusseau, 2004). Spinner dolphin resting behaviour is often interrupted or truncated by human activities and they leave resting bays in direct response to human disturbance (Courbis and Timmel, 2009). Rest is a vital component in the energy budgets of most animals (Cirelli and Tononi, 2008); as animals tire, they become less vigilant and more vulnerable to predators (Dukas and Clark, 1995). During night time foraging bouts spinner dolphins herd their prey to increase its density and then cooperatively feed on these high density aggregations (Benoit-Bird and Au, 2009). To recover from the energetically- demanding foraging activity and increase their vigilance, spinner dolphins return to these sheltered bays to rest (Johnston, 2014).

The results of the present study provide critical, but until now, missing evidence, i.e. that outside sheltered bays, spinner dolphins are unlikely to rest. If dolphins leave resting bays to avoid disturbance from human activities, our results indicate that they are unlikely to rest and recover from the ongoing energetic and cognitive costs associated with their rigid daily schedules.

5.5.1 Management of marine mammals The MMPA was originally designed to minimise the capture (or ‘take’), harassment and disturbance of marine mammals, primarily from by-catch from fisheries and cetacean hunting. The MMPA defines the term ‘take’ as “… hunting, killing, capture and harassment of a marine mammal or the attempt thereof”. Harassment is defined as “… any act of pursuit, torment, or annoyance which (i) has the potential

119 to injure a marine mammal or marine mammal stock in the wild; or (ii) has the potential to disturb a marine mammal or marine mammal stock in the wild by causing disruption of behavioral patterns, including, but not limited to, migration, breathing, nursing, breeding, feeding, or sheltering.” Most human-dolphin interactions (boat-based or swim-with) cause behavioural disruptions in dolphins, which by the above definition is ‘harassment’. The burden of proof in documenting dolphin behavioural changes as a consequence of human activities rests with the management agency. However, interpreting dolphin behavioural changes as a consequence of human activities is challenging and often clouded by arguments that any observed behavioural changes are a consequence of natural phenomena and not induced by human activity (Johnston, 2014). This highlights a need for an enforcement policy to make legislation more easily understood, less ambiguous and more fairly enforced. In 2005, NOAA published an Advance Notice of Proposed

Rulemaking to alert the public about the concerns surrounding human-dolphin interactions, and to solicit feedback on potential options for future regulations under the MMPA (NOAA, 2005).

5.5.2 Management implications The Hawaii Island associated spinner dolphin population may be especially vulnerable to disturbance because it is small (Chapter 3; Tyne et al., 2014a), genetically distinct (Andrews et al., 2010) and is unlikely to rest outside sheltered bays. On a daily basis humans seek out close-up interactions with spinner dolphins both inside and outside of important resting areas (Wiener et al., 2009, Courbis and

Timmel, 2009). Cumulative exposure to human interactions within resting habitats

120 may have a detrimental impact on spinner dolphins. Energetic models of spinner dolphins in Hawaiian waters indicate that they are less likely to rest when swimmers were within 150 m (J. Symons et al. University of Aberdeen unpublished data). Although the current level of swim-with exposure in this region does not appear to contribute to energetic deficits in spinner dolphins, research indicates that any further increase in intensity is likely to drive these dolphins into an energetic debt (J. Symons et al. University of Aberdeen unpublished data).

My results support management actions to reduce human access to preferred dolphin resting areas during important resting periods. Using environmental, spatial and temporal estimates of key habitats and guidance from energetic models (J.

Symons et al. University of Aberdeen unpublished data), I highlight two management approaches that should be considered. The following options to mitigate possible detrimental effects of human activity on spinner dolphins are based purely from a biological and conservation perspective. Other factors (e.g. cultural values and fishing activities) (Heenehan et al., 2015) also need to be taken into consideration. Management options include, but are not limited to:

1. Restricting all human activity throughout bays during dolphin rest periods.

2. Restricting human access to specific habitats (sandy bottom) within resting bays

during important dolphin rest periods in combination with implementing a

buffer zone, e.g., 150 m - 300 m, around these particular habitats.

Distances over water are difficult to estimated (Kinzey and Gerrodette, 2003).

Therefore I recommend that any management action implementing restrictions to

121 geographical regions should include surface markers to delineate the restricted areas (e.g. U.S. Fish and Wildlife Service, 2001).

These management options take into account the need for an easily enforceable policy by providing unambiguous solutions that can effectively protect the spinner dolphins from harassment within important resting areas. Importantly, when exploring measures to protect spinner dolphin resting habitat, decision makers should note conclusions of a recent review of marine protected areas which highlighted that the effectiveness of protected areas are dependent on protecting an area of adequate size and on compliance and enforcement (Edgar 2014).

Interactions between human activities and marine vertebrates are often negative, and the approach developed in this paper, where models explicitly link behaviour with habitat characteristics to identify important habitats for protection, is much needed. This approach is applicable to ongoing conservation conflicts, but could also be a component of recovery plans for depleted species. Such models could anticipate and avoid future conflicts as animals recover from exploitation and reoccupy portions of their ranges. For example, female and young calf humpback whales (Megaptera novaeangliae) use Exmouth Gulf, Western Australia, as a resting area (Braithwaite et al., 2012). Exmouth Gulf has been earmarked for possible future resource exploration and aquaculture development, and there is a pressing need to identify crucial resting habitats and transit corridors before development begins. This approach would also be useful for recovering species of pinnipeds, such as gray seals in the U.S. East Coast (Wood et al., 2011), where

122 combined assessments of breeding behaviour and colony habitat characteristics could anticipate where new breeding colonies may form and how these colonies may interact with coastal communities and other components of the marine ecosystem.

123

6 Cumulative exposure of Hawaiian spinner dolphins (Stenella longirostris) to human interactions: No rest for the weary

6.1 Abstract Repeated exposure of wildlife to disturbance may affect individual vital rates and ultimately population viability. Here, I quantify the cumulative exposure of human activity on a small, genetically-isolated spinner dolphin stock off Hawaii Island that is exposed to close-up human activity on a daily basis. I modelled the daytime cumulative activity budget (resting, socialising and travelling) of dolphins under impact (human activities within 100 m) and control (no human activities within 100 m) conditions. First, a systematic photo-identification study of individual dolphins (n

= 235) was used to record their presence/absence in each of four important dolphin resting bays, based on the proportion of time they were observed in each bay.

Secondly, concurrent passive acoustic recordings (n = 1,546,560 30-sec recordings) were used to document the daily presence/absence of dolphins within bays.

Thirdly, data from land-based and boat-based group focal follows (n = 428 hrs) inside and outside resting bays were used to provide behavioural time series of dolphins under control and impact conditions. Simulations were used to estimate location specific (inside bays/outside bays) activity budgets for individual dolphins.

Finally, the cumulative activity budget of the population was estimated. During the day, individual spinner dolphins spent between 49.5% and 69.4% of their time resting (mean=61.7%, SD=6.5) and were exposed to human activities for 82.7% of the time. Despite the high level of exposure, human activities seemingly did not have a significant effect on dolphin activity budgets. This result is, however, likely an artifact of the low level of control data available (< 18% of observations) to make robust comparisons between behavioural patterns in control and impact conditions 124

Furthermore, intervals between interactions were short (median duration =

10 min), before dolphins were exposed again, which may prevent individuals recovering from disturbance and deprive them of rest. Control observations, as defined in this study, may not accurately represent true resting behaviour of spinner dolphins. Specifically, it is likely I quantified resting behaviour when dolphins were in a “light” rather than a “deep” sleep behavioural state, which may lead to rest deprivation, impaired cognitive abilities and ultimately effects on population viability. The chronic exposure of spinner dolphins to human interactions and the concurrent use of the resting bays for recreational, commercial and subsistence purposes must now be of major concern for management.

6.2 Introduction The effect of repeated exposure to human interactions on wildlife populations is a concern for conservation management. Animals may respond to human interactions as they would in the presence of a natural predator (Frid and Dill,

2002), by engaging in anti-predator behaviours (e.g. increased vigilance and/or avoidance) at the costs of decreasing their time engaging in other important activities such as foraging or resting (Lima and Dill, 1990, Lima, 1998, Heithaus and

Dill, 2002). Repeated interactions with humans lead to changes in many aspects of dolphin behaviour and energetics, such as: activity budgets (Lusseau, 2003a), energetic deficits (Williams et al., 2006, Christiansen et al., 2014), reduced vigilance

(Johnston, 2014), alteration of social interactions with conspecifics (Lusseau and

Newman, 2004), disruption of day-time behavioural patterns and inadequate rest time (Lusseau, 2004) and displacement from prime habitat to less optimal habitat

125

(Gill et al., 2001). These effects may have negative impacts on individual vital rates, including survival and reproduction (National Research Council, 2005, Christiansen and Lusseau, 2015), exceeding those caused by natural predation (Ciuti et al., 2012) and ultimately having negative consequences for the viability of a population (Frid and Dill, 2002).

Since the early 1980s, demand for human interactions with free-ranging cetaceans has increased dramatically worldwide (O'Connor et al., 2009, Chapter 2; Tyne et al.,

2014b) Many coastal dolphin populations are now exposed to prolonged close-up human encounters, which may disrupt the natural behavioural pattern between interactions and prevent individuals from displaying their normal, undisturbed behaviours (Bejder et al., 2006a, Lusseau, 2003a, Christiansen et al., 2010,

Constantine et al., 2004). In the short-term, dolphins may be able to compensate for this temporary behavioural disruption (New et al., 2013) by, for example, feeding at other times and/or locations (Heithaus and Dill, 2002). The frequency of disturbance caused by human activities, however, may prevent recovery to a natural behavioural pattern between interactions (Lusseau, 2004). Thus, the effect of repeated short-term disruptions may lead to long-term avoidance strategies, such as the avoidance of important and preferred habitats (National Research

Council, 2005, Lusseau, 2004, Bejder et al., 2006b). Habitat avoidance strategies may be constrained, however, by the absence of suitable alternatives (Gill et al.,

2001, Bejder et al., 2009) and the condition of individuals, e.g., the animals being too weak to relocate (Beale and Monaghan, 2004). As a consequence, individuals may have no option but to remain and endure the disturbance, despite the

126 potential cost of this to the population viability (Gill et al., 2001, Beale and

Monaghan, 2004, Bejder et al., 2009). Therefore, understanding how cumulative behavioural disruptions affect the viability of populations is paramount to preventing long-term negative impacts and ensuring the conservation of the populations (National Research Council, 2005, Bejder et al., 2006b, Lusseau et al.,

2006, Currey et al., 2009). Furthermore, this information is valuable for developing effective mitigation approaches that reduce the cumulative exposure of dolphins to human interactions.

Hawaiian spinner dolphins (Stenella longirostris) exist in small (Chapter 3; Tyne et al., 2014a), isolated stocks with restricted ranges (Andrews et al., 2010), and have evolved a specialised diurnal behavioural pattern. They cooperatively forage offshore at night and return to sheltered bays to rest and socialise during the day

(Norris and Dohl, 1980, Norris et al., 1994, Benoit-Bird and Au, 2009). This temporal partitioning of behaviours allows spinner dolphins to maximize their foraging efficiency, while avoiding predation during periods of recovery from their night time foraging exertions (Johnston, 2014). As animals tire, they accrue what is often referred to as a vigilance decrement (Dukas and Clark, 1995). In higher vertebrates

(reptiles, birds and mammals), vigilance decrements can manifest as a decreased ability to detect camouflaged predators, which can lead to an increase in predation risk (Heithaus and Dill, 2002). To recover from a vigilance decrement, animals must rest (Cirelli and Tononi, 2008). The very structured, predictable behavioural pattern of spinner dolphins renders them especially vulnerable to interrupted resting bouts

127 during the day, as they have a limited ability to recover before embarking on another foraging bout the following evening.

Hawaiian spinner dolphins are sought out and targeted on a daily basis by humans eager to engage in close-up encounters (Norris, 1991). Throughout the day, spinner dolphins are repeatedly approached by kayakers, swimmers and vessels for close- up encounters (Courbis and Timmel, 2009) inside and outside their preferred resting habitats within sheltered bays (Chapter 5; Tyne et al., 2015). The cumulative effect of repeated interactions may have negative consequences for the spinner dolphins, as they are less likely to rest when swimmers approach within 150 m

(Symons et al., University of Aberdeen unpublished data). Furthermore, as a response to repeated interruption of resting bouts, it has been suggested that dolphins are spending less time in the sheltered bays during the day (Courbis and

Timmel, 2009, Timmel et al., 2008, Delfour, 2007). If the dolphins are displaced from their preferred resting habitat, they are unlikely to rest outside the bays

(Chapter 5; Tyne et al., 2015).

Concerns have been raised regarding the effects of human-dolphin interactions on the Hawaii Island spinner dolphins (Danil et al., 2005, Courbis and Timmel, 2009).

The National Oceanic and Atmospheric Administration (NOAA) are looking to implement a management strategy to reduce the number and intensity of human- dolphin interactions in Hawaii (NOAA, 2005). One management strategy under consideration is the implementation of time-area closures to restrict human access

(Chapter 2; Tyne et al., 2014b) in specific habitats that are important to the stock

128

(Chapter 5; Tyne et al., 2015). Spatial management is considered an effective approach to conserve marine mammal populations (Gormley et al., 2012), when implemented at appropriate spatial and temporal scales (Edgar et al., 2014, Chapter

2; Tyne et al., 2014b).

To help understand the effect of human interactions on spinner dolphins and to assist the development of an effective management strategy, data were collected to determine the impact of cumulative exposure of dolphins to human activities, both inside and outside preferred resting bays. These data came from a suite of complementary sampling techniques, including boat-based photo-identification surveys of individual dolphins, boat-based and land-based group focal follows and passive acoustic recordings. The information from this study will be of great value to management agencies in developing mitigation strategies for human-spinner dolphin interactions.

6.3 Materials and methods 6.3.1 Fieldwork In the waters off the Kona coast, on the leeward side of Hawaii Island (Figure 6.1), spinner dolphins are often observed within four bays during the day: Makako Bay,

Kealakekua Bay, Honaunau Bay and Kauhako Bay (Figure 1; Norris et al., 1994,

Thorne et al., 2012, Chapter 3; Tyne et al., 2014a). Between September 2010 and

March 2013, boat-based surveys were undertaken both inside and outside (within 1 km of the coastline) of these bays (see Chapter 5; Tyne et al., 2015 for protocol) to photographically-identify individual spinner dolphins and to record group behaviour

129

(see Chapter 3; Tyne et al., 2014a for protocol). Land-based group behavioural focal follows were undertaken via theodolite tracking from clifftops overlooking Kauhako

Bay (50 m elevation) and Kealakekua Bay (140 m elevation). Concurrently, acoustic data of dolphin vocalisations (echolocation clicks, burst pulse sounds and whistles) were recorded via bottom-mounted passive acoustic loggers in all four bays.

130

Figure 6.1 Map of the study area illustrating the four spinner dolphin resting bays, Makako Bay, Kealakekua Bay, Honaunau Bay and Kauhako Bay, along the Kona Coast of Hawaii Island.

131

6.3.2 Systematic photo-identification sampling From September 2010 to August 2012, boat-based photographic-identification surveys within the four preferred resting habitats of the Hawaii Island spinner dolphin stock were carried out using a systematic sampling design developed by

Tyne et al. (2014a). Each bay was surveyed on the same dates each month, regardless of whether dolphins were present or absent, providing consistent and even effort throughout the study period and area: four days in Kauhako Bay; two days in Honaunau Bay; four days in Kealakekua Bay; and two days in Makako Bay

(see Chapter 3; Tyne et al., 2014a for protocol). These data were used to investigate whether individual dolphins showed a preference for a particular bay.

6.3.3 Group focal follows Established group focal follow protocols were employed to collect positional and behavioural information on spinner dolphins during daylight hours from both boat- based and land-based (theodolite observations) platforms. Group focal follows consisted of a combination of continuous and instantaneous scan sampling procedures (Altmann, 1974, Mann, 1999). Instantaneous scan sampling recorded the predominant group activity (see Table 1 for definitions of dolphin group activities) at 10-min intervals (Altmann, 1974, Mann, 1999, Bejder et al., 2006a).

Behavioural data were categorised as control (undisturbed behaviour) or impact situations. A control situation was when no kayaks, swimmers or boats were with

100 m of the focal group, while an impact situation was when a kayak, swimmer and/or boat was within 100 m of the focal group. A group focal follow was terminated when dolphin behaviour could no longer be reliably determined

132 because of poor visibility, dolphins moving out of range or splitting into too many groups. Further details of boat-based and land-based group focal follow protocols are given in Chapter 5; Tyne et al. 2015.

Table 6-1 Definitions of spinner dolphin group activities, adapted from Norris et al (1994). Predominant group activity: Rest: Characterised by tight group, slow speed moving back forth generally Over sandy substrate or meandering movement. Individuals typically take multiple breaths; synchronous group diving; changing direction whilst underwater; spend long periods of time submerged (1.5-3 minutes). Social: Characterised by regular, consistent, aerial behaviours within the group; little time is spent below the surface; dives are brief. Travel: Characterised by regular and consistent spatial progress with respect to the bottom (in practice surface and shoreline features), i.e.directed swimming that is roughly straight. Travel speed is typically 3.2km/hr. Forage: Spinner dolphins do not forage during the day, but only at night offshore (Norris et al., 1994, Beniot-Bird and Au, 2003), thus this behavioural state was not observed.

6.3.4 Passive acoustic recordings From January 2011 through August 2012, 30-second, calibrated acoustic recordings were made every four minutes at a sampling rate of 80 kHz within each of the four resting bays via bottom-mounted DSG-Ocean acoustic instruments (Loggerhead

Instruments, Sarasota, FL, USA). Recorders were equipped with HTI-96-Min/3V

-1 hydrophones (sensitivity: within 1 dB of -186.6 dbV μPa , High Tech Inc, Gulfport,

MS, USA), a 16-bit computer board and 30.9 GB SD cards. Acoustic data were retrieved approximately every two weeks.

133

6.4 Data analyses 6.4.1 Cumulative activity budgets The cumulative activity budget of the dolphin population was estimated to evaluate disturbance of human activities. The cumulative activity budget considered the relative time that individual dolphins spent inside and outside of each of the four bays, throughout the study period (Jan 8th 2011–Aug 30th 2012). Model simulations estimated the cumulative activity budgets of the photographically-identified dolphins. For each day, dolphins were randomly allocated to different bays based on their relative occurrence in the bays, provided by the photo-identification data

(Figure 6.2A). When present in a bay, dolphins generally stayed there for the entire day and rarely left the bay or relocated to another bay (Tyne, Murdoch University, pers. obs.).

After allocating dolphins to bays, the passive acoustic data were used to confirm if dolphins (binary response based on whether or not dolphin vocalisations had been documented in recordings made within each bay) had indeed visited a particular bay on a given day. If no acoustic data were available for a given day (equipment malfunctions), dolphin presence was drawn at random for that day using a Bernoulli process informed by the probability that dolphins would be present in that bay

(dolphin presence/number of observations). If a bay had not been visited by dolphins on a given day, the dolphins allocated to that bay were removed and allocated to outside the bays (Figure 6.2B). Based on where an individual dolphin had been allocated, the time spent resting, socialising and travelling was estimated, based on the area-specific activity budgets (Figure 6.2C).

134

To account for uncertainty in the activity budget, a random proportion of the activity budget was drawn from the density distributions that were obtained for the area-specific activity budgets (see above) and allocated to a dolphin. This was repeated for every day of the study period, assuming no within-day movements between bays or movements from bays to outside of bays. The cumulative activity budget was estimated by taking the sum of the duration of the different activity states throughout the study period (Figure 6.2).

135

136

6.4.2 Calculation of dolphin activity budgets for each bay Activity budgets were defined as the proportion of time dolphins spent in each activity state and estimated directly from behavioural data time series. Boat-based data were used to estimate dolphin activity budgets for Makako Bay, Honaunau Bay and outside bays. Boat-based and land-based data were combined to estimate dolphin activity budgets within Kauhako Bay and Kealakekua Bay. For each bay, activity states were drawn at random (with replacement) from the original data set, with the total number of states equalling the original number of activity states for each bay. A new activity budget was then estimated for that bay. This was repeated

1,000 times to obtain a density distribution of the relative proportion of each activity state in the activity budget for each bay. From the resulting density distributions, 95% highest posterior density (HPD) intervals, analogous to 95% confidence intervals, were calculated. All calculations were performed using R 3.0

(R Core Team, 2014).

6.4.3 Factors affecting dolphin activity states To better understand which factors influence spinner dolphin activity states, I used the boat-based and land-based observational data to determine how the probability of dolphin resting, socialising and travelling were affected by different covariates, including: time-of-day (hour); day-of-year, dolphin group size; location

(inside or outside bays); number of boats/kayaks/swimmers present (within 100 m of the dolphin group); and distance between dolphin group and boats/kayaks/swimmers. Generalized Additive Mixed Models (GAMM; gamm in R package mgcv) were used with a thin plate regression spline smoother and a

137 binomial distribution and logit link function. In the model selection process, covariates and interactions between covariates were added sequentially to the null model. The F-statistic for the ANOVA F-test was estimated for each model and compared with that of the previous model. Covariates were added both as linear effects and non-linear smoothers to cover all possible relationships between the response and explanatory variables. As sequential observations within focal follows could not be considered independent, a temporal auto-correlation structure within follows was incorporated in the model, where the residuals at any given time were modelled as a function of the residuals of the previous time points. The most suitable auto-correlation structure was fitted by altering the number of auto- regressive and moving average parameters and then comparing the different models. Auto-correlation and partial auto-correlation function plots were used to detect patterns of auto-regressive and moving average parameters visually, before and after adding the different correlation structures.

The variance inflation factor (VIF) was used to investigate collinearity (high correlation) between the explanatory variables in the model. A threshold value of three was used to remove collinear variables one at a time until all VIF values were below three and no collinearity remained. For all models, model validation tests were run to identify potential violations of assumptions. Scatter plots of residuals versus fitted values and residuals against each explanatory variable were used to test the assumption of equal variances (i.e. homogeneity of variance) in the model.

Normality of residuals was interpreted from Quantile-Quantile plots and from residual histograms. Over-dispersion was tested for each model by dividing the

138 residual deviance by the residual degrees of freedom and a value of > 1.5 was used to indicate over-dispersion (Cox, 1983)).

6.4.4 Duration of interactions between dolphins and humans The elapsed time between repeated disturbances can influence the activity budget of dolphins (Lusseau, 2004), as exposed animals generally need time to return to their initial activity state following an interaction (Meissner et al., 2015, Bejder et al., 2006a). Insufficient time between interactions may prevent dolphins from returning to their initial activity state. Consequently, the activity state between interactions may not be a true representation of undisturbed dolphin activity state

(“control”). To investigate the elapsed time between human interactions with dolphins (“impact”), frequency histograms of control and impact bout durations

(continuous time spent in impact or control situations) were inspected and compared.

6.4.5 Passive acoustic recordings Daily spectrograms were generated primarily in Raven Pro 1.5 (Cornell University) but were also initially generated using XBAT (Cornell University), a bioacoustics toolbox in Matlab. Spectrograms were generated using a 512-point DFT, 50% overlap, and a 512 point (6.4 ms) Hann window. The presence of dolphin vocalisations was investigated for each bay per day through manual visual inspection, and was used to document the presence/absence of dolphins based on whether or not vocalisations had been documented on recordings. Each daily spectrogram was analyzed until dolphin vocalizations were documented. Once

139 dolphin vocalizations were found on that day, the data were entered into a spreadsheet, and this process continued for each day in the time series. Dolphin vocalisations included whistles, burst pulse sounds and echolocation clicks. To avoid mis-identification of background noise given the prevalence of snapping shrimp sounds, files were marked with the presence of dolphin vocalisations only when the following criterion were met: an echolocation click was found, the echolocation bout was clear and sharp, spanned most of the recording and was found in sequential recordings or it was followed by other dolphin vocalisations.

6.5 Results 6.5.1 Summary of photo-identification and behavioural sampling efforts The systematic sampling design developed in Chapter 2 (see also Tyne et al. (2014a) to collect individual spinner dolphin photo-identification data resulted in nearly

2,500 hours of on-water effort over 276 days of sampling (Table 6-2). Furthermore, approximately 4,000 hours of matching and grading spinner dolphin dorsal fin images were required to identify 235 highly distinctive individual spinner dolphins.

The sighting frequency of individual dolphins ranged from one to 48 during the study period (Figure 3), with a mean (± 1 SE) of 12 ± 0.52.

Table 6-2 Number of days, hours and mean length of photo-identification surveys conducted within the four resting bays between September 2010 and August 2012 Location Survey days Hours Mean survey length Makako Bay 46 395 8:35 ± 0:11 SE Kealakekua Bay 92 818 8:53 ± 0:13 SE Honaunau Bay 46 412 8:57 ± 0:15 SE Kauhako Bay 92 856 9:18 ± 0:10 SE Total 276 2481 9:00 ± 0.12 SE

140

10 Number of observationsofNumber

1 7

13 19 25 31 37 43 49 55 61 67 73 79 85 91 97

127 145 163 181 199 217 235 103 109 115 121 133 139 151 157 169 175 187 193 205 211 223 229 Individual dolphin

Figure 6.3 Sighting frequencies of 235 photographically-identified spinner dolphins in Makako Bay, Kealakekua Bay, Honaunau Bay and Kauhako Bay between September 2010 and August 2012.

From a total of 105 boat- and land-based dolphin group focal follows that were conducted over approximately 428 hours, 75 (71.4%) were from the boat-based platform and 30 (28.6%) were from the land-based platform (Table 6-3).

Table 6-3 Number and duration of dolphin focal follows collected from land-based and boat-based platforms inside bays and outside of resting bays along the Kona Coast, Hawaii Island. Number of Mean focal focal Total focal follow duration Focal follow follows follow hours (hh:mm ± SE) Land-based Kealakekua Bay 23 189 8:27 ± 0:19 Kauhako Bay 7 38 3:25 ± 1:17 Total 30 227 7:57 ± 0:22

Boat-based Makako Bay 13 26 2:00 ± 0:26 Kealakekua Bay 10 16 1:36 ± 0:15 Honaunau Bay 5 21 4:12 ± 0:15 Kauhako Bay 10 16 1:36 ± 0:48 117 Outside Bays 37 3:10 ± 0:13 Total 75 201 2:41 ± 0:10

Overall total 105 428 4:08 ± 0:51

141

6.5.2 Bay use by individual dolphins Observations of individual spinner dolphins would be inflated in Kealakekua Bay and Kauhako Bay due to the higher sampling effort in these bays (Table 1). To take this into account the data were standardised by:

 Randomly selecting two of the four sampling days in Kealakekua and

Kauhako Bay 100 times.

 Identifying the frequency of individual dolphin observations in each bay on

both of the selected days.

 Calculating the proportion of observations of each spinner dolphin in each

bay over the study period.

 Averaging those proportions across the 100 samples.

Most dolphins (94%, n=220) were observed in Makako Bay, followed by Kealakekua

(55%, n=130), Honaunau (53%, n=124) and Kauhako (36%, n=85; Figure 6.4).

100 90 80 70 60 50 40 30 20

Proportion ofindividuals Proportion 10 0 Makako Bay Kealakekua Bay Honaunau Bay Kauhako Bay Resting Bays

Figure 6.4 Proportion of the 235 identified spinner dolphins were documented in each of the four resting bays. Data from Kealakekua Bay and Kauhako Bay were standardised and error bars are shown.

142

6.5.3 Passive acoustic monitoring efforts In each of the four bays, bottom-mounted acoustic loggers were simultaneously deployed during 601 days. A total of 1,546,560 30-sec recordings were made over the study period (Table 6-4). Acoustic recordings confirmed the presence of dolphins during 90% of monitoring days in Makako Bay, 65% in Kealakekua Bay,

37% in Honaunau Bay and 51% in Kauhako Bay (Table 6-4)

Table 6-4 Number of days spinner dolphins were present, absent, logger malfunctions and number of recordings in each resting bay from 601 days of passive acoustic recordings in each bay. The acoustic loggers recorded for 30 seconds every four minutes. Days with No. of days that Total number No. of days No. of days acoustic acoustic loggers of 30 s dolphins dolphins Bay data malfunctioned recordings were present were absent Makako 565 36 406,800 506 59 Kealakekua 484 117 348,480 315 169 Honaunau 563 38 405,360 209 354 Kauhako 546 65 385,920 274 262

6.5.4 Dolphin activity budgets inside and outside resting bays Spinner dolphins spent most of their time resting, followed by socialising and travelling (Figure 6.5). The proportion of time spent resting was higher inside bays

(>60%) than outside (<40%; Figure 6.5). While time spent socialising was approximately the same (35%) inside and outside bays, dolphins spent a substantially higher proportion of time travelling outside (30%) than inside bays

(5%; Figure 6.5). There was little variation in the dolphins’ activity budget between bays, with the exception of Makako Bay, where dolphins spent more time resting

(72.6%), and less time socialising (19.0%) than in the other bays (Figure 6.5).

143

6.5.5 Cumulative activity budget for the sampled population Simulations showed that during daytime individual spinner dolphins spent most of their time inside bays (76% of time), while spending 24% of time outside bays

(Figure 6.5). The cumulative activity budget showed that Individual dolphins spent between 49.5% and 69.4% of their time resting (mean=61.7%, SD=6.5). There was a peak in resting activity around 70% (Figure 6.6) as a consequence of dolphins spending most of their time in Makako Bay, where resting was relatively higher

(Figure 6.5). Socialising activity showed a similar pattern, with dolphins spending between 20.2 and 34.7% of their time socialising (mean=26.1%, SD=5.0), with a peak around 20% (Figure 6.6), corresponding to the activity budget at Makako Bay

144

(Figure 6.6). While both resting and socialising activities showed relatively large individual variations, there was not much variation in the proportion of time dolphins spent travelling, ranging between 10.3 and 16.9% of their time

(mean=12.2%, SD=1.6; Figure 6.6).

6.5.6 Factors affecting dolphin activity The most parsimonious GLMM explaining the daytime activity of spinner dolphins included time-of-day as the only explanatory variable (Figure 6.7). There was a curvilinear relationship between resting probability and time-of-day

(F5.6,2506.4 = 16.9, P < 0.001), with dolphins having a peak in resting in the middle of the day (between 12:00 and 14:00) and a lower probability of resting during early 145 morning and late afternoon (Figure 6.7A). An inverse curvilinear relationship was found between socialising (F5.0,2507.0 = 11.7, P < 0.001) and travelling probabilities

(F4.5,2507.5 = 7.1, P < 0.001) and time-of-day. While the peak in the probability of socialising occurred early in the morning (between 07:00 and 09:00) (Figure 6.7B), the peak in the probability of travelling occurred a bit later in the morning (between

09:00 and 11:00) (Figure 6.7C). Both activity states showed a second peak late in the afternoon (between 16:00 and 18:00) (Figure 6.7). All three models included an auto-correlation within focal follows, with a lag of one (10 mins). The dispersion parameters (φ) for the resting, socializing and travelling models were 0.96, 0.97 and

0.79, respectively, which indicated no over-dispersion of the Binomial GLMMs.

Time-of-day explained 25.7%, 18.7% and 2.7% of the deviance in the data, for resting, socialising and travelling, respectively.

146

Figure 6.7 The probability of spinner dolphins (A) resting, (B) socialising and (C) travelling as a function of time-of-day, estimated from the 428 hours of behavioural focal follow observations.

None of the human activity covariates (presence of boats/kayaks/swimmers, distance between dolphins and boats/kayaks/swimmers) had a significant effect on the probability of dolphins resting, socialising or travelling. All human activity

147 covariates (presence of boat/kayak/swimmer) were collinear, and were also collinear with time-of-day, preventing the use of more than one covariate in each model. The control data (no boats/kayaks/swimmers present within 100m), constituted only 27.7% of the land-based data, and 5.3% of the boat-based data. In total, spinner dolphins were exposed to human activities (impact scenario; human activity within 100m) during 82.7% of the time (focal follow observations).

6.5.7 Duration of interactions The frequency histograms of control and impact bout durations (i.e. continuous time spent in each situation) showed that control bouts were shorter compared to impact bouts (Figure 6.8). The median bout durations during control situations were 10 minutes for both the boat- and land-based data, while the corresponding impact durations were 70 and 30 minutes, respectively.

148

6.6 Discussion Data collected from a suite of sampling methods were analysed in combination to derive an estimate of the cumulative exposure of the spinner dolphin stock to human activities off the Kona Coast of Hawaii Island. During daytime hours, dolphins were exposed to human activities for > 82% of the time. This level of exposure appears to be much higher (> 25%) than those previously reported for any other dolphin species and similar to those reported previously for spinner dolphins

(Table 6-5). Despite the high level of exposure, however, human activities seemingly did not have a significant effect on the probability of spinner dolphins resting, socialising or travelling. This result is, however, likely an artifact of the low 149 level of control data available (< 18% of observations) to make robust comparisons between behavioural patterns in control and impact conditions. Furthermore, control bouts (no human activity within 100 m of dolphins) had a median duration of only ten minutes before dolphins were exposed to human activity again. As dolphins need time to recover from disturbance to return to a pre-disturbed activity state (Lusseau, 2004, Bejder et al., 2006a), it is likely that the short time intervals between interactions may be insufficient for the dolphins to recover from disturbance (Moberg, 2000). For example, bottlenose dolphins in Milford Sound,

New Zealand, required at least 68 minutes between interactions to recover to their pre-disturbed behaviour (Lusseau, 2004). Consequently, the control observations, as defined in this current study, may not accurately represent natural resting behaviour of spinner dolphins and may represent a behavioural state of “light sleep” rather than in “deep sleep”. This could explain why no significant differences were detected in the probabilities of observing specific activity states during impact and control situations. Future research should obtain behavioural data in an undisturbed environment on day time activities of Hawaiian spinner dolphins, e.g., from another Hawaiian Island (Molakai) location with lower population and fewer people seeking to interact with dolphins than along the Kona Coast of Hawaii Island.

This would provide important information on undisturbed dolphin behavioural patterns and activity budgets (resting, socialising, travelling) for comparing with tourism-exposed dolphins, such as those off the Kona Coast.

150

Proportion of time exposed Impact to human distance Source of Behavioural Species activities % (m) disturbance response Study Bottlenose dolphin (Tursiops truncatus) 9 400 MV, K Yes (Lusseau, 2003a) Bottlenose dolphin T. truncatus 10.8 400 MV, K Yes (Lusseau, 2004) Bottlenose dolphin T. truncatus 12.8 400 MV, K Yes (Lusseau, 2004) Bottlenose dolphin T. truncatus 15.5 400 MV, K Yes (Lusseau, 2006) Common dolphin (Meissner et al., Delphinus sp. 21 300 MV, SUP, K Yes 2015) Hectors dolphin Hectori hectori 23.6 200 MV No (Bejder et al., 1999) Bottlenose dolphin T. truncatus 24 50 MV, S Yes (Peters et al., 2012) Common dolphin Delphinus sp. 29 300 MV Yes (Stockin et al., 2008) Killer whale (Orcinus orca) 28.5 100 MV Yes (Lusseau et al., 2009) Dusky dolphin (Lagenorhynchus obscurus) 31 200 MV Yes (Dans et al., 2008) Killer whale (O. orca) 37.6 100 MV Yes (Lusseau et al., 2009) Bottlenose dolphin (Stensland and (T.truncatus) 45 50 MV, S Yes Berggren, 2007) Dusky dolphin (Lundquist et al., (L. obscurus) 51.6 300 MV Yes 2012) Bottlenose dolphin (Constantine et al., (T. truncatus) 58 300 MV Yes 2004)

Spinner dolphin Not (S.longirostris) 77 300 MV, K, S reported (Timmel et al., 2008) Spinner dolphin Insufficient (S.longirostris) 82.7 100 MV, K, S control data This study

During periods of activity, animals usually exhibit enhanced brain function, which is often referred to as vigilance (Dukas and Clark, 1995). Vigilance is required for many

151 activities including foraging, socialising and predator avoidance. As animals undertake these cognitively challenging activities they tire, and accrue what is referred to as a vigilance decrement (Dukas and Clark, 1995). In higher vertebrates, vigilance decrements can manifest in a decreased ability to detect predators or prey

(Dukas and Clark, 1995). While no less serious, vigilance decrement can also manifest in less obvious ways such as reduced cognitive capabilities (Dukas and

Clark, 1995). To recover from their cognitively challenging night time foraging activities (Benoit-Bird and Au, 2009), spinner dolphins need to rest (Cirelli and

Tononi, 2008). Resting in dolphins has been highlighted as the most sensitive activity to interactions with humans (Lusseau and Higham, 2004). Spinner dolphins need to spend at least 60% of their day time activity budget resting to accrue a positive activity budget for the day (Symons et al. University of Aberdeen unpublished data). I have shown that individual spinner dolphins spent between

49.5 and 69.4% of their daytime resting (mean=61.7%, SD=6.5), suggesting that some individuals may be deprived of the rest they need.

Rest deprivation can induce an increase in sleep pressure (Oleksenko et al., 1992), which can lead to impaired cognitive and decision-making abilities (Cirelli and

Tononi, 2008), and in extreme cases death (Rechtschaffen et al., 1983). The decision to rest by free-ranging animals is influenced by the risk of predation (Lima and Dill, 1990, Lima et al., 2005). To reduce predation risk, spinner dolphins gather in sheltered bays to rest during daytime (Norris and Dohl, 1980, Norris et al., 1994,

Chapter 5; Tyne et al., 2015). However, in these same bays spinner dolphins

152 experience chronic exposure to human interactions, which may alter their state of rest to one of a more vigilant nature, i.e. light sleep (Lima et al., 2005).

Depriving dolphins of deep rest may reduce the recovery time from their foraging induced vigilance decrement (Dukas and Clark, 1995), and could result in impaired cognitive abilities (Tartar et al., 2006, Rattenborg et al., 2004, Ganguly-Fitzgerald et al., 2006) leading to increased vulnerability to predation and reduced foraging efficiency. During their cooperative night time foraging bouts, pairs of spinner dolphins take turns to forage within dense prey aggregations (Benoit-Bird and Au,

2009). Resting and socialising in sheltered bays may contribute to the success of the collaborative foraging strategy employed by spinner dolphins (Benoit-Bird and Au,

2009), by developing and reinforcing social bonds between conspecifics. Impaired cognition may affect social interactions which, in turn, can adversely affect the social cohesion of a community (Lusseau and Newman, 2004). The success of the cooperative foraging strategy employed by spinner dolphins (Benoit-Bird and Au,

2009) may be compromised as a consequence. Mothers and calves may be particularly susceptible to sleep deprivation if the ability of mothers to properly care for, feed and protect their calves is compromised (Siegel, 2008). Any reduction in calf survival affects the population viability of the next generation.

Elsewhere, dolphin communities with considerably less cumulative exposure to human activities (Table 6-5) have had their natural behavioural patterns disrupted

(Bejder et al., 2006a, Constantine et al., 2004, Christiansen et al., 2010) and energy budgets affected (Williams et al., 2006). Repeated exposure to human activities has

153 resulted in long-term habitat abandonment (Lusseau, 2004, Bejder et al., 2006a), which has led to longer-term strategies such as the avoidance of important habitats

(Lusseau, 2004), and subsequently to biologically negative impacts on populations

(Lusseau et al., 2006, Bejder et al., 2006b, National Research Council, 2005). For example, in Doubtful Sound, New Zealand, the short term avoidance of tour boats by bottlenose dolphins led to a long-term avoidance of preferred habitat (Lusseau,

2004, Lusseau et al., 2006) and in Shark Bay, Western Australia, the relative abundance of bottlenose dolphins within a tourism area declined in response to an increase in tour vessel activity (Bejder et al., 2006b). Given the adverse effects on the dolphin populations of Doubtful Sound, New Zealand and Shark Bay, Western

Australia, it is likely that the spinner dolphins of Hawaii may be under similar pressure.

The resting bays are subject to considerable use by humans, some of which seek out spinner dolphins for prolonged close-up encounters (Heenehan et al., 2015).

Concerns have been raised regarding the effect that repeated human-dolphin interactions may have on spinner dolphins (Courbis and Timmel, 2009, Delfour,

2007). The cumulative exposure of spinner dolphins to human activities may affect their vital rates (Christiansen and Lusseau, 2015, National Research Council, 2005), alter their energetic budget (Williams et al., 2006) and ultimately affect population viability (Bejder et al., 2006b, Lusseau et al., 2006). I demonstrate that dolphins were chronically exposed to human activities during 83% of observations, a magnitude significantly higher than reported for other dolphin species around the world. Symons et al. (University of Aberdeen unpublished data) showed that

154 spinner dolphins need to be in a resting state during at least 60% of their day-time activity budget to maintain a positive activity budget. Results from my study found that individual spinner dolphins spent between 49.5 and 69.4% of their time resting

(mean=61.7%, SD=6.5; Chapter 6). Therefore, some individuals may be in energetic deficit and, perhaps more importantly, may be deprived of resting opportunities with ramifications for their cognitive abilities (Cirelli and Tononi, 2008, Lima et al.,

2005). The results of this study didn’t show any significant negative effect on the behavioural response of the spinner dolphins to human activities. This could be a consequence of the lack of control data and the tolerance of the dolphins to the chronic exposure to human activities. However, chronic exposure experienced by the spinner dolphins and the negative behavioural responses from populations elsewhere suggest a precautionary approach to the management of this spinner dolphin population.

6.6.1 Management implications The tourism industry is an important economic resource in Hawaii, and spinner dolphin excursions are available year-round (Hu et al., 2009). From 2006–2007, dolphin tour boats were operating close to 100% in the peak winter season

(December to February) and above 60% at other times during the year (Hu et al.,

2009). At present, there are no entry restrictions for new dolphin excursion operators, which may lead to an unsustainable increase in dolphin excursion numbers (Hu et al., 2009). It has been ten years since NOAA published an Advanced

Notice of Proposed Rulemaking raising concerns on spinner human-dolphin interactions and seeking feedback on potential future regulations under the MMPA

155

(NOAA, 2005). Within this time, scientific evidence has confirmed that Hawaii Island spinner dolphins live in small (Chapter 3; Tyne et al., 2014a; Chapter 4) genetically- isolated (Andrews et al., 2010) stocks, and rely on sheltered bays to rest and socialise during the day (Norris et al., 1994, Chapter 5; Tyne et al., 2015). The research in this Chapter has shown that they are exposed to chronic and repeated human interactions and this may deprive them of rest, impair cognitive abilities, increase predation risk, decrease foraging efficiency, accrue an energetic budget deficit (Symons et al., University of Aberdeen unpublished data) and ultimately effect population viability. The concurrent very high use of the resting bays for recreational, subsistence and tourism purposes along the Kona Coast, (Heenehan et al., 2015) must now be of major concern for management.

Management frameworks exist where legislation requires cetacean-watch operators to obtain a license to engage in commercial activities. Such a framework provides managers with the opportunity to regulate the level of exposure of a cetacean population to tourism operations. In such cases, if the industry has a detrimental effect, managers have the power to revoke licenses following an adaptive management procedure (Higham et al., 2008). For example, an unprecedented management decision was made to ensure the long term sustainability of the dolphin-watch operations in Shark Bay, Western Australia. The number of cetacean-watch operators was reduced from two to one within a

156 area of tourism (Bejder et al., 2006b, Higham and Bejder, 2008). In light of the research findings presented here, and the detrimental impacts of dolphin-human interactions documented elsewhere, it would be prudent to restrict the exposure of spinner dolphins to human activities off the Kona coast. Unlike New Zealand and

Australia, cetacean-watching in Hawaii is not licensed and is managed under the more general framework of Marine Mammal Protection Act (MMPA) (Roman et al.,

2013); restricting human-dolphin interactions by licensing is therefore not feasible.

What other options are available to protect cetacean populations against impact(s) from human activities within the framework of the MMPA? (Chapter 2; Tyne et al.,

2014b). The current management intervention being considered for reducing the exposure of spinner dolphins to human exposure is the implementation of time- area closures in the four spinner dolphin resting bays (Chapter 5; Tyne et al., 2015,

NOAA, 2005). This management approach appears to be the most appropriate given the available legislative framework and the context in which human-dolphin interactions are occurring in the bays.

157

7 Conclusions

Decision making for the conservation and management of wildlife populations is critical and is most effective when based on rigorous scientific investigations and rigorous monitoring regimes. In my thesis, I have provided an assessment of the abundance, survival, habitat use and cumulative exposure of human activities on the Hawaii Island spinner dolphin stock. This was achieved through four discrete research components: a robust sampling regime in combination with capture- recapture models to estimate abundance and survival rates; providing an evaluation of the ability to detect changes (decline/increase) in population abundance from various sampling scenarios; quantifying the important habitat for resting spinner dolphins; and determining the cumulative exposure of spinner dolphins to human activities. This research provides a strong scientific foundation for informing management decisions for developing an approach to mitigate the effects of human interactions on the Hawaii Island spinner dolphin stock.

The rigorous systematic sampling regime developed here, in combination with capture-recapture models, was designed specifically to estimate the abundance and the first survival rates for this stock of spinner dolphins (Chapter 3; Tyne et al.,

2014a; Chapter 4). The estimated abundance and survival rates for two consecutive years were very similar (Chapter 3; Tyne et al., 2014a; Chapter 4). Both abundance estimates were much lower (by about 35%) than any estimates previously reported for this stock (Chapter 3; Tyne et al., 2014a). While direct comparisons between previously reported abundance estimates and those presented here should be made with caution, the much lower estimates for the number of dolphins off the

158

Kona Coast (Chapter 4) documented here are a concern. The management plans for this species should be reviewed in light of these estimates.

The ability to detect small changes (decline/increase) in abundance should be an integral part of an effective population monitoring strategy (Thompson et al., 2000,

Wilson et al., 1999). Long-lived animals, however, are a management challenge, as their slow growth rate and low fecundity can lead to slow recovery from disturbance. These life-history characteristics require that monitoring is carried out over tens of years to understand trends in abundance, particularly to detect small rates of change that could be sufficient to reduce the population to seriously low levels before a change is detected, and management intervenes. Here, an increase in sampling effort increased the precision of abundance estimates and thus reduced the time taken to detect a change in abundance (Chapter 4). However, even with the most intensive sampling effort and with a 95% power of detection, the spinner dolphin stock may be in a precipitous decline before a change is detected (Chapter

4). In light of the prolonged time period to detect change and the potential for a large stock decline prior to detection, the precautionary principle suggests that a lower power level (e.g., 80%) should be used as a potential indicator for management intervention. This reduces the time for detection of a change from 12 to nine years and from seven to six years, with monitoring intervals of three years and one year respectively, and depending on the size of the population, rate of decline and precision of the estimates.

159

Identifying habitat that is important to wildlife populations is a key consideration in conservation management. To date, studies evaluating resting habitat of spinner dolphins have focussed on areas inside bays only. My study integrated both boat- based and land-based, group focal-follow sampling regimes inside and outside bays and gradient boosting Generalised Additive Models (GAMs) to identify habitat features that contribute to the occurrence of resting spinner dolphins in coastal waters off Hawaii Island (Chapter 5; Tyne et al., 2015). The results show very clearly that spinner dolphins are most likely to rest inside bays over a sandy substrate and between 10 am and 2 pm. They also show that should spinner dolphins be displaced from resting bays, they are unlikely to engage in rest behaviour elsewhere, which further demonstrates the significance of the bays as critical habitat.

These bays are also subject to considerable use by humans, some of which seek out spinner dolphins for prolonged close-up encounters (Chapter 6; Heenehan et al.,

2015). Concerns have been raised regarding the effect that repeated human- dolphin interactions may have on spinner dolphins (Courbis and Timmel, 2009,

Delfour, 2007). The cumulative exposure of spinner dolphins to human activities may affect their vital rates (Christiansen and Lusseau, 2015, National Research

Council, 2005), alter their energetic budget (Williams et al., 2006) and ultimately affect population viability (Bejder et al., 2006b, Lusseau et al., 2006). In Chapter 6, I demonstrate that dolphins were chronically exposed to human activities during

83% of observations, a magnitude significantly higher than reported for other dolphin species around the world (Chapter 6). Symons et al. (University of

160

Aberdeen unpublished data) showed that spinner dolphins need to be in a resting state during at least 60% of their day-time activity budget to maintain a positive activity budget. Results from my study found that individual spinner dolphins spent between 49.5 and 69.4% of their time resting (mean=61.7%, SD=6.5; Chapter 6).

Therefore, some individuals may be in energetic deficit and, perhaps more importantly, may be deprived of resting opportunities with ramifications for their cognitive abilities (Cirelli and Tononi, 2008, Lima et al., 2005).

Sleep deprivation has induced an increase in sleep pressure (the need to sleep) in dolphins (Oleksenko et al., 1992), impaired cognitive and decision making abilities

(Ganguly-Fitzgerald et al., 2006, Rattenborg et al., 2004, Tartar et al., 2006), and in extreme cases death (e.g. rats, Rechtschaffen et al., 1983). The decision by free- ranging animals on where and when to rest is influenced by the risk of predation

(Lima and Dill, 1990, Lima et al., 2005). To reduce predation risk spinner dolphins gather in sheltered bays to rest (Norris et al., 1994, Chapter 5; Tyne et al., 2015,

Norris and Dohl, 1980). However, in these same bays, dolphins experience chronic exposure to human interactions (Chapter 6), which may alter their state of rest to one of a more vigilant nature, i.e. lighter sleep (Lima et al., 2005). Depriving dolphins of deep rest may reduce the recovery time from their foraging induced vigilance decrement (Dukas and Clark, 1995), and could result in impaired cognitive abilities (Rattenborg et al., 2004, Tartar et al., 2006, Cirelli and Tononi, 2008), leading to an increased vulnerability to predation (Lima et al., 2005), reduced foraging efficiency and ultimately reduced fecundity and population viability.

161

Hawaii Island spinner dolphins live in small (Chapter 3; Tyne et al., 2014a; Chapter

4) genetically-isolated (Andrews et al., 2010) stocks, and rely on sheltered bays to rest and socialise during the day (Norris et al., 1994, Chapter 5; Tyne et al., 2015).

They are chronically exposed to human activities that may deprive them of rest, impair cognitive abilities, increase predation risk, decrease foraging efficiency, and ultimately effect population viability (Chapter 6). Elsewhere, dolphin communities with considerably less exposure to human activities (Chapter 6), have had their natural behavioural patterns disrupted (Bejder et al., 2006a, Constantine et al.,

2004, Christiansen et al., 2010), and energy budgets effected (Williams et al., 2006) which has led to longer-term strategies such as the avoidance of important habitats

(Lusseau, 2004), and subsequently to biologically negative impacts on populations

(Lusseau et al., 2006, Bejder et al., 2006b, National Research Council, 2005). The considerable and concurrent use of the resting bays for recreational, subsistence and tourism purposes, (Heenehan et al., 2015) must now be of major concern for management.

7.1 Concluding remarks In conclusion, my research documented that a) the genetically isolated Hawaii

Island spinner dolphin stock is small and less numerous than previously thought; b) dolphins prefer sheltered bays for rest and are unlikely to rest outside bays; c) dolphins are chronically exposed to humans inside resting bays; and d) the spinner dolphin stock may be in a precipitous decline before a change is detected. The results from my study provide NOAA and the NFMS in Hawaii with scientific

162 information to evaluate various management options, and develop an effective management strategy, to protect spinner dolphins from negative effects of human activities.

163

Appendix I: Peer-reviewed publications during candidature

1. de Freitas, M., Jensen, F.H., Tyne, J.A., Bejder, L., and Madsen, P.T. (2015)

Echolocation parameters of Australian humpback dolphins (Sousa sahulensis)

and Indo-Pacific bottlenose dolphins (Tursiops aduncas) in the wild. Journal of

the Acoustical Society of America.

2. Tyne, J.A., Johnston, D.W., Rankin, R., Loneragan, N.R. and Bejder, L. (2015). The

importance of spinner dolphin (Stenella longirostris) resting habitat:

Implications for management. Journal of Applied Ecology. doi: 10.1111/1365-

2664.12434

3. Heenehan, H., Basurto, X., Bejder, L., Tyne, J.A., Higham, J. and Johnston, D.W.

(2015). Using Ostrom’s common pool research theory to build and integrate

ecosystem-based sustainable cetacean tourism systems in Hawai`i. Journal of

Sustainable Tourism. 23(4): 536-556. doi: 10.1080/09669582.2014.986490

4. Tyne, J.A., Loneragan, N.R. and Bejder, L. (2014). The use of area-time closures

as a tool to manage cetacean-watch tourism. In Whale-watching, sustainable

tourism and ecological management (eds J. Higham, L.Bejder and R. Williams),

pp. 242-260. Cambridge University Press, Cambridge.

5. Anderson, D., Kobryn, H., Norman, B., Bejder L., Tyne, J.A. and Loneragan, N.R.

2014. Spatial and temporal patterns of nature-based tourism interactions with

whale sharks (Rhincodon typus) at Ningaloo Reef, Western Australia. Estuarine,

Coastal and Shelf Science 148: 109-119.

6. Tyne, J.A., Pollock, K.H., Johnston, D.W. and Bejder, L. (2014). Abundance and

Survival Rates of the Hawaii Island Associated Spinner Dolphin (Stenella

longirostris) Stock. PLoS ONE 9, e86132. doi:10.1371/journal.pone.0086132

164

7. Allen SJ, Tyne JA, Kobryn HT, Bejder L, Pollock KH and Loneragan NR. (2014)

Patterns of Dolphin Bycatch in a North-Western Australian Trawl Fishery. PLoS

ONE 9(4): e93178. doi: 10.1371/journal.pone.0093178.

8. Thorne, L. H., Johnston, D.J., Urban, D.L., Tyne, J., Bejder, L., Baird, R.W., Yin, S.,

Rickards, S.H., Deakos, M., Mobley, J.R. Jr., Pack, A.A. and, Chapla-Hill, M. (2012)

Predictive modeling of spinner dolphin (Stenella longirostris) resting habitat in

the main Hawaiian Islands. PLoS ONE. 7(8): e43167.

doi:10.1371/journal.pone.0043167

9. Tyne JA, Loneragan NR, Kopps AM, Allen, SJ, 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.

10. Tyne, J.A., Loneragan, N., Krützen, M., Allen, S., and Bejder, L. (2010) An

integrated data management and video system to sample aquatic benthos.

Journal of Marine and Freshwater Research. 61: 1023-1028.

165

Appendix II: Model selection tables

POPAN models fitted to the capture histories of Hawaiian spinner dolphins to estimate population parameters in Kauhako Bay along the Kona coast of Hawaii Island between September 2010 and August 2011 ĉ = 1.35.

Model QAICc ∆QAICc QAICc Weight 휑(. )푝(푡)푏(푡) 468.0579 0.0000 0.98968 휑(푡)푝(푡)푏(푡) 477.1855 9.1276 0.01031 휑(푡)푝(. )푏(푡) 493.6691 25.6112 0.00000 휑(. )푝(. )푏(푡) 561.0855 93.0276 0.00000

Model QAICc ∆QAICc QAICc Weight 휑(. )푝(푡)푏(푡) 1682.9801 0.0000 0.60027 휑(푡)푝(푡)푏(푡) 1683.7933 0.8132 0.39973 휑(푡)푝(. )푏(푡) 1759.5230 76.5429 0.00000 휑(. )푝(. )푏(푡) 1812.0369 129.0568 0.00000

166

References

Akaike, H. 1974. A new look at the statistical model identification. Automatic Control, IEEE Transactions on, 19, 716-723. Allen, S., Smith, H., Waples, K. & Harcourt, R. 2007. The voluntary code of conduct for dolphin watching in Port Stephens, Australia: is self-regulation an effective management tool? Journal of Cetacean Resource Management, 9, 159-166. Allen, S. J., Cagnazzi, D. D., Hodgson, A. J., Loneragan, N. R. & Bejder, L. 2012. Tropical inshore dolphins of north-western Australia: Unknown populations in a rapidly changing region. Pacific Conservation Biology, 18, 56-63. Allen, S. J., Tyne, J. A., Kobryn, H. T., Bejder, L., Pollock, K. H. & Loneragan, N. R. 2014. Patterns of Dolphin Bycatch in a North-Western Australian Trawl Fishery. PLoS ONE, 9, e93178. Alter, E. S., Simmonds, M. P. & Brandon, J. R. 2010. Forecasting the consequences of climate-driven shifts in human behavior on cetaceans. Marine Policy, 34, 943-954. Altmann, J. 1974. Observational study of behavior: sampling methods. Behaviour, 49, 227- 267. Anderson, D. R., Burnham, K. P. & White, G. C. 1994. AIC model selection in overdispersed capture-recapture data. Ecology, 75, 1780-1780. Andrews, K. R. 2009. Barriers to gene flow in the spinner dolphin (Stenella longirostris). PhD dissertation for the Department of Zooology. University of Hawai'i. Andrews, K. R., Karczmarski, L., Au, W. W. L., Rickards, S. H., Vanderlip, C. A., Bowen, B. W., Gordon Grau, E. & Toonen, R. J. 2010. Rolling stones and stable homes: social structure, habitat diversity and population genetics of the Hawaiian spinner dolphin (Stenella longirostris). Molecular Ecology, 19, 732-748. Armstrong, J. B., Schindler, D. E., Ruff, C. P., Brooks, G. T., Bentley, K. E. & Torgersen, C. E. 2013. Diel horizontal migration in streams: Juvenile fish exploit spatial heterogeneity in thermal and trophic resources. Ecology, 94, 2066-2075. Aschettino, J. M., Baird, R. W., McSweeney, D. J., Webster, D. L., Schorr, G. S., Huggins, J. L., Martien, K. K., Mahaffy, S. D. & West, K. L. 2012. Population structure of melon- headed whales (Peponocephala electra) in the Hawaiian Archipelago: Evidence of multiple populations based on photo identification. Marine Mammal Science, 28, 666-689. Ashe, E., Noran, D. P. & Williams, R. 2010. Animal behaviour and marine protected areas: incorporating behavioural data into the selection of marine protected areas for an endangered killer whale population. Animal Conservation, 13, 196-203. Baird, R. W., Gorgone, A. M., McSweeney, D. J., Webster, D. L., Salden, D. R., Deakos, M. H., Ligon, A. D., Schorr, G. S., Barlow, J. & Mahaffy, S. D. 2008. False killer whales (Pseudorca crassidens) around the main Hawaiian Islands: Long-term site fidelity, inter-island movements, and association patterns. Marine Mammal Science, 24, 591-612. Ballantyne, R., Packer, J. & Falk, J. 2010. Visitors' learning for environmental sustainability: Testing short- and long-term impacts of wildlife tourism experiences using structural equation modelling. Tourism Management, doi: 10.1016/j.tourman.2010.11.003 Balmford, A., Beresford, J., Green, J., Naidoo, R., Walpole, M. & Manica, A. 2009. A Global Perspective on Trends in Nature-Based Tourism. PLoS Biol, 7, e1000144. Barlow, J. 2006. Cetacean abundance in Hawaiian waters estimated from a summer/fall survey in 2002. Marine Mammal Science, 22, 446-464. Beale, C. M. & Monaghan, P. 2004. Behavioural responses to human disturbance: a matter of choice? Animal Behaviour, 68, 1065-1069.

167

Bearzi, G., Politi, E. & di Sciara, G. N. 1999. Diurnal behavior of free-ranging bottlenose dolphins in the Kvarnerić (Northern Adriatic Sea)1. Marine Mammal Science, 15, 1065-1097. 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. Bearzi, G., Pierantonio, N., Bonizzoni, S., Notarbartolo di Sciara, G. & Demma, M. 2010. Perception of a cetacean mass stranding in Italy: the emergence of compassion. Aquatic Conservation: Marine and Freshwater Ecosystems, 20, 644-654. Beasley, I., Bejder, L. & Marsh, H. 2010. Dolphin-watching tourism in the Meekong River, Cambodia: A case study of economic interests influencing conservation. International Whaling Commissions 62nd Annual Meeting in Agadir, Morroco, 2010. Bejder, L., Dawson, S. M. & Harraway, J. A. 1999. Responses by hector's dolphins to boats and swimmers in Porpoise Bay, New Zealand. Marine Mammal Science, 15, 738- 750. Bejder, L. & Samuels, A. 2003. Evaluating the effects of nature-based tourism on cetaceans. In: N. M. Gales, M. Hindell & R. Kirkwood (eds.) Marine mammals and humans: Towards a sustainable balance. Collingwood, Australia: CSIRO Publishing. Bejder, L., Samuels, A., Whitehead, H. & Gales, N. 2006a. Interpreting short-term behavioural responses to disturbance within a longitudinal perspective. Animal Behaviour, 72, 1149-1158. Bejder, L., Samuels, A., Whitehead, H., Gales, N., Mann, J., Connor, R., Heithaus, M., Watson-Capps, J., Flaherty, C. & Krützen, M. 2006b. Decline in relative abundance of bottlenose dolphins exposed to long-term disturbance. Conservation Biology, 20, 1791-1798. Bejder, L., Samuels, A., Whitehead, H., Finn, H. & Allen, S. 2009. Impact assessment research: use and misuse of habituation, sensitisation and tolerance in describing wildlife responses to anthropogenic stimuli. Marine Ecology Progress Series, 395, 177-185. Bejder, L., Hodgson, A., Loneragan, N. & Allen, S. 2012. Coastal dolphins in North-Western Australia: The need for re-evaluation of species listings and short-comings in the environmental impact assessment process. Pacific Conservation Biology, 18, 22-25. Beniot-Bird, K. J. & Au, W. W. L. 2003. Prey dynamics affect foraging by a pelagic predator (Stenella longirostris) over a range of spatial and temporal scales. Behavioral Ecology and Sociobiology, 53, 364-373. Benoit-Bird, K. J. & Au, W. W. L. 2009. Cooperative prey herding by the pelagic dolphin, Stenella longirostris. The Journal of the Acoustical Society of America, 125, 125-137. Borchers, D. L. & Efford, M. G. 2008. Spatially Explicit Maximum Likelihood Methods for Capture–Recapture Studies. Biometrics, 64, 377-385. Braithwaite, J. E., Meeuwig, J. J. & Jenner, K. C. S. 2012. Estimating Cetacean Carrying Capacity Based on Spacing Behaviour. PLoS ONE, 7, e51347. Brown, A. M., Kopps, A. M., Allen, S. J., Bejder, L., Littleford-Colquhoun, B., Parra, G. J., Cagnazzi, D., Thiele, D., Palmer, C. & Frère, C. H. 2014. Population Differentiation and Hybridisation of Australian Snubfin (Orcaella heinsohni) and Indo-Pacific Humpback (Sousa chinensis) Dolphins in North-Western Australia. PLoS ONE, 9, e101427. Bühlmann, P. & Hothorn, T. 2007. Boosting Algorithms: Regularization, Prediction and Model Fitting. Statistical Science, 22, 477--505.

168

Burnham, K. P., Anderson, D. R., White, G. C., Brownie, C. & Pollock, K. H. 1987. Design and analysis methods of fish survival experiments based on release-recapture. American Fisheries Society Monograph, 5. Cantor, M., Wedekin, L. L., Daura-Jorge, F. G., Rossi-Santos, M. R. & Simões-Lopes, P. C. 2012. Assessing population parameters and trends of Guiana dolphins (Sotalia guianensis): An eight-year mark-recapture study. Marine Mammal Science, 28, 63- 83. Carlson, C. 2009. A review of whale watch guidelines and regulations around the world version 2009. Carretta, J. V., Oleson, E., Weller, D. W., Lang, A. R., Forney, K. A., Baker, J., Hanson, B., Martien, K., M, M., Orr, A. J., Huber, H., Lowry, M. S., Barlow, J., Lynch, D., Carswell, L., Brownwell Jr, R. L. & Mattila, D. K. 2013. U.S. Pacific Marine Mammal Stock Assessment: 2013. U.S. DEPARTMENT OF COMMERCE, National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Southwest Fisheries Science Center. Chilvers, B. L., Lawler, I. R., Macknight, F., Marsh, H., Noad, M. & Paterson, R. 2005. Moreton Bay, Queensland, Australia: an example of the co-existence of significant marine mammal populations and large-scale coastal development. Biological Conservation, 122, 559-571. Christensen, A., Rowe, S. & Needham, M. D. 2007. Value orientations, awareness of consequences, and participation in a whale watching education program in Oregon. Human Dimensions of Wildlife, 12, 275-279. Christiansen, F., Lusseau, D., Stensland, E. & Berggren, P. 2010. Effects of tourist boats on the behaviour of Indo-Pacific bottlenose dolphins off the south coast of Zanzibar. Endangered Species Research, 11, 91-99. Christiansen, F., Rasmussen, M. H. & Lusseau, D. 2014. Inferring energy expenditure from respiration rates in minke whales to measure the effects of whale watching boat interactions. Journal of Experimental Marine Biology and Ecology, 459, 96-104. Christiansen, F. & Lusseau, D. 2015. Linking Behavior to Vital Rates to Measure the Effects of Non-Lethal Disturbance on Wildlife. Conservation Letters, n/a-n/a. Cirelli, C. & Tononi, G. 2008. Is sleep essential? PLoS Biol, 6, e216. Cisneros-Montemayor, A. M., Sumaila, U. R., Kaschner, K. & Pauly, D. 2010. The global potential for whale watching. Marine Policy, 34, 1273-1278. Ciuti, S., Northrup, J. M., Muhly, T. B., Simi, S., Musiani, M., Pitt, J. A. & Boyce, M. S. 2012. Effects of Humans on Behaviour of Wildlife Exceed Those of Natural Predators in a Landscape of Fear. PLoS ONE, 7, e50611. Clapham, P. J. 2000. The humpback whale: seasonal feeding and breeding in a baleen whale. In: J. Mann, P. L. Tyack, R. Connor & H. Whitehead (eds.) Cetacean Societies. University of Chicago Press. Clark, C., Ellison, W., Southall, B., Hatch, L., Parijs, S. V., Frankel, A. & Ponirakis, D. 2009. Acoustic masking in marine ecosystems: intuitions, analysis, and implication. Marine Ecology Progress Series, 395, 201-222. Clutton-Brock, T. & Sheldon, B. C. 2010. Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology & Evolution, 25, 562-573. Congdon, J. D., Dunham, A. E. & Van Loben Sels, R. C. 1993. Delayed Sexual Maturity and Demographics of Blanding's Turtles (Emydoidea blandingii): Implications for Conservation and Management of Long-Lived Organisms. Conservation Biology, 7, 826-833.

169

Constantine, R. 2001. Increased avoidance of swimmers by wild bottlenose dolphins (Tursiops truncatus) due to long-term exposure to swim-with-dolphin tourism. Marine Mammal Science, 17, 689-702. Constantine, R., Brunton, D. H. & Dennis, T. 2004. Dolphin-watching tour boats change bottlenose dolphin (Tursiops truncatus) behaviour. Biological Conservation, 117, 299-307. Corkeron, P. J., Van Parijs, S. M., John, H. S., Karl, K. T. & Steve, A. T. 2001. Marine Mammal Migrations and Movement Patterns. Encyclopedia of Ocean Sciences. Oxford: Academic Press. Corkeron, P. J. 2004. Whale Watching, Iconography, and Marine Conservation. Conservation Biology, 18, 847-849. Corkeron, P. J. 2006. How shall we watch whales? In: D. M. Lavigne (ed.) Gaining Ground: In Pursuit of Ecological Sustainability. Guelph, Canada: International Fund for Animal Welfare. Courbis, S. 2007. Effect of spinner dolphin presence on level of swimmer and vessel activity in Hawai'ian bays. Tourism in Marine Environments, 4, 1-14. Courbis, S. & Timmel, G. 2009. Effects of vessels and swimmers on behavior of Hawaiian spinner dolphins (Stenella longirostris) in Kealake'akua, Honaunau, and Kauhako bays, Hawai'i. Marine Mammal Science, 25, 430-440. Cox, D. R. 1983. Some remarks on overdispersion. Biometrika, 70, 269-274. Culik, B. M. & Wilson, R. P. 1995. Penguins disturbed by tourists. Nature, 376, 301-302. Currey, R. J. C., Dawson, S. M., Slooten, E., Schneider, K., Lusseau, D., Boisseau, O. J., Haase, P. & Williams, J. A. 2009. Survival rates for a declining population of bottlenose dolphins in Doubtful Sound, New Zealand: an information theoretic approach to assessing the role of human impacts. Aquatic Conservation: Marine and Freshwater Ecosystems, 19, 658-670. Danil, K., Maldini, D. & Marten, K. 2005. Patterns of use of Maku'a Beach, O'ahu, Hawai'i by spinner dolphins (Stenella longirostris) and potential effects of swimmers on their behavior. Aquatic Mammals, 31, 403-412. Dans, S. L., Crespo, E. A., Pedraza, S. N., Degrati, M. & Garaffo, G. V. 2008. Dusky dolphin and tourist interaction: effect on diurnal feeding behavior. Marine Ecology Progress Series, 369, 287-296. Delfour, F. 2007. Hawaiian spinner dolphins and the growing dolphin watching activity in Oahu. Journal of the Marine Biological Association of the UK, 87, 109-112. Department of Parks and Wildlife 2013. Lalang-garram / Camden Sound Marine Park management plan 73 2013–2023. Department of Parks and Wildlife, Perth. Dill, L. M. 1987. Animal decision making and its ecological consequences: the future of aquatic ecology and behaviour. Canadian Journal of Zoology, 65, 803-811. Dudley, N. 2008. Guidelines for Applying Protected Area Management Categories, Gland, Switzerland, IUCN. Dukas, R. & Clark, C. W. 1995. Sustained vigilance and animal performance. Animal Behaviour, 49, 1259-1267. Edgar, G. J., Stuart-Smith, R. D., Willis, T. J., Kininmonth, S., Baker, S. C., Banks, S., Barrett, N. S., Becerro, M. A., Bernard, A. T. F., Berkhout, J., Buxton, C. D., Campbell, S. J., Cooper, A. T., Davey, M., Edgar, S. C., Forsterra, G., Galvan, D. E., Irigoyen, A. J., Kushner, D. J., Moura, R., Parnell, P. E., Shears, N. T., Soler, G., Strain, E. M. A. & Thomson, R. J. 2014. Global conservation outcomes depend on marine protected areas with five key features. Nature, 506, 216-220. Ferretti, F., Worm, B., Britten, G. L., Heithaus, M. R. & Lotze, H. K. 2010. Patterns and ecosystem consequences of shark declines in the ocean. Ecology Letters, 13, 1055- 1071.

170

Frid, A. & Dill, L. M. 2002. Human-caused disturbance stimuli as a form of predation risk. Conservation Ecology, 6, 94-109. Friday, N. A., Smith, T. D., Stevick, P. T. & Allen, J. 2000. Measurement of photographic quality and individual distinctiveness for the photographic identification of humpback whales, Megaptera novaeangliae. Marine Mammal Science, 16, 355- 374. Friedman, J., Hastie, T. & Tibshirani, R. 2000. Additive logistic regression: a statistical view of boosting. The Annals of Statistics, 28, 337-407. Gailey, G. & Ortega-Ortiz, J. G. 2002. A note on a computer-based system for theodolite tracking of cetaceans. Journal of Cetacean Resource Management, 4, 213-218. Gales, N. J., Kasuya, T., Clapham, P. J. & Brownell, R. L. 2005. Japan's whaling plan under scrutiny. Nature, 435, 883-884. Ganguly-Fitzgerald, I., Donlea, J. & Shaw, P. J. 2006. Waking Experience Affects Sleep Need in Drosophila. Science, 313, 1775-1781. Garrod, B. & Fennell, D. A. 2004. An analysis of whale watching codes of conduct. Annals of Tourism Research, 31, 334-352. Gero, S., Bejder, L., Whitehead, H., Mann, J. & Connor, R. C. 2005. Behaviourally specific preferred associations in bottlenose dolphins, Tursiops spp. Canadian Journal of Zoology, 83, 1566-1573. Gerrodette, T. 1987. A Power Analysis for Detecting Trends. Ecology, 68, 1364-1372. Gill, J. A., Norris, K. & Sutherland, W. J. 2001. Why behavioural responses may not reflect the population consequences of human disturbance. Biological Conservation, 97, 265-268. Goetz, K., Montgomery, R., Ver Hoef, J., Hobbs, R. & Johnson, D. 2012. Identifying essential summer habitat of the endangered beluga whale Delphinapterus leucas in Cook Inlet, Alaska. Endangered Species Research, 16, 135-147. Gormley, A. M., Dawson, S. M., Slooten, E. & Bräger, S. 2005. Capture-recapture estimates of hector's dolphin abundance at Banks Peninsula, New Zealand. Marine Mammal Science, 21, 204-216. Gormley, A. M., Slooten, E., Dawson, S., Barker, R. J., Rayment, W., du Fresne, S. & Bräger, S. 2012. First evidence that marine protected areas can work for marine mammals. Journal of Applied Ecology, 49, 474-480. Gowans, S. & Whitehead, H. 2001. Photographic identification of northern bottlenose whales (Hyperoodon Ampullatus): sources of heterogeneity from natural marks. Marine Mammal Science, 17, 76-93. Guidetti, P., Milazzo, M., Bussotti, S., Molinari, A., Murenu, M., Pais, A., Spanò, N., Balzano, R., Agardy, T., Boero, F., Carrada, G., Cattaneo-Vietti, R., Cau, A., Chemello, R., Greco, S., Manganaro, A., Notarbartolo di Sciara, G., Russo, G. F. & Tunesi, L. 2008. Italian marine reserve effectiveness: Does enforcement matter? Biological Conservation, 141, 699-709. Hammond, P. S. 1986. Estimating the size of naturally marked whale populations using capture-recapture techniques. Report of the International Whaling Commission (special issue) 8, 252-282. Hammond, P. S., Mizroch, S. A. & Donovan, G. P. 1990. Individual recognition of cetaceans: Use of photo-identification and other techniques to estimate population parameters. Report of the International Whaling Commission (Special Issue 12). Harwood, J. & Wilson, B. 2001. The implications of developments on the Atlantic Frontier for marine mammals. Continental Shelf Research, 21, 1073-1093. Hastie, T., Tibshirani, R. & Friedman, J. 2009. The Elements of Statistical Learning: Data Mining, Inference, and Prediction. 2nd edn, New York, Springer.

171

Heenehan, H., Basurto, X., Bejder, L., Tyne, J., Higham, J. E. S. & Johnston, D. W. 2015. Using Ostrom's common-pool resource theory to build toward an integrated ecosystem- based sustainable cetacean tourism system in Hawai`i. Journal of Sustainable Tourism, 23, 536-556. 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., Frid, A., Wirsing, A. J. & Worm, B. 2008. Predicting ecological consequences of marine top predator declines. Trends in Ecology & Evolution, 23, 202-210. Higham, J. & Carr, A. 2002. Ecotourism visitor experiences in Aotearoa/New Zealand: challenging the environmental values of visitors in pursuit of pro-environmental behaviour. Journal of Sustainable Tourism, 10, 277-294. Higham, J. & Lusseau, D. 2007. Defining critical habitats: the spatio-ecological approach to managing tourism--wildlife interactions. Critical Issues in Ecotourism. Oxford: Butterworth-Heinemann. Higham, J., Bejder, L. & Williams, R. 2014. Whale-watching: Sustainable Tourism and Ecological Management, Cambridge, UK, Cambridge University Press. Higham, J. E. S., Bejder, L. & Lusseau, D. 2008. An integrated and adaptive management model to address the long-term sustainability of tourist interactions with cetaceans. Environmental Conservation, 35, 294-302. Higham, J. E. S. & Bejder, L. 2008. Managing Wildlife-based Tourism: Edging Slowly Towards Sustainability? . Current Issues in Tourism, 11, 75-83. Hilborn, R., Stokes, K., Maguire, J.-J., Smith, T., Botsford, L. W., Mangel, M., Orensanz, J., Parma, A., Rice, J., Bell, J., Cochrane, K. L., Garcia, S., Hall, S. J., Kirkwood, G. P., Sainsbury, K., Stefansson, G. & Walters, C. 2004. When can marine reserves improve fisheries management? Ocean & Coastal Management, 47, 197-205. Hinch, P. R. & De Santo, E. M. 2010. Factors to consider in evaluating the management and conservation effectiveness of a whale sanctuary to protect and conserve the North Atlantic right whale (Eubalaena glacialis). Marine Policy, In Press, Corrected Proof. Hofner, B. 2011. Boosting in structured additive models. PhD, Ludwig-Maximilians- Universität München. Hofner, B., Mayr, A., Robinzonov, N. & Schmid, M. 2014. Model-based boosting in R: a hands-on tutorial using the R package mboost. Computational Statistics, 29, 3-35. Holdo, R. M., Fryxell, J. M., Sinclair, A. R. E., Dobson, A. & Holt, R. D. 2011. Predicted Impact of Barriers to Migration on the Serengeti Wildebeest Population. PLoS ONE, 6, e16370. Hothorn, T., Bühlmann, P., Kneib, T., Schmid, M. & Horfner, B. 2010a. Model-based Boosting 2.0. Journal of Machine Learning Research, 11, 2109-2113. Hothorn, T., Müller, J., Schröder, B., Kneib, T. & Brandl, R. 2010b. Decomposing environmental, spatial, and spatiotemporal components of species distributions. Ecological Monographs, 81, 329-347. Hoyt, E. 2001. World wide tourism numbers, expenditures and expanding socioeconomic benefits. A special report from the International Fund for Animal Welfare. Hoyt, E. 2005. Marine protected areas for whales, dolphins, and porpoises : a world handbook for cetacean habitat conservation, London, Earthscan. Hoyt, E. 2007. A blueprint for dolphin and whale watching development, Washington, DC, USA, Humane Society International. Hoyt, E. 2011. Marine protected areas for whales, dolphins, and porpoises : a world handbook for cetacean habitat conservation and planning, London, Earthscan. Hu, W., Boehle, K., Cox, L. & Pan, M. 2009. Economic values of dolphin excursions in Hawaii: A stated choice analysis. Marine Resource Economics, 24, 61-76. IWC 2006. Report of the Scientific Committee. Cetacean Research and Management, 1-160.

172

Jaramillo-Legorreta, A., Rojas-Bracho, L., Brownell, R. L., Read, A. J., Reeves, R. R., Ralls, K. & Taylor, B. L. 2007. Saving the Vaquita: Immediate Action, Not More Data. Conservation Biology, 21, 1653-1655. Jefferson, T. A., Hung, S. K. & Würsig, B. 2009. Protecting small cetaceans from coastal development: Impact assessment and mitigation experience in Hong Kong. Marine Policy, 33, 305-311. Jensen, F. H., Bejder, L., Wahlberg, M., Aguilar Soto, N., Johnson, M. & Madsen , P. T. 2009. Vessel noise effects on delphinid communication. Marine Ecology Progress Series, 395, 161-175. Johnson, C. J., Seip, D. R. & Boyce, M. S. 2004. A quantitative approach to conservation planning: using resource selection functions to map the distribution of mountain caribou at multiple spatial scales. Journal of Applied Ecology, 41, 238-251. Johnston, D. W., Thorne, L. H. & Read, A. J. 2005a. Fin whales Balaenoptera physalus and minke whales Balaenoptera acutorostrata exploit a tidally driven island wake ecosystem in the Bay of Fundy. Marine Ecology Progress Series, 305, 287-295. Johnston, D. W., Westgate, A. J. & Read, A. J. 2005b. Effects of fine-scale oceanographic features on the distribution and movements of harbour porpoises Phocoena phocoena in the Bay of Fundy. Marine Ecology Progress Series, 295, 279-293. Johnston, D. W. & Read, A. J. 2007. Flow-field observations of a tidally driven island wake used by marine mammals in the Bay of Fundy, Canada. Fisheries Oceanography, 16, 422-435. Johnston, D. W., Bowers, M. T., Friedlaender, A. S. & Lavigne, D. M. 2012. The Effects of Climate Change on Harp Seals (Pagophilus groenlandicus). PLoS ONE, 7, e29158. Johnston, D. W. 2014. Vigilance, resilience and failures of science and management. In: J. Higham, L. Bejder & R. Williams (eds.) Whale-watching: Sustainable Tourism and Ecological Management. Cambridge: Cambridge University Press. Jolly, G. M. 1965. Explicit estimates from capture-recapture data with both death and immigration-stochastic model. Biometrika, 52, 225-247. Karanth, K. K. & DeFries, R. 2011. Nature-based tourism in Indian protected areas: New challenges for park management. Conservation Letters, 4, 137-149. Keane, A., Jones, J. P. G., Edwards-Jones, G. & Milner-Gulland, E. J. 2008. The sleeping policeman: understanding issues of enforcement and compliance in conservation. Animal Conservation, 11, 75-82. Keane, A., Ramarolahy, A. A., Jones, J. P. G. & Milner-Gulland, E. J. 2011. Evidence for the effects of environmental engagement and education on knowledge of wildlife laws in Madagascar. Conservation Letters, 4, 55-63. Kelleher, G. 1999. Guidelines for Marine Protected Areas. IUCN, Gland, Switzerland and Cambridge, UK, IUCN Kessler, M. & Harcourt, R. 2010. Aligning tourist, industry and government expectations: A case study from the swim with whales industry in Tonga. Marine Policy, 34, 1350- 1356. Kinzey, D. & Gerrodette, T. 2003. Distance measurements using binoculars from ships at sea: accuracy, precision and effects of refraction Journal of Cetacean Research and Management, 5, 159-171. Lammers, M. O. 2004. Occurrence and behavior of Hawaiian spinner dolphins (Stenella longirostris) along Oahu's leeward and south shores. Aquatic Mammals, 30, 237- 250. Larsen, F. & Hammond, P. S. 2004. Distribution and abundance of West Greenland humpback whales (Megaptera novaeangliae). Journal of Zoology, 263, 343-358.

173

Lauck, T., Clark, C. W., Mangel, M. & Munro, G. R. 1998. Implementing the precautionary principle in fisheries management through marine reserves. Ecological Applications, 8, 72-78. Lebreton, J.-D., Burnham, K. P., Clobert, J. & Anderson, D. R. 1992. Modeling Survival and Testing Biological Hypotheses Using Marked Animals: A Unified Approach with Case Studies. Ecological Monographs, 62, 67-118. Lima, S. L. & Dill, L. M. 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Canadian Journal of Zoology, 68, 619-640. Lima, S. L. 1998. Nonlethal effects in the ecology of predator-prey interactions. Bioscience, 48, 25-34. Lima, S. L., Rattenborg, N. C., Lesku, J. A. & Amlaner, C. J. 2005. Sleeping under the risk of predation. Animal Behaviour, 70, 723-736. Lück, M. 2003. Education on marine mammal tours as agent for conservation--but do tourists want to be educated? Ocean & Coastal Management, 46, 943-956. Lundquist, C. J. & Granek, E. E. 2005. Strategies for successful marine conservation: Integrating socioeconomic, political and scientific factors. Conservation Biology, 19, 1771-1778. Lundquist, D., Gemmell, N. J. & Würsig, B. 2012. Behavioural Responses of Dusky Dolphin Groups (Lagenorhynchus obscurus) to Tour Vessels off Kaikoura, New Zealand. PLoS ONE, 7, e41969. Lusseau, D. 2003a. The effects of tour boats on the behavior of bottlenose dolphins:Using Markov chains to model anthropogenic impacts. Conservation Biology, 17, 1785 - 1793. Lusseau, D. 2003b. Male and female bottlenose dolphins Tursiops spp. have different strategies to avoid interactions with tour boats in Doubtful Sound, New Zealand. Marine Ecology Progress Series, 257, 267-274. Lusseau, D. & Newman, M. E. J. 2004. Identifying the role that animals play in their social networks. Proceedings of the Royal Society B: Biological Sciences, 271, S477-S481. Lusseau, D. 2004. The hidden cost of tourism: detecting long-term effects of tourism using behavioural information. Ecology and Society, 9. Lusseau, D. & Higham, J. E. S. 2004. Managing the impacts of dolphin-based tourism through the definition of critical habitats: the case of bottlenose dolphins (Tursiops spp.) in Doubtful Sound, New Zealand. Tourism Management, 25, 657-667. Lusseau, D. 2005. Residency patten of bottlenose dolphins Tursiops spp. in Milford Sound, New Zealand is related to boat traffic. Marine Ecology Progress Series, 295, 265- 272. Lusseau, D. 2006. The short-term behavioral reactions of bottlenose dolphins to interactions with boats in Doubtful Sound, New Zealand. Marine Mammal Science, 22, 802-818. Lusseau, D., Slooten, L. & Currey, R. J. C. 2006. Unsustainable dolphin-watching tourism in Fiordland, New Zealand. Tourism in Marine Environments, 3, 173-178. Lusseau, D., Bain, E. D., Williams, R. & Smith, J. C. 2009. Vessel traffic disrupts the foraging behaviour of southern resident killer whales Orcinus orca. Endangered Species Research, 6, 211-221. Magurran, A. E., Baillie, S. R., Buckland, S. T., Dick, J. M., Elston, D. A., Scott, E. M., Smith, R. I., Somerfield, P. J. & Watt, A. D. 2010. Long-term datasets in biodiversity research and monitoring: assessing change in ecological communities through time. Trends in Ecology & Evolution, 25, 574-582. Maloney, K. O., Schmid, M. & Weller, D. E. 2012. Applying additive modelling and gradient boosting to assess the effects of watershed and reach characteristics on riverine assemblages. Methods in Ecology and Evolution, 3, 116-128.

174

Mangel, J. C., Alfaro-Shigueto, J., Van Waerebeek, K., Cáceres, C., Bearhop, S., Witt, M. J. & Godley, B. J. 2010. Small cetacean captures in Peruvian artisanal fisheries: High despite protective legislation. Biological Conservation, 143, 136-143. Mann, J. 1999. Behavioral sampling methods for cetaceans: a review and critique. Marine Mammal Science, 15, 102-122. Martien, K. K., Baird, R. W., Hedrick, N. M., Gorgone, A. M., Thieleking, J. L., McSweeney, D. J., Robertson, K. M. & Webster, D. L. 2012. Population structure of island- associated dolphins: Evidence from mitochondrial and microsatellite markers for common bottlenose dolphins (Tursiops truncatus) around the main Hawaiian Islands. Marine Mammal Science, 28, 208-232. Meinshausen, N. & Bühlmann, P. 2010. Stability selection. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 72, 417-473. Meissner, A. M., Christiansen, F., Martinez, E., Pawley, M. D. M., Orams, M. B. & Stockin, K. A. 2015. Behavioural Effects of Tourism on Oceanic Common Dolphins, Delphinus sp., in New Zealand: The Effects of Markov Analysis Variations and Current Tour Operator Compliance with Regulations. PLoS ONE, 10, e0116962. MMPA 1972. Marine Mammal Protection Act of 1972. 16 U.S.C. et seq. and 50 CFR part 216. (1972). Moberg, G. P. 2000. Biological Response to Stress: Implications for Animal Welfare. In: G. P. Moberg & J. A. Mench (eds.) Biology of Animal Stress : Basic Principles and Implications for Animal Welfare. Wallingford, Oxon, GBR: CABI Publishing. Mobley, J. R., Spitz, S. S., Forney, K. A., Grotefendt, R. A. & Forestall, P. H. 2000. Distribution and abundance of odontocete species in Hawaiian waters: preliminary results of 1993-98 aerial surveys Southwest Fisheries Science Center, National Marine Fisheries Service, P.O. Box 271, La Jolla, CA 92038. Müllner, A., Eduard Linsenmair, K. & Wikelski, M. 2004. Exposure to ecotourism reduces survival and affects stress response in hoatzin chicks (Opisthocomus hoazin). Biological Conservation, 118, 549-558. Mustika, P. L. K., Birtles, A. & Marsh, H. 2011. The possible effects of dolphin tourism on coastal cetaceans: A case study from Bali, Indonesia. Determining and quantifying threats to coast cetaceans: A regional collaborative workshop February 22-24. Myers, R. A. & Worm, B. 2003. Rapid worldwide depletion of predatory fish communities. Nature, 423, 280-283. Nagelkerke, N. J. D. 1991. A note on a general definition of the coefficient of determination. Biometrika, 78, 691-692. National Research Council 2005. Marine Mammal Populations and Ocean Noise: Determining When Noise Causes Biologically Significant Effects, Washington DC, National Academy Press. Nellemann, C., Jordhøy, P., Støen, O.-G. & Strand, O. 2000. Cumulative Impacts of Tourist Resorts on Wild Reindeer ( Rangifer tarandus tarandus ) during Winter. New, L. F., Harwood, J., Thomas, L., Donovan, C., Clark, J. S., Hastie, G., Thompson, P. M., Cheney, B., Scott-Hayward, L. & Lusseau, D. 2013. Modelling the biological significance of behavioural change in coastal bottlenose dolphins in response to disturbance. Functional Ecology, 27, 314-322. Nicholson, K., Bejder, L., Allen, S. J., Krützen, M. & Pollock, K. H. 2012. Abundance, survival and temporary emigration of bottlenose dolphins (Tursiops sp.) off Useless Loop in the western gulf of Shark Bay, Western Australia. Marine and Freshwater Research, 63, 1059-1068. NOAA 2005. Protecting spinner dolphins in the main Hawaiian Islands from human activities that cause ‘‘Take,’’ as defined in the Marine Mammal Protection Act and

175

its implementing regulations, or to otherwise adversely affect the dolphins NOAA 051110296–5296–01; I.D.102405A. Norris, K. S. & Dohl, T. P. 1980. Behavior of the Hawaiian spinner dolphin, Stenella longirostris. Fish. Bull., 77, 821-849. Norris, K. S. 1991. Dolphin Days, New York, WW Norton and Co., New York. Norris, K. S., Wursig, B., Wells, S. & Wursig, M. 1994. The Hawaiian Spinner Dolphin, Berkeley, CA, University of California Press. Notarbartolo di Sciara, G., Hanafy, M. H., Fouda, M. M., Afifi, A. & Costa, M. 2008. Spinner dolphin (Stenella longirostris) resting habitat in Samadai Reef (Egypt, Red Sea) protected through tourism management. Journal of the Marine Biological Association of the UK, 89, 211-216. Nowacek, D. P., Thorne, L. H., Johnston, D. W. & Tyack, P. L. 2007. Responses of cetaceans to anthropogenic noise. Mammal Review, 37, 81-115. O'Connor, S., Campbell, R., Cortez, H. & Knowles, T. 2009. Whale Watching Worldwide: tourism numbers, expenditures and expanding economic benefits, a special report from the International Fund for Animal Welfare. Yarmouth, MA, USA. Oleksenko, A. I., Mukhametov, L. M., Polyakova, I. G., Supin, A. Y. & Kovalzon, V. M. 1992. Unihemispheric sleep deprivation in bottlenose dolphins. Journal of Sleep Research, 1, 40-44. Orams, M. B. 1996. Using interpretation to manage nature-based tourism. Journal of Sustainable Tourism, 4, 81-94. Orams, M. B. 1997. The effectivenes of environmental education: Can we turn tourists into 'Greenies'? Progress in Tourism and Hospitality Research, 3. Orams, M. B. 1999. Marine tourism: development, impacts and management, London, Routledge. Ostman-Lind, J., Driscoll-Lind, A. & Rickards, S. 2004. Delphinid abundance, distribution and habitat use off the western coast of the island of Hawai'i. Administrative Report. Southwest Fisheries Science Center Ostman-Lind, J. 2008. Impacts of human activities on spinner dolphins (Stenella longirostris) in their resting areas. Final report to the National Marine Fisheries Service, Pacfic Island Regional Office. Ostman, J. 1994. Social organization and social behavior of Hawaiian spinner dolphins (Stenella longirostris). PhD, University of California, Santa Cruz. Panigada, S., Pesante, G., Zanardelli, M., Capoulade, F., Gannier, A. & Weinrich, M. T. 2006. Mediterranean fin whales at risk from fatal ship strikes. Marine Pollution Bulletin, 52, 1287-1298. Parent, C. & Weatherhead, P. J. 2000. Behavioral and life history responses of eastern massasauga rattlesnakes (Sistrurus catenatus catenatus) to human disturbance. Oecologia, 125, 170-178. Pauly, D., Christensen, V., Guenette, S., Pitcher, T. J., Sumaila, U. R., Walters, C. J., Watson, R. & Zeller, D. 2002. Towards sustainability in world fisheries. Nature, 418, 689-695. Pergams, O. R. W. & Zaradic, P. A. 2008. Evidence for a fundamental and pervasive shift away from nature-based recreation. Proceedings of the National Academy of Sciences, 105, 2295-2300. Perrin, W. F. 1990. Subspecies of Stenella longirostris (Mammalia: Cetacea: Delphinidae). Proceedings of the Biological Society of Washington 103, 453-463. Perrin, W. F., Dolar, M. L. L. & Robineau, D. 1999. Spinner dolphins (Stenella longirostris) of the western Pacific and southeast Asia: Pelagic and shallow-water forms. Marine Mammal Science, 15, 1029-1053.

176

Peters, K. J., Parra, G. J., Skuza, P. P. & Möller, L. M. 2012. First insights into the effects of swim-with-dolphin tourism on the behavior, response, and group structure of southern Australian bottlenose dolphins. Marine Mammal Science, 29, 484-497. Pirotta, E., Merchant, N. D., Thompson, P. M., Barton, T. R. & Lusseau, D. 2015. Quantifying the effect of boat disturbance on bottlenose dolphin foraging activity. Biological Conservation, 181, 82-89. Pollock, K. H., Nichols, J. D., Brownie, C. & Hines, J. E. 1990. Statistical inference for capture- recapture experiments. Wildlife Monographs, 107, 1-97. Quiros, A. L. 2007. Tourist compliance to a Code of Conduct and the resulting effects on whale shark (Rhincodon typus) behavior in Donsol, Philippines. Fisheries Research, 84, 102-108. R Core Team 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/. Rattenborg, N. C., Mandt, B. H., Obermeyer, W. H., Winsauer, P. J., Huber, R., Wikelski, M. & Benca, R. M. 2004. Migratory Sleeplessness in the White-Crowned Sparrow (Zonotrichia leucophrys gambelii). PLoS Biol, 2, e212. Read, A. J., Urian, K. W., Wilson, B. & Waples, D. M. 2003. Abundance of bottlenose dolphins in the bays, sounds, and estuaries of North Carolina. Marine Mammal Science, 19, 59-073. Rechtschaffen, A., Gilliland, M. A., Bergmann, B. M. & Winter, J. B. 1983. Physiological Correlates of Prolonged Sleep Deprivation in Rats. Science, 221, 182-184. Reijnders, P. J. H., Aguilar, A. & Borrell, A. 2009. Pollution and Marine Mammals. In: F. P. Perrin, B. Wã¼Rsig & J. G. M. Thewissen (eds.) Encyclopedia of Marine Mammals second edition. Academic Press. Ripley, B. 2013. ODBC database access. Roman, J., Altman, I., Dunphy-Daly, M. M., Campbell, C., Jasny, M. & Read, A. J. 2013. The Marine Mammal Protection Act at 40: status, recovery, and future of U.S. marine mammals. Annals of the New York Academy of Sciences, 1286, 29-49. Ross, P. S., Barlow, J., Jefferson, T. A., Hickie, B. E., Lee, T., MacFarquhar, C., Christien Parsons, E., Riehl, K. N., Rose, N. A., Slooten, E., Tsai, C.-Y., Wang, J. Y., Wright, A. J. & Chu Yang, S. 2011. Ten guiding principles for the delineation of priority habitat for endangered small cetaceans. Marine Policy, 35, 483-488. Samuels, A. & Spradlin, T. R. 1995. Quantitative behavioural study of bottlenose dolphins in swim-with-dolphin programs in the United States. Marine Mammal Science, 11, 520-544. Scarpaci, C., Dayanthi, N. & Corkeron, P. J. 2003. Compliance with Regulations by “Swim- with-Dolphins” Operations in Port Phillip Bay, Victoria, Australia. Environmental Management, 31, 0342-0347. Schaffar, A., Garrigue, C. & Constantine, R. 2010. Exposure of humpback whales to unregulated whalewatching activities in their main reproductive area in New Caledonia. Journal of Cetacean Research and Management, 11, 147-152. Schwartz, C. J. & Arnason, A. N. 1996. A General Methodology for the Analysis of Capture- Recapture Experiments in Open Populations. Biometrics, 52, 860-873. Seber, G. A. F. 1965. A note on the multiple-recapture census. Biometrika, 52, 249-259. Siegel, J. M. 2008. Do all animals sleep? Trends in Neurosciences, 31, 208-213. Silva, M., Magalhaes, S., Prieto, R., Santos, R. & Hammond, P. 2009. Estimating survival and abundance in a bottlenose dolphin population taking into account transience and temporary emigration. Marine Ecology Progress Series, 392, 263-276. Smith, H. C., Pollock, K., Waples, K., Bradley, S. & Bejder, L. 2013. Use of the Robust Design to Estimate Seasonal Abundance and Demographic Parameters of a Coastal Bottlenose Dolphin (Tursiops aduncus) Population. PLoS ONE, 8, e76574.

177

Smolker, R. A., Richards, A. F., Connor, R. C. & Pepper, J. 1992. Association patterns among bottlenose dolphins in Shark Bay, Western Australia. Behaviour, 123, 38-69. Stalmaster, M. V. & Newman, J. R. 1978. Behavioral Responses of Wintering Bald Eagles to Human Activity. The Journal of Wildlife Management, 42, 506-513. Steckenreuter, A., Harcourt, R. & Möller, L. 2012. Are Speed Restriction Zones an effective management tool for minimising impacts of boats on dolphins in an Australian marine park? Marine Policy, 36, 258-264. Stensland, E. & Berggren, P. 2007. Behavioural changes in female Indo-Pacific bottlenose dolphins in response to boat-based tourism. Marine Ecology Progress Series, 332, 225-234. Stockin, K. A., Lusseau, D., Binedell, V., Wiseman, N. & Orams, M. B. 2008. Tourism affects the behavioural budget of the common dolphin Delphinus sp. in the Hauraki Gulf, New Zealand. Marine Ecology Progress Series, 355, 287-295. Tartar, J. L., Ward, C. P., McKenna, J. T., Thakkar, M., Arrigoni, E., McCarley, R. W., Brown, R. E. & Strecker, R. E. 2006. Hippocampal synaptic plasticity and spatial learning are impaired in a rat model of sleep fragmentation. European Journal of Neuroscience, 23, 2739-2748. Taylor, B. L. & Gerrodette, T. 1993. The Uses of Statistical Power in Conservation Biology: The Vaquita and Northern Spotted Owl. Conservation Biology, 7, 489-500. Taylor, B. L., Martinez, M., Gerrodette, T., Barlow, J. & Hrovat, Y. N. 2007. Lessons from monitoring trends in abundance of marine mammals. Marine Mammal Science, 23, 157-175. Thompson, P. M., Wilson, B., Grellier, K. & Hammond, P. S. 2000. Combining Power Analysis and Population Viability Analysis to Compare Traditional and Precautionary Approaches to Conservation of Coastal Cetaceans. Conservation Biology, 14, 1253- 1263. Thorne, L. H., Johnston, D. W., Urban, D. L., Tyne, J., Bejder, L., Baird, R. W., Yin, S., Rickards, S. H., Deakos, M. H., Mobley, J. R., Jr., Pack, A. A. & Chapla Hill, M. 2012. Predictive modeling of spinner dolphin (Stenella longirostris) resting habitat in the main Hawaiian Islands. PLoS ONE, 7, e43167. Timmel, G., Courbis, S., Sargeant-Green, H. & Markowitz, H. 2008. Effects of human traffic on the movement patterns of Hawaiian spinner dolphins (Stenella longirostris) in Kealakekua Bay, Hawaii. Aquatic Mammals, 34, 402-411. Turvey, S. T., Pitman, R. L., Taylor, B. L., Barlow, J., Akamatsu, T., Barrett, L. A., Zhao, X., Reeves, R. R., Stewart, B. S., Wang, K., Wei, Z., Zhang, X., Pusser, L. T., Richlen, M., Brandon, J. R. & Wang, D. 2007. First human-caused extinction of a cetacean species? Tyack, P. L. 2008. Implications for marine mammals of large-scale changes in the marine acoustic environment. Journal of Mammalogy, 89, 549-558. Tyne, J. A., Pollock, K. H., Johnston, D. W. & Bejder, L. 2014a. Abundance and Survival Rates of the Hawai’i Island Associated Spinner Dolphin (Stenella longirostris) Stock. PLoS ONE, 9, e86132. Tyne, J. A., Loneragan, N. & Bejder, L. 2014b. The use of area–time closures as a tool to manage cetacean-watch tourism. In: J. Higham, L. Bejder & R. Williams (eds.) Whale-watching: Sustainable Tourism and Ecological Management. Cambridge: Cambridge University Press. Tyne, J. A., Johnston, D. W., Rankin, R., Loneragan, N. R. & Bejder, L. 2015. The importance of spinner dolphin (Stenella longirostris) resting habitat: Implications for management doi: 10.1111/1365-2664.12434. Journal of Applied Ecology, 52, 621=630.

178

U.S. Fish and Wildlife Service 2001. Florida Manatee Recovery Plan, (Trichechus manatus latirostris),Third Revision. U.S. Fish and Wildlife Service. Atlanta, Georgia. . Urian, K., Hohn, A. A. & Hansen, L. J. 1999. Status of the photo-identification catalog of coastal bottlenose dolphins of the western North Atlantic. Report of a workshop of catalog contributors. NOAA Technical Memo Urian, K., Gorgone, A., Read, A., Balmer, B., Wells, R. S., Berggren, P., Durban, J., Eguchi, T., Rayment, W. & Hammond, P. S. 2015. Recommendations for photo-identification methods used in capture-recapture models with cetaceans. Marine Mammal Science, 31, 298-321. Vanzella-Khouri, A. Year. The CaMPAM experience: a social network to enhance MPA management in the Wider Caribbean and links to marine mammal conservation. In: First international conference on marine mammal protected areas, 2009 Maui, Hawaii. Wade, P. R. 1998. Calculating limits to the allowable human-caused mortality of cetaceans and pinnipeds. Marine Mammal Science, 14, 1-37. Ward, N. & MacDonald, C. Year. Sister sanctuary: care of marine mammal protected areas beyond borders - an innovative management tool for transboundary species. In: First international conference on marine mammal protected areas, 2009 Maui, Hawaii. 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. White, G. C. & Burnham, K. P. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study, 46, S120-S139. Whitt, A. D. & Read, A. J. 2006. Assessing Compliance to Guidelines By Dolphin-watching Operators in Clearwater, Florida, USA. Tourism in Marine Environments, 3, 117-130. Wiener, C. S., Needham, M. D. & Wilkinson, P. F. 2009. Hawaii's real life marine park: interpretation and impacts of commercial marine tourism in the Hawaiian Islands. Current Issues in Tourism, 12, 489-504. Williams, B. K., Nichols, J. D. & Conroy, M. J. 2002a. Analysis and management of animal populations, Academic Press. Williams, J. A. 1993. The abundance and distribution of bottlenose dolphins (Tursiops truncatus) in Doubtful Sound, New Zealand. Can J Zool, 71, 2080-2088. Williams, R., Trites, A. W. & Bain, D. E. 2002b. Behavioural responses of killer whales (Orcinus orca) to whale-watching boats: opportunistic observations and experimental approaches. Journal of Zoology, 256, 255-270. Williams, R., Lusseau, D. & Hammond, P. S. 2006. Estimating relative energetic costs of human disturbance to killer whales (Orcinus orca). Biological Conservation, 133, 301-311. Williams, R. & Ashe, E. 2007. Killer whale evasive tactics vary with boat number. Journal of Zoology, 272, 390-397. Williams, R. & Thomas, L. 2009. Cost-effective abundance estimation of rare animals: Testing performance of small-boat surveys for killer whales in British Columbia. Biological Conservation, 142, 1542-1547. Williams, R., Lusseau, D. & Hammond, P. S. 2009. The role of social aggregations and protected areas in killer whale conservation: The mixed blessing of critical habitat. Biological Conservation, 142, 709-719. Williams, R., Grand, J., Hooker, S. K., Buckland, S. T., Reeves, R. R., Rojas-Bracho, L., Sandilands, D. & Kaschner, K. 2014. Prioritizing global marine mammal habitats using density maps in place of range maps. Ecography, 37, 212-220.

179

Williams, R. O. B., Hedley, S. L., Branch, T. A., Bravington, M. V., Zerbini, A. N. & Findlay, K. P. 2011. Chilean Blue Whales as a Case Study to Illustrate Methods to Estimate Abundance and Evaluate Conservation Status of Rare Species Ballenas Azules Chilenas como Caso de Estudio para Ilustrar Métodos para Estimas la Abundancia y Evaluar el Estatus de Conservación de Especies Raras. Conservation Biology, 25, 526-535. Wilson, B., Hammond, P. S. & Thompson, P. M. 1999. Estimating size and assessing trends in a coastal bottlenose dolphin population. Ecological Applications, 9, 288-300. Wilson, B., Reid, R. J., Grellier, K., Thompson, P. M. & Hammond, P. S. 2004. Considering the temporal when managing the spatial: a population range expansion impacts protected areas-based management for bottlenose dolphins. Animal Conservation, 7, 331-338. Wolanski, E. & Hamner, W. M. 1988. Topographically controlled fronts in the ocean and their biological influence. Science, 241, 177-181. Wood, S. A., Frasier, T. R., McLeod, B. A., Gilbert, J. R., White, B. N., Bowen, W. D., Hammill, M. O., Waring, G. T. & Brault, S. 2011. The genetics of recolonization: an analysis of the stock structure of grey seals (Halichoerus grypus) in the northwest Atlantic. Canadian Journal of Zoology, 89, 490-497. Worm, B., Davis, B., Kettemer, L., Ward-Paige, C. A., Chapman, D., Heithaus, M. R., Kessel, S. T. & Gruber, S. H. 2013. Global catches, exploitation rates, and rebuilding options for sharks. Marine Policy, 40, 194-204. Würsig, B., Cipriano, F. & Würsig, M. 1991. Dolphin movement patterns: Information from radio and theodolite tracking studies. In: K. S. Norris & K. Pryor (eds.) Dolphin Societies. Los Angeles: University of California Press. Zimmer, W. M. X. & Tyack, P. L. 2007. Repetitive shallow dives pose decompression risk in deep-diving beaked whales. Marine Mammal Science, 23, 888-925.

180