Foraging ecology of sperm whales at Kaikōura
Marta Guerra Bobo
A thesis submitted for the degree of
Doctor of Philosophy
at the University of Otago, Dunedin,
Aotearoa – New Zealand
October 2018
i
Abstract
The submarine canyon of Kaikōura, New Zealand, is a foraging ground for male sperm whales (Physeter macrocephalus). The abundance of whales feeding in this area during summer has almost halved over the last three decades, for reasons that are unknown. The decline may reflect a shift in distribution away from the area, possibly caused by underlying oceanographic or ecological changes. It is therefore important to understand what sustains the whales’ diet and what environmental factors drive their distribution. First, I examined the whales’ food web using stable isotope analyses. Isotope ratios of carbon and nitrogen were measured in samples of sloughed whale skin, potential prey and primary producers.
Generalised additive models suggested that occasional visitors to Kaikōura had more diverse and lower isotope ratios than more frequent visitors (by c. -1‰ δ¹³C and -2‰ δ¹⁵N, n = 90), likely reflecting a range of foraging areas further south and/or offshore. The ultimate contribution of pelagic phytoplankton vs coastal macroalgae to the whales’ food web could not be determined precisely, but there was strong support for it being sustained mostly by pelagic production. Whales appeared to feed on a mixture of squid and demersal fish, and differences in the whales’ isotope ratios between summer and winter suggested seasonal variability in diet. Summer food resources were likely to comprise a high proportion of squid.
Surveys of foraging whales (n = 334) in conjunction with in situ oceanographic sampling (n =
486) were carried out over three years. I used species-distribution models to investigate the topographic and oceanographic characteristics that determine good foraging habitat.
Physical factors that may contribute to aggregating prey seemed particularly important; these included strong thermal stratification in the water column, steep slopes, and slope orientations likely to concentrate food resources through the interaction with local currents.
Habitat preferences differed between summer and winter, consistent with stable isotope ratios and previously recorded patterns in diving behaviour, suggesting seasonal fluctuations in targeted prey. An evaluation of the overlap between foraging habitat and the recently established Hikurangi Marine Reserve suggested little meaningful protection for sperm whales from the potential impacts of fishing. The importance of thermal stratification for foraging habitat in summer suggested that whales may be susceptible to changes in ii
oceanographic conditions of the ecosystem. I investigated the long-term variability in whale abundance during summer in relation to environmental conditions, using abundance estimates from 1990–2017 and remotely sensed oceanographic data. Linear models suggested that warmer sea temperatures in winter, and lower inputs of Subantarctic water to the south of Kaikōura in summer, were correlated with fewer whales foraging in the area, explaining
59% of inter-annual variability. These correlations may be mediated by indirect effects on prey availability, potentially contributing to the decline in whale abundance. Precautionary management of anthropogenic activities affecting the whales and their habitat is necessary to increase the resilience of the population to climate-driven changes in the ecosystem. This is important to ensure the persistence of sperm whales at Kaikōura and protect the canyon ecosystem that supports them.
iii
Sperm whale diving. Kaikōura, New Zealand.
iv
Acknowledgements
This PhD has been an incredible journey and an experience that I will always treasure. It would not have been possible, and no way nearly as enjoyable, without the support from so many people.
First and foremost, a massive thank you to my supervisors: Will Rayment, Steve Dawson, Lucy
Wing and Liz Slooten. Thanks for the wonderful opportunity that this PhD has been. Will, your help and support have been absolutely invaluable. Thank for your endless enthusiasm, thorough and constructive feedback, and guidance with just about everything – from boat driving to study design, from using common sense in interpreting results to where to get the best blueberry- custard-buns. There hasn’t been a time when being busy stopped you from helping, and I am grateful to have learnt so much from you. Steve, thank you for continuing to be such a dedicated and inspiring mentor, for your tremendous support, for teaching me to build hydrophones and much more, and for your extensive advice on science and practical matters. Lucy, thank you so much for your help with stable isotope ecology, for all the time and effort you’ve dedicated to this, and for your detailed feedback. Liz, thanks for your awesome ‘big picture’ advice and overall support.
Thanks to Michael Moore for creating an opportunity for me to meet the sperm whales of
Kaikōura and participate in the research cruise that led to my starting of this particular PhD.
For funding this research, I am extremely grateful to Whale Watch Kaikoura (special thanks to
Roger Williams and Kauahi Ngapora), and also to the New Zealand Whale and Dolphin Trust, the Ministry for Primary Industries and the University of Otago. The research vessel Grampus and a lot of the research equipment were supplied by the New Zealand Whale and Dolphin Trust, without which this project would not have been possible. Thanks also to the University of Otago for supporting me during my research with a PhD scholarship.
For help with fieldwork at Kaikōura I owe a huge debt of gratitude to everyone that generously gave their time to join me in the early mornings, long days, pulling and winding of CTDs, and generally putting up with me. I’m particularly thankful to Tamlyn Somerford and Will Rayment for the fantastic times in the field over multiple seasons. Also huge thanks to: Tom Brough, Sara
Niksic, Becci Jewell, Rob Lewis, Madda Fumagalli (best quake-adventure-buddy ever!), Liz
Slooten, Lindsay Wickman, Charlotte Taplin & Elisa Chillingworth (the Jeffries!), Erica Page, Jesu v
Valdés, Megan Shapiro, Toby Dickson and Rebecca Bakker. Big thanks also to Tom Brough and
Matt Desmond for help with collecting seaweed samples.
For assistance with logistics back at Otago University, big thanks to Chris Fitzpatrick for solving any question to do with finance and admin (and general running of the universe), Kim Hageman for welcoming me into her lab to use the ASE machine, and Linn Hoffmann for providing lab space, equipment and advice with examining SPOM. Huge thanks to Rob van Hale and Dianne
Clarke at Iso-Trace. Also Bob Dagg for help with setting up and dealing with the RBR, Daryl
Coup for help with palmtop computers and IT (Internet Things), and Helen Dunn and JA Parsons for general assistance. Thanks also to Reuben Pooley and Linda Groenewegen for seamlessly providing the necessary equipment when I didn’t even know what a GF/F filter looked like.
Many people offered help throughout my thesis. Big thanks to Brian Miller for advice on acoustics and for providing the Pamguard plug-in used in Chapter 2. My attempt at 3D localisation of whales did not make it through, but it certainly wasn’t from lack of your sharing of extensive knowledge. Rob Smith, your advice with oceanography and satellite data has been incredibly helpful, thank you truly. Big thanks to Steve Wing for helpful advice on food webs and isotopes. Thanks also to Iliana Ruiz-Cooley and Rebecca Guest for advice on isotope analysis,
Chris Lalas for help with squid-beak identification, Paulo Lagos for krill identification, Christina
Riesselman for advice on SPOM sampling, and Tim Jowett for stats advice. And to many other people who, through a conversation or e-mail, shared their tips and knowledge with me.
To my sperm whale sister, Tamlyn Somerford: thanks so much for your endless support and for sharing a passion for sperm whales, and also for sharing the data on seasonal whale abundance, used in Chapter 5. I would also like to thank everyone who worked on the long-term study at
Kaikōura: especially Steve and Liz, for starting it and running it, Will, for looking after it in the last few years, and Natalie Jaquet, Brian Miller, Simon Childerhouse, Miranda van der Linde,
Leslie Douglas and Abe Growcott for their research contributions.
Many people helped to make my time at Kaikōura quite special. Huge thanks to the staff and crew of Whale Watch Kaikoura, for their friendliness out on the water, their interest in the research, and for being true kaitiaki of the Kaikōura sperm whales. Very special thanks to Roger
Williams for his great support and for the ‘whaley updates’ between field seasons. I am super grateful to Ben Cooper for letting me join him and his crew out fishing – thanks for letting a
‘scientist’ in after I had almost given up hope of getting my mitts on some prey samples. Ben and
Tracey, thanks so much for your support and for some lovely evenings of great company. Thanks vi
to Murray and Nick at Kaikōura Marine for keeping our boat ship-shape and our truck truck- shape. Thanks also to Ted & Ailsa, Jody, Ngaire, Bernd & Mel. And to the excellent folks at New
Zealand Maritime Radio, for keeping a watchful ear on Grampus.
I would not have come out in one piece from this PhD if it wasn’t for the wonderful support I’ve had in Dunedin. Biggest thanks to my friends who are my family and an amazing bunch of humans – Olga, Em, Sze-En, Trudi, Manon, Jim, Charles, Will, Madda, Jun, Fatima, Rob, Sorrel,
Santi, Anna K. And to Emma & Clare, Scott, Rachel, Jen, Diego, Aimee, Rocío, Naomi and Manna, who were there at the beginning – Dunedin misses you. I am very lucky to have been surrounded by a bunch of fun, super helpful, slightly nuts lab mates. Thanks to Tom, Madda, Eva, Tamlyn,
Rob, Steph, David, Lindsay, Jesu, Toby, Will C, Rosa, Mick, Chuck and Tony. You rock. Special thanks to Eva, for your crucial help with Matlab and superhuman kindness. Thanks also to the peeps at Marine Science for the fun morning teas and cheerful atmosphere.
To my family back in Spain: Mamá, Papá, Chavi y Alicia. I can’t thank you enough for your tremendous encouragement, love, and enthusiasm. Your support knows no bounds and gives me so much strength. Os quiero muchísimo. También a Toñín y Jose. Y a mis cotufas preferidas:
Adriana y Alba. Gracias también a mis queridos Bobos y Guerras, os echo de menos cada día. Y a Marina, Rodri, Andrea, Sarini y Ana (mi perlika), por estar tan presentes, os quiero.
A massive thank you goes to the Brough/Cordery/Denham family, for loving me as one of your own. Super special mega thanks go to Isla, for your kindness and beautiful nature, and for cheerfully going along with our slightly chaotic schedules. Thank you for sharing your time with us and bringing us so much fun and joy. I love you. Thanks Ali & James for the great company, delicious dinners and always an open door. And Ian, Jenny & Bob, for your beautiful support and good times.
Tom, corazón, I don’t think the dictionary has a big enough word for the thank-you I need to express! Thank you so f. much for your inexhaustible support, love and help throughout these years. Sharing the PhD journey with you has been so fun, and I wouldn’t have made it without your encouragement throughout my ups and downs and waves of panic. Your unfaltering belief that I was totally going to make it was confusing but always a source of strength. Now on to our next adventure!
And to Kaikōura, its mountains and its sperm whales, for the magic.
vii
Table of contents
Abstract ...... i
Acknowledgements ...... iv
List of figures, tables and abbreviations ...... ix
Chapter 1. Introduction
1.1. Ecological significance of top predators in marine systems ...... 1
1.2. Studying foraging ecology of top predators ...... 2
1.3. Submarine canyons: hotspots for top predators ...... 7
1.4. Sperm whales ...... 8
1.5. Sperm whales at Kaikōura ...... 10
1.6. Thesis goals ...... 14
1.7. Thesis structure ...... 15
Statement about the data used in this thesis ...... 17
Chapter 2. Individual and seasonal variation in the foraging ecology of sperm whales at
Kaikōura, based on stable isotope analysis
2.1. Introduction ...... 19
2.2. Methods ...... 22
2.3. Results ...... 31
2.4. Discussion ...... 40
Chapter 3. Trophic interactions in the Kaikōura Canyon: using stable isotope ratios to identify the food resources of sperm whales
3.1. Introduction ...... 49
3.2. Methods ...... 52
3.3. Results ...... 61
3.4. Discussion ...... 74
viii
Chapter 4. Habitat preferences of sperm whales in the Kaikōura Canyon: environmental drivers of foraging distribution
4.1. Introduction ...... 83
4.2. Methods ...... 86
4.3. Results ...... 99
4.4. Discussion ...... 110
Chapter 5. Long-term variability in abundance of sperm whales at Kaikōura in relation to ocean conditions
5.1. Introduction ...... 119
5.2. Methods ...... 123
5.3. Results ...... 133
5.4. Discussion ...... 143
Chapter 6. Synthesis and conclusions
6.1. Summary of findings ...... 151
6.2. Ecological implications ...... 152
6.3. Implications for management ...... 156
6.4. Future research ...... 159
6.5. Concluding remarks ...... 162
Appendices
A1. Appendix 1. Supporting figures and tables for chapters 2 and 3 ...... 163
A2. Appendix 2. Supporting figures and tables for chapter 4 ...... 171
A3. Appendix 3. Supporting figures and tables for chapter 5 ...... 187
A4. Appendix 4. Publication associated with this thesis ...... 190
References ...... 203
ix
List of figures
Figure 1.1. Map of the Kaikōura Canyon, New Zealand ...... 11
Figure 1.2. Thesis road map ...... 16
Figure 2.1. Locations of sperm whale skin collection at the Kaikōura Canyon ...... 23
Figure 2.2. Sperm whale skin ...... 24
Figure 2.3. Frequency distributions of the number of seasons (summers and winters) that the
37 sperm whales included in this analysis were (a) encountered and (b) sampled
off Kaikōura between Jan-2014 and Jan-2017 ...... 32
Figure 2.4. Variation in sperm whale stable isotope ratios with (a) season, (b) sighting
frequency and (c) body length ...... 34
Figure 2.5. GAMM smoother functions for the variability in δ¹³C of sperm whales ...... 37
Figure 2.6. GAMM smoother functions for the variability in δ¹⁵N of sperm whales ...... 39
Figure 3.1. Locations of samples collected for stable isotope analysis ...... 53
Figure 3.2. Isotopic composition of sperm whales, potential prey, krill, and organic matter
source pools of Kaikōura ...... 62
Figure 3.3. Isotope ratios of candidate primary carbon sources and sperm whales ...... 66
Figure 3.4. δ13C:δ15N correlations of potential food web scenarios ...... 70
Figure 3.5. Estimated isotopic signature of the diet mixture of sperm whales in relation to
their prey ...... 72
Figure 3.6. Warty squid (Onykia ingens) from Kaikōura ...... 73
Figure 4.1. Study site around the Kaikōura Canyon, New Zealand ...... 86
Figure 4.2. Extraction of absence data for SDMs of sperm whales foraging off Kaikōura . . 91
Figure 4.3. Relative importance indices of habitat variables included in the models to explain
presence of sperm whales foraging off Kaikōura ...... 101
Figure 4.4. Effect of habitat variables on sperm whale distribution during summer, estimated
with the highest-ranking generalised additive model ...... 104
Figure 4.5. Effect of habitat variables on sperm whale distribution during winter, estimated
with the highest-ranking generalised additive mixed model ...... 105
Figure 4.6. Habitat preferences of sperm whales foraging off Kaikōura during summer . . 107 x
Figure 4.7. Habitat preferences of sperm whales foraging off Kaikōura during winter . . . 108
Figure 5.1. Oceanographic setting of the Kaikōura Canyon ...... 121
Figure 5.2. Locations of data collection for the analysis of correlations between abundance of
sperm whales off Kaikōura and satellite-derived oceanographic variables . . . 124
Figure 5.3. Temporal variability in abundance of sperm whales off Kaikōura, 1990-2017. . .134
Figure 5.4. Relative importance indices of environmental variables at different time lags,
derived from models predicting the monthly abundance index of sperm whales
off Kaikōura ...... 135
Figure 5.5. Correlation coefficients of environmental variables used as predictors of monthly
relative abundance of sperm whales ...... 137
Figure 5.6. Relative importance indices of environmental variables at different time lags,
derived from models predicting the summer abundance of sperm whales . . . 138
Figure 5.7. Correlation coefficients of environmental variables used as predictors of summer
abundance of sperm whales at Kaikōura ...... 140
Figure 5.8. Summertime abundance of sperm whales at Kaikōura 1990-2017 ...... 141
Figure 5.9. Time-series data for the duration of the study (1990-2017) for: (a) SSTA off
Kaikōura in May-July, and (b) the Mixing Ratio at the Mernoo Saddle (MRMS)
during Nov-Jan ...... 142
Figure 6.1. Schematic representation of habitat use by sperm whales foraging at Kaikōura, as
suggested by the results from this thesis ...... 153
xi
List of tables
Table 2.1. Summary of explanatory variables used to describe individual traits of sperm
whales foraging at Kaikōura ...... 28
Table 2.2. Summary of samples of sloughed skin from sperm whales off Kaikōura analysed
for δ¹³C and δ¹⁵N ...... 31
Table 2.3. Within-individual differences in stable isotope ratios of sperm whales ...... 33
Table 2.4. Correlation between isotopic signatures of sperm whales and sighting frequencies
at different temporal scales ...... 35
Table 2.5. Correlation among explanatory variables used to model the variability in isotope
ratios of sperm whales ...... 35
Table 2.6. Model selection of GAMMs to explain intra-population variation of δ¹³C of sperm
whales ...... 37
Table 2.7. Model selection of GAMMs to explain intra-population variation in δ¹⁵N of sperm
whales ...... 39
Table 3.1. Measured values of δ15N, δ13C and C:N (mean ± 1 SD) of organisms collected at
Kaikōura ...... 63
Table 3.2. Seasonal differences in measured values of δ15N, δ13C and C:N of sperm whales
and primary source pools ...... 65
Table 3.3. Sensitivity analysis in the estimation of (a) relative contribution of SPOM and
macroalgae to the diet of sperm whales at Kaikōura, and (b) trophic level of sperm
whales ...... 67
Table 3.4. Seasonal differences in the relative contribution of SPOM vs macroalgae to the diet
of sperm whales at Kaikōura, and trophic level of sperm whales ...... 68
Table 3.5. Isotope ratios of source pools and TDF used in the mixing models to estimate the
relative contribution of SPOM and macroalgae to the food web of whales . . . . . 68
Table 4.1. Candidate explanatory variables for the SDMs used to investigate the habitat
preferences of sperm whales foraging off Kaikōura ...... 93
Table 4.2. Summary of survey effort and sperm whale presences and pseudo-absences . . 100 xii
Table 4.3. Model selection of (a) summer GAMs and (b) winter GAMMs used to explain
presence of sperm whales ...... 102
Table 4.4. Smooth terms from (a) summer GAMs and (b) winter GAMMs used to explain
presence of sperm whales ...... 103
Table 4.5. Metrics of model performance for the (a) summer GAMs and (b) winter GAMMs
used to predict habitat suitability for sperm whales ...... 109
Table 5.1. Temporal scale of response and explanatory variables considered in the time-series
analysis ...... 130
Table 5.2. Ranking of LMMs used to explain the monthly relative abundance of sperm whales
off Kaikōura from 1994 to 2017 ...... 137
Table 5.3. Parameter estimates from LMMs used to model the relationship between monthly
relative abundance of sperm whales and environmental variables ...... 138
Table 5.4. Ranking of GLMs used to explain the summer abundance of sperm whales off
Kaikōura from 1990 to 2017 ...... 139
Table 5.5. Parameter estimates from GLMs used to model the relationship between summer
abundance of sperm whales and environmental variables ...... 140
xiii
List of abbreviations
AI – Abundance Index NW – Neritic Water
AICc – Akaike’s Information Criteria NZ – New Zealand
adjusted for small sample sizes SAM – Southern Annular Mode
ASE – Accelerated Solvent Extraction SAW – Subantarctic surface water
AUC – Area Under the ROC Curve SD – Standard Deviation chl-a – Chlorophyll a SDM – Species Distribution Model
CI – Confidence Intervals SE – Standard Error
CV – Coefficient of Variation SF – Sighting Frequency
DOC – Department of Conservation (NZ) SIA – Stable Isotope Analysis
ENSO – El Niño Southern Oscillation SOI – Southern Oscillation Index
GAM – Generalised Additive Model SPOM – Suspended Particulate Organic
GAMM – Generalised Additive Mixed Matter
Model SSC – Sea Surface Chlorophyll
GLM – Generalized Linear Model SSCA – Sea Surface Chlorophyll
GPS – Geographic Positioning System Anomaly
IPI – Inter-Pulse Interval SST – Sea Surface Temperature
IRMS – Isotope Ratio Mass Spectrometry SSTA – Sea Surface Temperature
LMM – Linear Mixed Model Anomaly
MMRG – Marine Mammal Research STF – Subtropical Front
Group (University of Otago, NZ) STW – Subtropical surface water
MPA – Marine Protected Area TDF – Trophic Discrimination Factor
MPI – Ministry for Primary Industries TL – Trophic Level
(NZ) TSS – True Skill Statistic
MR.MS – Mixing Ratio at the Mernoo VIF – Variance Inflation Factor
Saddle
xiv
Chapter 1.
Introduction.
1.1. Ecological significance of top predators in marine systems
For thousands of years, humans have used the ocean for its resources. Over the last century, the rate of population growth and technological advancement has increased exponentially, and our pressure on the marine environment is unprecedented (Vitousek et al. 1997, Halpern et al. 2008, Butchart et al. 2010). The impacts on marine top predators have been particularly noticeable, with widespread declines in the abundance of species and ecosystem biodiversity
(Baum et al. 2003, Myers & Worm 2003, Estes et al. 2006, Estes et al. 2011). Top predators
(typically marine mammals, seabirds, sharks and large teleost fish) occupy the highest trophic level within food webs, playing a key role in the function and structure of marine ecosystems (Heithaus et al. 2008, Rosenblatt et al. 2013, Markel & Shurin 2015). They can regulate community structure through top-down control and behaviour-mediated interactions, as well as increase food web stability (Estes & Duggins 1995, Estes et al. 2001,
Burkholder et al. 2013). The loss of top predators from marine ecosystems has had far- reaching effects. In a classic example, the intensive hunting of sea otters in the North Pacific rim suppressed predation on sea urchins, with a significant increase in herbivory and the consequent loss of kelp habitat (Estes & Palmisano 1974, Simenstad et al. 1978, Estes &
Duggins 1995). Although the consequences of predator declines on marine food webs are not fully understood (Grubbs et al. 2016), the protection of apex predators should be part of the solution to safeguard the stability of marine ecosystems.
Understanding the impact of anthropogenic activities on top predators is valuable, as it allows the development of management strategies for the protection of their populations.
Marine top predators can be directly affected by human-induced mortality through fishing, whaling and bycatch (Whitehead et al. 1997, Pauly et al. 1998a, D’Agrosa et al. 2000, Baum et al. 2003, Reeves et al. 2013), but also indirectly through alterations to their environment.
These include changes in prey abundance due to fisheries (Hammond et al. 2013), habitat Chapter 1 - Introduction 2 loss (Würsig & Gailey 2002, Jefferson et al. 2009), ecological impacts of climate change
(MacLeod 2009, Hazen et al. 2013a), and disturbance from noise pollution and boat traffic
(Mate et al. 1994, Bejder et al. 2006, Forney et al. 2017). Investigating environmental variability within a species’ habitat can help explain trends in population dynamics. For example, temperature anomalies can lead to decreased pup production in pinnipeds
(Forcada et al. 2005, Baker et al. 2007), and fluctuations in oceanographic conditions can change the large-scale distribution of cetaceans, introducing variability in the abundance of a species within a fixed surveyed region (Forney 2000, Sprogis et al. 2017).
1.2. Studying foraging ecology of top predators
For all predators, but especially those with high energy requirements, access to good foraging habitat is crucial. This determines body condition, which in marine mammals is often linked to migratory capacity, health and reproductive success (Guinet et al. 1998, Brock et al. 2013, Pirotta et al. 2018). Knowledge on the food and habitat requirements of top predators is therefore essential to understanding the status of their populations. This, in turn, is critical for deciding what management strategies are necessary to protect a population and ensure its long-term survival. There are two main components in the study of the foraging ecology of marine predators: their food web, and the habitat supporting it.
1.2.1. Marine food webs
In a simplified depiction of a food web, energy is transferred from primary producers (or autotrophs, which use inorganic compounds to produce organic matter) to consumers at higher trophic levels (or heterotrophs, including grazers, carnivores and heterotrophic microorganisms; Kaiser et al. 2005). Decomposers (including microbial organisms and scavengers) contribute to the recycling of organic matter, making inorganic material available to primary producers (Azam et al. 1993, Kaiser et al. 2005). Due to the energy lost with increasing trophic level (through movement, waste and heat), the total energy required to support an organism increases with its trophic level (Lindeman 1942, Hussey et al. 2014).
The sources of organic carbon for marine food webs are provided by primary producers.
Pelagic phytoplankton (including microalgae and bacteria) and coastal macroalgae are the most common sources in temperate ecosystems, with other sources including seagrass, mangroves, chemoautotrophic organisms and terrestrial plant input (Kaiser et al. 2005). Chapter 1 - Introduction 3
The high diversity of organisms and habitats in the marine environment, make marine food webs different to those in freshwater and terrestrial systems. Marine food webs typically have higher connectivity (i.e. high number of inter-species links) and longer chain lengths
(i.e. number of links between primary producers and top predators; Cohen 1994, Link 2002,
Dunne et al. 2004). While high complexity promotes increased stability in food webs, effects of perturbations can be transmitted rapidly and widely through marine ecosystems
(Williams et al. 2002, Dunne et al. 2004). Marine taxa at high trophic levels tend to have particularly high connectivity (Rooney et al. 2006), with intensive exploitation of top predators impacting the structure and robustness of marine food webs (Pauly et al. 1998a,
Jackson et al. 2001).
1.2.2. Studying food webs
To understand what energy resources are important for a predator it is necessary to identify their prey as well as the primary producers that sustain their food web. Traditional dietary studies of marine mammals have used direct field observations (e.g. Bräger 1998), scat analysis (e.g. Treacy & Crawford 1981) and the examination of stomach contents (e.g., Santos et al. 1999, Evans & Hindell 2004). These methods can give valuable information on diet composition and feeding habits (Pierce & Boyle 1991, Smith & Whitehead 2000), but provide little insight into the base of the food web and are not always suitable. In the case of sperm whales (Physeter macrocephalus), in particular, traditional methods offer limited knowledge.
Faecal plumes do not give a representative sample of prey items (Smith & Whitehead 2000), field observations are impractical because sperm whales feed at depth, and opportunities for examination of stomach contents are uncommon because sperm whales rarely strand. More recently, the use of biochemical tracers that are transferred from prey to a predator’s tissues during metabolism has greatly advanced our ability to study foraging ecology (Hobson et al. 1994, Budge et al. 2006, Fry 2006, Newsome et al. 2010a). Predators integrate biochemical tracers (such as stable isotopes, fatty acids and organic pollutants) acquired from their diet, which reflect both the prey consumed and the regions from which they were taken. These natural biomarkers can be used to trace energy from food sources to consumers (Iverson et al. 2004, Budge et al. 2006, Krahn et al. 2007, Newsome et al. 2010a). Chapter 1 - Introduction 4
Stable isotopes have been used to address a wide range of ecological questions, including predator-prey relationships, pathways of organic matter through food webs, and movement patterns (DeNiro & Epstein 1978, 1981, Peterson & Fry 1987, Fry 2006, Newsome et al. 2010a).
The most commonly used isotope ratios are those of carbon (¹³C:¹²C, expressed as δ¹³C) and nitrogen (¹⁵N:¹⁴N, or δ¹⁵N). The natural abundances of heavy to light isotopes differ among primary producers, due to the differences in isotope ratios of their external environment, physiology and metabolism (Rau et al. 1982, Peterson & Fry 1987). This variability exists between types of primary producers (e.g., macroalgae vs phytoplankton) and spatially (e.g., with latitude or distance to land). For example, phytoplankton with high growth rates from productive nearshore regions are associated with higher δ¹³C values, compared to phytoplankton with slower growth rates further offshore (Goericke & Fry 1994, Graham et al. 2010). The difference in isotope ratios among primary producers allows the tracking of these sources into higher trophic levels.
When a consumer ingests food, the heavier isotopes are assimilated into tissue at faster rates, leading to increases in the stable isotope ratios of their tissues (Fry 2006). This process is termed ‘fractionation’, and the change in isotopic values from diet to consumer is referred to as ‘trophic discrimination’ or ‘isotopic shift’ (Peterson & Fry 1987, Fry 2006). Values of δ¹³C change little (0.5 – 1‰) as carbon is passed up the food chain, providing a way to identify which primary producers support a consumer – either directly or indirectly (De Niro &
Epstein 1978, Peterson & Howarth 1987). Values of δ¹⁵N change more substantially with each trophic step, increasing by 2 – 4‰ with every step in the food web (DeNiro & Epstein 1981,
Caut et al. 2009). Thus, δ¹⁵N is useful for estimating the trophic position of a consumer (Post
2002). Together, analyses of δ¹³C and δ¹⁵N are a powerful tool to determine predator-prey relationships and trace energy transfer through food webs. For example, the comparison of stable isotopes in the skin of dolphins to those in potential prey and primary producers in a fiord, determined the importance of local rocky reefs vs external pelagic subsidies to the dolphins (Lusseau & Wing 2006). Stable isotope analyses have also been used to identify the indirect but substantial contribution of kelp-derived carbon for pelagic predatory fish in nearshore ecosystems (Koenings et al. 2015, Markel & Shurin 2015, von Biela et al. 2016). In addition, because δ¹³C and δ¹⁵N of primary producers vary spatially, isotope ratios can provide insights into the regions where top predators forage (Hobson et al. 1997, Graham et Chapter 1 - Introduction 5 al. 2010, Lorrain et al. 2015). For example, isotopic values in penguins indicated that two sympatric sub-populations used different foraging habitats across an inshore-offshore gradient (Cherel & Hobson 2007).
Stable isotope analyses of carbon and nitrogen in bulk tissue do not always yield clear outcomes; complementary methods are increasingly used to improve resolution. Isotope ratios of sulphur (³⁴S:³²S) can help resolve the contribution of organic sources to consumers, such as distinguishing between benthic and pelagic sources or the contribution of chemosynthetic biomass (MacAvoy et al. 2002, Connolly et al. 2004). Fatty acid analyses can also be used to investigate diet and trace the sources of primary production in a food web
(e.g., Hooker et al. 2001, MacLeod & Wing 2007). This method is based on the natural variability in fatty acid composition among primary consumers and the lack of modification in fatty acids as these are incorporated into a consumer’s diet (Budge et al. 2006). Fatty acid analyses require samples with high lipid content, such as blubber, which cannot easily be obtained from free-ranging predators. Another approach is the stable isotope analysis of specific compounds, such as amino acids (McClelland & Montoya 2002, Ohkouchi et al.
2017). In contrast to bulk tissue, which contains molecules with different origins and metabolisms, analysis of individual amino acids can increase specificity (Popp et al. 2007,
Zupcic-Moore et al. 2017). This technique can address some of the limitations of bulk tissue isotope analysis, but requires larger samples, is substantially more expensive and analytically complex.
1.2.3. Stable isotope analysis: assumptions, strengths and limitations
Ecological interpretations of stable isotope data require information of two main sources of isotopic variation: (1) the transformation of isotope ratios as a food source is assimilated into the tissues of a consumer (‘trophic discrimination’), and (2) the time taken for the isotopic value of diet to be integrated into the tissues of a consumer (‘turnover time’; Fry 2006). These values can vary widely among species, individuals, tissues and diets (Vanderklift & Ponsard
2003, Newsome et al. 2010a), and are difficult to measure in the wild, especially in the case of large and mobile marine predators. When the values of trophic discrimination and turnover times are unknown, they are typically assumed based on values in similar taxa. These assumptions are a source of uncertainty in stable isotope studies. Chapter 1 - Introduction 6
When the isotopic signatures of potential prey or primary organic sources are known,
‘mixing models’ (Phillips & Gregg 2001, Phillips 2012), can estimate their relative contribution (or mix) to a predator’s diet. However, an important limitation to this approach is that it cannot identify the contributions of different prey, unless the suite of potential prey is known (Newsome et al. 2010a). Furthermore, the isotope ratios of food sources need to be isotopically distinct, a requirement which is not always met. In spite of these caveats, stable isotope analysis is a powerful tool, and has some key advantages over other methods used to study trophic ecology. Compared to stomach contents analysis, it allows for continued study of live animals, and provides measurements of food assimilated over time, not just recent food uptake (Pierce & Boyle 1991). In the case of cetaceans, the use of sloughed skin
(e.g., Marcoux et al. 2007) provides a non-invasive source of samples for analysis, averting the use of techniques such as biopsy sampling.
1.2.3. Habitat selection by marine top predators
The preference for particular habitats leads animals to have uneven distributions throughout their range (Hall et al. 1997). Information on where a species is found and what are its habitat requirements is central to understanding its ecology, but is also useful to assess the overlap with anthropogenic impacts, prioritise areas for protection, and guide conservation strategies (Bailey & Thompson 2009, Embling et al. 2010, Becker et al. 2012).
Prey availability is one of the main factors determining the distribution of marine mammal top predators (Benoit-Bird & Au 2003, Hastie et al. 2004, Friedlaender et al. 2006, Hazen et al. 2011). In turn, environmental characteristics, such as primary productivity, topography, oceanic fronts, and seawater properties, influence prey availability and can thus correlate with habitat selection (Yen et al. 2004, Croll et al. 2005, Torres et al. 2008, Forney et al. 2015).
Intrinsic biological factors, such as physiology, sex, or breeding status, can also influence habitat choice, and hence the distribution of a population (Leung et al. 2012, Rayment et al.
2014).
Species-distribution models are a useful way to quantify the habitat preferences of a species
(Guisan & Zimmermann 2006) and are widely used in ecological studies of cetaceans
(Redfern et al. 2006, Pirotta et al. 2011, Forney et al. 2012). By relating the locations of individuals to the characteristics of their habitat, the models quantify the relative importance Chapter 1 - Introduction 7 of different environmental variables in determining the spatial distribution of a species
(Guisan & Zimmermann 2000, Redfern et al. 2006).
The diet of marine mammals can vary seasonally according to the availability and distribution of their prey (e.g., Similä et al. 1996, Hall-Aspland et al. 2005). This, in turn, determines their seasonal movements and foraging distribution (Jaquet et al. 2000, Andrews
& Harvey 2013). Thus, studies of habitat use greatly benefit from examining foraging patterns at a seasonal scale (Becker et al. 2014). At a longer time-scale, fluctuations in ocean conditions can change food availability, which can impact foraging patterns, breeding success and population dynamics of top predators (Smith & Whitehead 1993, Xavier et al.
2013, Bost et al. 2014, Sprogis et al. 2017). For example, high sea-surface-temperature anomalies due to El Niño Southern Oscillation around South Georgia correlated with extreme lows in pup production of Antarctic fur seals, thought to be mediated by a decline in krill abundance (Forcada et al. 2005). Identifying the inter-annual variation in habitat use helps us understand how populations respond to environmental variability and anthropogenic activities (Forney 2000, Rolland et al. 2009).
1.3. Submarine canyons: hotspots for top predators
Although many top predators are highly mobile species with wide ranges, their distribution in the world’s oceans is not homogenous: there are ‘hotspot’ habitats where diversity and abundance are particularly high (Worm et al. 2005, Hazen et al. 2013b). Submarine canyons are one such place, providing key habitat for many cetacean species (Yen et al. 2004, Vetter et al. 2010, Moors-Murphy 2014). This association seems particularly strong for sperm whales and beaked whales (family Ziphiidae) (Whitehead et al. 1992, Jaquet et al. 2000,
MacLeod & Zurr 2005, Johnson et al. 2016), deep-diving predators that specialise in hunting deep-water squid and fish. Although the association is clearly driven by high food availability (Moors-Murphy 2014), the mechanisms behind this relationship are poorly understood.
Submarine canyons are steep-walled valleys that connect shallow continental shelves to deep ocean basins (Shepard and Dill 1966), and are often characterised by high productivity and habitat heterogeneity (Vetter 1994, McClain & Barry 2010). Research in the last two decades has shown that the function and biodiversity of submarine canyon ecosystems can Chapter 1 - Introduction 8 be seriously impacted by anthropogenic activities, such as fishing, pollution and extraction of oil and gas (Danovaro et al. 2008, Ramirez-Llodra et al. 2010, Fernandez-Arcaya et al.
2017). In particular, there is growing concern that the effects of climate change will modify the hydrography of canyons, with consequent impacts on their biological communities
(Levin & Le Bris 2015). To ensure the long-term health of canyon ecosystems and the top predators they support, it is important to maximise their protection from anthropogenic disturbance and manage their resources in a sustainable way. This requires a robust understanding of the ecological communities within them, and knowledge on how they are affected by changes in their environment (Worm et al. 2006, Fernandez-Arcaya et al. 2017).
1.4. Sperm whales
The sperm whale (Physeter macrocephalus, L. 1758) belongs to the order of cetaceans, suborder odontocetes – the toothed whales, dolphins and porpoises, all of which use echolocation for foraging and navigating their surroundings (Norris & Harvey 1972, Au 1993). Sperm whales are animals of extremes. They are the largest odontocete, own the world’s largest nose, produce the loudest biological sounds, and perform some of the deepest and longest dives
(Cranford 1999, Møhl et al. 2003, Whitehead 2017). These traits reflect highly specialised anatomical and physiological adaptations, which have allowed sperm whales to become very successful deep-diving predators. Using a specialised nasal complex, they produce loud, broadband ‘click’ sounds (Goold & Jones 1995) used for short and long-range echolocation (Møhl et al. 2000, Fais et al. 2015), as well as communication (Weilgart &
Whitehead 1993). Foraging dives typically last 40-60 minutes and reach depths of 400-1200 m, although sperm whales occasionally hunt at shallow depths and can dive deeper than
2000 m (Watkins et al. 1993, Watwood et al. 2006, Teloni et al. 2008, Miller et al. 2013a).
Sperm whales are widely distributed throughout the deep waters of both hemispheres (Rice
1989). They are concentrated in ‘grounds’ that are typically areas of high productivity (Jaquet
& Whitehead 1996, Jaquet et al. 1996). At a broad scale, sperm whale distribution is often associated with steep topography, such as the edges of continental shelves and submarine canyons (Whitehead et al. 1992, Jaquet et al. 2000, Pirotta et al. 2011), or with dynamic oceanographic features, such as frontal zones and eddies (Griffin 1999, Gannier & Praca 2007,
Wong & Whitehead 2014). The spatial distribution of males and females is starkly different. Chapter 1 - Introduction 9
Females and calves live in nursery groups closer to the tropics, between latitudes of 50°N and 40°S (Rice 1989, Whitehead 2003). Males disperse from their natal groups at the age of
6-10 years, and migrate to higher latitudes, apparently seeking more productive feeding grounds (Best 1979, Rice 1989, Mendes et al. 2007a, b). Adult males return periodically to warmer latitudes to breed from around 25 years of age (Lockyer 1989, Whitehead 2017), but these visits are short compared to the time spent in feeding grounds (Mendes et al. 2007a).
Sperm whales have strong sexual dimorphism in size and sociality, in addition to distribution. Adult females typically weigh 15 tons and reach 11 m in length, while mature males weigh about 45 tons and measure 16 m long, sometimes reaching up to 18 m (Rice
1989). In addition, while females and their calves live in tight social groups (Whitehead 2003,
Coakes & Whitehead 2004), males tend to be solitary. When they leave their nursery group they form loose ‘bachelor schools’, and as they age and range to colder latitudes their aggregations become smaller, with the largest males usually found alone (Christal &
Whitehead 1997, Whitehead 2017).
Sperm whales are top predators of mesopelagic1 ecosystems in the deep ocean. Information on the diet of sperm whales is limited, with most knowledge coming from the examination of hunted and stranded animals. Stomach content analyses indicate that their primary prey are mesopelagic and bathypelagic squid (Okutani & Nemoto 1964, Rice 1989, Santos et al.
1999). In some high-latitude regions, however, demersal fish contribute significantly to their diet (Kawakami 1980, Martin & Clarke 1986, Rice 1989). Diet composition appears to vary according to sex and age, as well as among regions, seasons and years (Clarke et al. 1993,
Best 1999, Ruiz-Cooley et al. 2012). Adult males have very high energy requirements, and are thought to eat 1-2 tonnes of food a day (Clarke 1977, Lockyer 1981).
Multiple studies have investigated the drivers of sperm whale distribution around the world
(e.g., Whitehead et al. 1992, Jaquet & Gendron 1992, Praca et al. 2009, Pirotta et al. 2011).
Although the species is generally associated with deep water and steep topographic features, the characteristics of their preferred foraging habitat are highly variable. A lot of this
1 The following terms are used throughout this thesis to refer to vertical strata in the ocean and organisms found therein: ‘pelagic’ refers to the water column in the open ocean; ‘demersal’ relates to the layer in proximity to the seafloor; ‘benthic’ relates to organisms living at or in the seafloor; ‘benthopelagic’ refers to organisms that move between benthic and pelagic strata; ‘mesopelagic’ refers to intermediate depths in the water column (c. 200 – 1000 m); ‘bathypelagic’ refers to depths in the water column of c. 1000 – 3000 m. Chapter 1 - Introduction 10 apparent variability stems from the different spatial and temporal scales at which the relationships are examined (Jaquet 1996). Understanding the key influences on sperm whale distribution is challenging due to their large home-ranges, deep-water feeding habits, dynamic environment, and limited knowledge concerning their prey (Jaquet 1996). In addition, their high trophic level means that temporal and spatial lags between primary productivity and sperm whale occurrence can be large (Jaquet 1996).
Sperm whales are currently listed as ‘vulnerable’ by the International Union for the
Conservation of Nature (Taylor et al. 2008, IUCN 2017). This conservation status is mainly due to the global decline caused by whaling during the 19th and 20th centuries (Whitehead
2002). Sperm whales have extremely low rates of population growth (at a maximum of 1% per year, Whitehead 2002); although commercial whaling has stopped, its flow-on effects still linger. Despite decades of protection, most heavily exploited sperm whale populations are not recovering (Whitehead et al. 1997, Whitehead 2003, Reeves & Notarbartolo Di Sciara
2006, Carroll et al. 2014) and some are in decline (Whitehead et al. 1997, Notarbartolo Di
Sciara 2014, Gero & Whitehead 2016). Anthropogenic stressors that may hinder recovery include entanglement in fishing gear (Barlow & Cameron 2003, Hucke-Gaete et al. 2004,
Reeves et al. 2013, Rendell & Franzis 2016), ecological impacts from fisheries (Trites et al.
1997), climate change (Cantor et al. 2017), seismic exploration (Bowles et al. 1994, Farmer et al. 2018), tourism disturbance (Richter et al. 2006), collisions with ships (Carrillo & Ritter
2010, Fais et al. 2016), toxic contaminants (Nielsen et al. 2000, Pinzone et al. 2015) and ingestion of debris and plastic (Jacobsen et al. 2010, Unger et al. 2016). The effects of these anthropogenic impacts are poorly understood at a population level.
1.5. Sperm whales at Kaikōura
The Kaikōura Canyon (-42.5 S, 173.7 E) is located off the east coast of New Zealand’s South
Island (Figure 1.1). It has been described as the most productive non-chemosynthetic habitat recorded to date in the deep ocean (De Leo et al. 2010). The canyon harbours an exceptional biomass of demersal fish and benthic invertebrates (De Leo et al. 2010, Leduc et al. 2012,
2014). This richness is probably derived from the high primary productivity in waters above the canyon, and potentially increased by macrophyte detritus exported from the coast (De
Leo et al. 2010, Leduc et al. 2012, 2014), although these mechanisms remain untested. Chapter 1 - Introduction 11
Furthermore, the presence of many top-predators that target mesopelagic prey (Benoit-Bird et al. 2004, Boren et al. 2006, Miller et al. 2013a) suggests that the canyon hosts a highly productive pelagic system. Surprisingly for such an accessible system, the drivers of the high productivity in the Kaikōura Canyon are unknown, and the reasons behind the habitat’s importance to marine megafauna remain to be studied.
Figure 1.1. The Kaikōura Canyon. The research area for sperm whale surveys is indicated by the broken red line. The Kaikōura Marine Management Area includes a no-take marine reserve (Hikurangi Marine Reserve, in green), and a whale sanctuary (Te Rohe o Te Whānau Puha, in yellow) where seismic surveying is restricted. Depth contours show 250, 500, 1000, 1500 and 2000 m isobaths.
Kaikōura is one of a few places worldwide where sperm whales can be found near-shore.
The canyon is a foraging ground for bachelor and mature males (Childerhouse et al. 1995,
Growcott et al. 2011). The whales have an important cultural value as taonga (‘treasure’) species, and are a key natural asset for the local tourism industry, which is the main driver of the economy at Kaikōura (Curtin 2003). The population has been studied since 1990 by Chapter 1 - Introduction 12 the Marine Mammal Research Group of the University of Otago (MMRG), with studies focusing on population abundance and dynamics (Childerhouse et al. 1995, van der Linde
2009, Somerford 2018), size estimation and growth rates (Dawson et al. 1995, Rhinelander &
Dawson 2004, Growcott et al. 2011, Miller et al. 2013b), acoustic and diving behaviour (Jaquet et al. 2000, 2001, Douglas et al. 2005, Miller et al. 2013a, Guerra et al. 2017) and the effects of whale-watching tourism (Richter et al. 2006). Fieldwork has typically been carried out in summer and winter. Sperm whales can be found at Kaikōura almost year-round, except for a few weeks in spring (September/October; MMRG unpublished data, Sagnol et al. 2014a).
The whales have a wide range of residency rates (Childerhouse et al. 1995, Jaquet et al. 2000) and distributions (MMRG unpublished data), suggesting potential differences in habitat use among individuals.
The abundance of sperm whales within the study area has declined over the duration of the
28-year study, from 89 individuals (95% CI: 60 to 131) in 1991 to 40 (95% CI: 33 to 49) in 2017
(van der Linde 2009, Somerford 2018). The reasons for this change are unknown, but the trend is driven by a decrease in the number of whales visiting Kaikōura in spring and summer (Somerford 2018), and there appears to have been a decline in the daily density of whales year-round (Sagnol et al. 2014a, MMRG unpublished data). With no evidence for impacts on survival (Somerford 2018), and considering the large ranges of sperm whales
(Whitehead et al. 2008), the decline probably reflects a shift in distribution away from the study area. For example, whales may be favouring areas further offshore, or changing how often, or for how long, they visit Kaikōura. This could be caused by underlying ecological or oceanographic changes that affect the availability of their prey, or by human activities in the area (such as tourism and fishing). Although the sperm whales of Kaikōura are well studied, their foraging ecology is poorly understood. With the canyon functioning as a foraging hotspot for the population, it is important to better understand what sustains the whales’ diet and the environmental features that drive their distribution.
Most of what is known about the diet of sperm whales in New Zealand comes from one study. Gaskin & Cawthorn (1967) examined the stomach contents of sperm whales caught in the wider Cook Strait/Kaikōura area. Their analysis showed that their main prey was
Onychoteuthid squid (known as warty squid), but that demersal fish made up an unusually high proportion of their diet. Sperm whales off Kaikōura forage throughout the water Chapter 1 - Introduction 13 column, targeting pelagic and demersal habitats (Miller et al. 2013a, Guerra et al. 2017). The frequency of demersal foraging at Kaikōura is higher than that reported in any previous studies, suggesting that the extreme benthic productivity of the canyon might be reflected in their prey preferences (Guerra et al. 2017). What sperm whales are eating in the Kaikōura
Canyon, specifically, remains unknown.
The spatial distribution of sperm whales over the Kaikōura Canyon is not homogenous
(Jaquet et al. 2000). Significant differences in the spatial distribution and diving behaviour between summer and winter suggest that the whales may change their diet in response to changes in prey availability and distribution (Jaquet et al. 2000). While underwater topography plays a role in shaping the whales’ distribution (Jaquet et al. 2000, Sagnol et al.
2014b), there have been no published studies addressing the influence of oceanographic factors on the whales’ habitat preferences.
The influence of environmental fluctuations on sperm whales at Kaikōura has not been investigated. It is possible that changes in oceanographic conditions over the last few decades have affected prey availability or impacted the primary organic sources at the base of the whales’ food web. Oceanic squid, for example, are restricted to relatively narrow temperature ranges (Jackson et al. 2000), making them sensitive to climate variability. Ocean warming in southeast New Zealand over the last 50 years (Shears & Bowen 2017) may have affected the distribution of squid and influenced other ecosystem processes. Quantifying environmental change may help explain the observed decline in the number of sperm whales foraging in the area. Thanks to the long-term dataset of individual sightings recorded since
1990 it is possible to estimate population trends over decades. This allows examination of long-term correlations between oceanographic conditions and whale abundance.
The Kaikōura Marine Management Area was established in October 2014, centred over the
Kaikōura Canyon (Fig. 1.1). The management area includes a no-take marine reserve (the
‘Hikurangi Marine Reserve’) and a whale sanctuary (‘Te Rohe o Te Whānau Puha’), where seismic surveying activity is restricted (DOC 2014). Although marine reserves can provide significant conservation benefits (Baum et al. 2003, Halpern 2003), the effectiveness of the
Hikurangi Marine Reserve in protecting the canyon ecosystem is uncertain. It is extremely narrow in some parts (e.g., 1.95 km wide where it joins the coastline), and has a complex Chapter 1 - Introduction 14 shape, with boundaries adjusted to accommodate fishing interests. The value of the reserve in protecting the sperm whale population remains unknown.
1.6.Thesis goals
The primary motivation for this research was to understand the decline in the number of sperm whales foraging in and around the Kaikōura Canyon over the last three decades. The drivers of this change are unknown, preventing development of protection measures. With this thesis I aimed to identify the food resources and environmental drivers of distribution for sperm whales at Kaikōura, in order to better understand their habitat requirements.
Given that the decline is apparent in summer, I examined the whales’ foraging ecology within a seasonal context. A range of methods, including stable isotope analyses, species- habitat surveys and oceanographic sampling, were used to address the thesis objectives. My specific research goals were to:
(1) Investigate the food resources utilised by sperm whales foraging at Kaikōura:
- What are the primary carbon sources sustaining the whales’ food web?
- What can we learn about the contribution of different prey groups to the whales’ diet
using stable isotope analyses?
- How does diet vary seasonally?
- Is there individual variability in the use of food resources?
(2) Identify the whales’ habitat preferences:
- What are the environmental drivers of foraging distribution at Kaikōura?
- What are the characteristics of the whales’ preferred habitat?
- What is the spatial overlap between foraging habitat and the current MPA?
(3) Investigate the role of environmental change in the decline of whale numbers:
- Is inter-annual variability in ocean conditions correlated with whale abundance?
- Could long-term changes in the whales’ habitat be contributing to their decline?
I used this information to propose conservation measures that would contribute to the protection of sperm whales and their habitat at Kaikōura. In a wider ecological context, I aimed to improve our understanding of how submarine canyons provide important habitat for marine top predators. Chapter 1 - Introduction 15
1.7. Thesis structure
In the first two chapters, I describe the use of stable isotope analyses to investigate the food web of sperm whales at Kaikōura. In Chapter 2, I examine variability in stable isotope signatures of sperm whales by analysing seasonal patterns in diet and differences among individuals according to their residency rates and body size. In Chapter 3, I assess the whales’ stable isotope signatures in a food web context, to determine the relative contribution of pelagic primary production relative to coastal macroalgae to the food web sustaining sperm whales at Kaikōura. In this chapter, I also explore the contribution of fish and squid to the whales’ diet. In Chapter 4, I use sightings of foraging sperm whales and habitat characteristics (including topographic and in situ oceanographic data) to create species-distribution models. These models quantify the spatial distribution of sperm whales and establish the environmental drivers of habitat preferences in summer and winter. I also examine the overlap between the Hikurangi Marine Reserve and foraging habitat. Chapter
5 is an exploratory analysis of the long-term correlations between sperm whale abundance and ocean conditions around the Kaikōura Canyon over the last 28 years. In this chapter, climatic indices and remotely sensed oceanographic data are used to quantify the inter- annual variability in the whales’ environment, and identify potential environmental factors contributing to the decline in the number of whales using the area. Lastly, Chapter 6 is a general discussion in which the conclusions from the previous chapters are brought together, and I outline recommendations for the management of sperm whales at Kaikōura. This structure is represented graphically in Fig. 1.2.
Chapter 1 - Introduction 16
Figure 1.2. Thesis road map.
17
Statement about the data used in this thesis
This thesis is part of the long-term study of sperm whales at Kaikōura carried out by the
Marine Mammal Research Group (MMRG), which started in 1990. The research group is led by Steve Dawson, Liz Slooten and Will Rayment (University of Otago). My research draws from, and builds on, this study.
I personally gathered the data used in this thesis, with one exception: the long-term series of whale abundance estimates (1990-2017). In Chapter 5, I present a time-series analysis which examines the correlation between abundance of sperm whales at Kaikōura and environmental variability. I use two variables to describe whale abundance: (i) relative abundance for each month, based on photo-ID sightings, and (ii) seasonal abundance estimated for each summer (‘summertime abundance’), based on photo-ID capture- recapture analyses. The long-term database of whale sightings necessary to calculate these variables is held and owned by the MMRG. The estimates of summertime abundance were calculated by Tamlyn Somerford for her MSc thesis at the University of Otago (Somerford
2018), working under the supervision of Rayment, Dawson and Slooten, and alongside me in the field. I have used these estimates as a response variable in the analyses of Chapter 5, with the permission of Somerford and her supervisors. Both Somerford and I have contributed to the long-term dataset by collecting sighting and photo-identification data, and by matching fluke photographs to the Kaikōura photo-ID catalogue. My contribution spans from 2014 to 2018. With the exception of summertime abundance estimation, I carried out all analyses presented in this thesis (including the calculation of relative abundance of sperm whales per month). I aim to adapt Chapter 5 into a manuscript for publication in a peer- reviewed journal, in which Somerford, Rayment, Dawson and Slooten will be co-authors.
My involvement with sperm whale research at Kaikōura began in 2013, with a research cruise led my Michael Moore (Woods Hole Oceanographic Institute) in collaboration with the MMRG. One of the objectives of the research was to study the whales’ diving behaviour.
During my PhD, I used some of the data collected during the cruise to investigate the whales’ foraging behaviour. This led to the publication ‘Diverse foraging strategies by a marine top predator: sperm whales exploit pelagic and demersal habitats in the Kaikōura Canyon’ 18
(Guerra et al. 2017, included in Appendix 4). This study helped to address some of the questions posed in this thesis.
On the 14th November 2016, a 7.8 magnitude earthquake affected Kaikōura. Some of the data collected throughout this PhD, in addition to data from the long-term database held by the
MMRG, were used in a study to assess the potential impacts of the Kaikōura Earthquake on sperm whales (Guerra et al. 2018). The report compared data from before (Jan 2014 – Nov
2016) and after (Dec 2016 – Jan 2018) the earthquake. This study was presented in a report for Fisheries New Zealand (formerly known as Ministry for Primary Industries), and reviewed by the Aquatic Environment Working Group. Specifically, this report included long-term data on seasonal abundance of whales (1990-2017), as well as shorter-term data on stable isotope ratios of sperm whale skin, spatial distribution of whales, and acoustic and surface behaviour (2014-2018). The report is available on request from Fisheries New
Zealand ([email protected]) or from myself directly ([email protected]).
Chapter 2.
Individual and seasonal variation in the foraging ecology of sperm whales at Kaikōura, based on stable isotope analysis
2.1. Introduction
Despite the connectedness of marine ecosystems, food resources are heterogeneously distributed over a wide range of spatial and temporal scales (Levin & Dayton 2009). This has important consequences for how consumers use their habitat, as it is a key influence on their distribution and foraging behaviour (Ito et al. 2009, Benoit-Bird et al. 2013, Kuhn et al. 2015).
Traditionally, resource use has been investigated at the species or population level, but more recently it has been shown that variability among individuals plays an important role in shaping ecological communities (Bolnick et al. 2003, Matich et al. 2011, Kernaléguen et al.
2015, Samarra et al. 2017). In this context, individuals sharing an environment can differ in their foraging patterns by targeting different habitats or prey. Diversity in foraging strategies among conspecifics can influence population stability, social interactions, and food web structure (Bolnick et al. 2011, Schreiber et al. 2011). Investigating individual variation in foraging ecology can be helpful for understanding the population dynamics of marine top predators and their food webs (Schreiber et al. 2011).
The feeding habits of many marine top predators vary seasonally with the availability of food resources (Similä et al. 1996, Hall-Aspland et al. 2005, O’Toole et al. 2015). Knowledge of such variability is useful for understanding seasonal changes in predator distribution, their ecological role, and their overlap with anthropogenic impacts (Chilvers et al. 2003,
Matich & Heithaus 2014). Sperm whales are deep-diving predators, specialised in hunting squid and fish (Okutani & Nemoto 1964, Gaskin & Cawthorn 1967, Martin & Clarke 1986).
Very little is known about the seasonality of their diet, but analyses of stomach contents suggest that some prey are preferentially targeted at certain times of the year (Gaskin & Cawthorn 1967, Best 1999). Chapter 2 – Intra-population variability of isotope ratios 20
The Kaikōura Canyon is a foraging ground for male sperm whales (Childerhouse et al. 1995).
The decline in abundance over the last three decades is driven by a decrease in the number of whales visiting the area during spring and summer (Somerford 2018). This highlights the urgency to better understand the trophic dynamics of the population and its seasonal variability. The spatial distribution and diving behaviour of the whales differ between summer and winter (Jaquet et al. 2000). It has been proposed that this is driven by fluctuations in prey availability (Jaquet et al. 2000); however, seasonal variations in diet have never been directly addressed.
Sperm whales at Kaikōura have a wide range of residency patterns. Some individuals are seen once or twice over a season, while others spend months at a time foraging at Kaikōura
(Jaquet et al. 2000). On a longer time-scale, some whales have been seen only once over the duration of the 28-year study, while others have been re-sighted consistently for up to 25 years (Childerhouse et al. 1995, Somerford 2018). There also appear to be differences in habitat use among individuals, with some whales often found foraging inshore, while others tend to forage in deeper and further offshore waters (MMRG, unpublished data). These patterns suggest individual variability in foraging ecology, which may be reflected in the whales’ diet. There is also wide variation in body sizes among sperm whales, with total length varying by up to 30% among individuals (Growcott et al. 2011). Body size can influence competition for resources, with larger animals expected to occupy better habitats
(Smith & Parker 1976, Renison et al. 2002). The echolocation clicks produced by sperm whales have a multi-pulse structure that is caused by reflections within their head (Norris &
Harvey 1972), conveying reliable information about an individual’s size (Rhinelander &
Dawson 2004, Growcott et al. 2011). Echolocation clicks may act as acoustic signals that mediate competition among males (Weilgart & Whitehead 1988) by settling size contests without the risk of physical interactions. If body size plays a role in competition for foraging habitat, or if larger animals can target larger or deeper prey, size variation might influence differences in diet among whales (e.g., Thomson et al. 2012, Nifong et al. 2015).
Stable isotope ratios in the tissues of a predator reflect the integrated isotopic composition of its prey, providing a useful tool for investigating diet (Newsome et al. 2009a, Matich &
Heithaus 2014, Kernaléguen et al. 2015). Isotope ratios of carbon (¹³C:¹²C, expressed as δ¹³C) and nitrogen (¹⁵N:¹⁴N, or δ¹⁵N) transform in a more or less predictable way as a food source Chapter 2 – Intra-population variability of isotope ratios 21
is assimilated into the tissues of a consumer (De Niro & Epstein 1978, 1981, Fry 2006). Values of δ¹³C change little (0.5 – 1‰) with each trophic step, providing a way to identify which primary producers support a food web (Peterson & Howarth 1987). Values of δ¹⁵N change more substantially, increasing by 2 – 4‰ with every step in the food web (DeNiro & Epstein
1978, 1981, Caut et al. 2009). Thus, δ¹⁵N is useful for estimating the trophic level of a consumer (Post 2002). Importantly for marine food webs, the baseline isotope ratios in autotrophic sources (such as pelagic phytoplankton) show strong spatial variation within ocean basins (Rau et al. 1982, Somes et al. 2010, Magozzi et al. 2017). This means that the isotope ratios of a predator will not only be influenced by its trophic level and the autotrophic sources supporting its diet, but also by the region where it forages (Cherel &
Hobson 2007, Díaz-Gamboa et al. 2017). These patterns need to be considered when comparing isotope ratios among animals with large home ranges.
Sperm whales slough skin naturally and regularly, providing a reliable source of tissue that can be obtained non-invasively for use in dietary studies (Whitehead et al. 1990). In combination with photo-identification, skin samples can be assigned to specific individuals in a population. This approach allows monitoring of individual isotopic signatures over time
(e.g., Marcoux et al. 2007), as well as relating signatures to individual traits in order to explore intra-population variability in foraging (e.g., Samarra et al. 2017). Stable isotope analyses have been used to identify variation in the diet of sperm whales among populations and geographical areas (Mendes et al. 2007b, Ruiz-Cooley et al. 2012, Zupcic-Moore et al.
2017), as well as between sexes (Ruiz-Cooley et al. 2004), among social groups (Marcoux et al. 2007) and in relation to ontogenetic movements (Mendes et al. 2007a, Borrell et al. 2013).
In this study, I investigate the variability in the stable isotope ratios of sperm whales found in and around the Kaikōura submarine canyon, to identify seasonal patterns and differences in foraging among individuals. Specific questions included:
1) Do foraging patterns change between summer and winter?
2) Are there differences in isotopic signatures among individual whales?
3) Do whales with different residency rates at Kaikōura differ in their foraging patterns?
4) Do larger whales have different foraging patterns than smaller whales? Chapter 2 – Intra-population variability of isotope ratios 22
To address these questions, values of δ¹³C and δ¹⁵N in sloughed skin were modelled in relation to seasonal and individual traits, including sighting frequency, long-term association with the canyon, and body size. By examining the fine-scale variation in isotopic signatures of sperm whales I also attempted to establish a framework for studying the whales’ foraging patterns in relation to the food web of the submarine canyon (addressed in
Chapter 3).
2.2. Methods
2.2.1. Sample and data collection
Samples of sperm whale skin were collected from the Kaikōura Canyon and surrounding areas (Fig. 2.1) between January 2014 and January 2017. This period included sampling over four spring/summers (November to February; hereafter summer) and three autumn/winters
(May to July; hereafter winter). Data collection was conducted aboard a 6 m outboard- powered boat, RV Grampus, within a research area of ca. 20 x 15 nautical miles (nmi). To maximise the chances of evenly sampling all individuals present at Kaikōura at any given time, a standardised survey protocol was followed. For this purpose, the research area was divided into blocks of 4x4 nmi and the decision in which block to start a survey was made each day based on previous effort and weather conditions. The start location within each block was randomly generated. Sperm whales were tracked acoustically with a custom-built directional hydrophone (Dawson 1990) until they were visually located at the surface. For each whale encounter, I aimed to obtain:
- A photograph of the tail flukes for individual identification, taken at the time of
diving (‘fluke-up’), using a digital SLR camera (Nikon D750, D2H or D3, with a
Nikkor 300 mm lens)
- An acoustic recording to estimate size via inter-pulse intervals
- Sloughed skin for stable isotope analysis
Search effort and encounter data were logged via a custom written program running on a palmtop computer (HP 200LX) interfaced with a GPS, or on a tablet (Samsung Galaxy Tab
A) running Cybertracker software. Chapter 2 – Intra-population variability of isotope ratios 23
Figure 2.1. Locations of sperm whale skin collection at the Kaikōura Canyon, New Zealand. Depth contours show 500, 1000, 1500 and 2000 m isobaths. The study area is bounded by the red dotted line (offshore bounds are 12 nmi navigational limits). Only the locations of collected samples that were subsequently analysed are included in the map.
The identity of each whale was determined via photographic identification of the nicks and notches along the tail flukes (Childerhouse et al. 1995, 1996), and matching to the existing
Kaikōura photo-ID catalogue held by the MMRG.
Samples of sloughed skin (Fig. 2.2) were collected using a dip-net while following a whale at the surface or by entering the ‘slick’ after the whale had fluked. Samples were kept on ice packs until the end of each field day and stored frozen until analysis. Skin type was classified as type 1 (sheet-like, with a robust structure; Fig. 2.2b) or type 2 (strands lacking a robust structure, and presumably less fresh than type 1 skin; Fig. 2.2c), with the two types being clearly distinguishable. This classification was done to account for potential differences in
δ15N and δ13C due to skin freshness, as skin degradation could result in the uneven loss of isotopes and thus alter their ratios. Chapter 2 – Intra-population variability of isotope ratios 24
a b c
10 cm 1 cm 1 cm
Figure 2.2. Sloughed sperm whale skin. (a) Collection from the sea surface; (b) skin type 1; (c) skin type 2.
The body length of each whale was estimated from the structure of echolocation clicks
(section 2.2.3), using acoustic recordings obtained after the whale dived. To obtain a recording, a custom-built stereo hydrophone array (Barlow et al. 2008a) on a 50 m cable was deployed while the whale was at the surface and recovered 10 min after fluke-up. The hydrophone elements were 5 m apart and each contained a 40 dB pre-amplifier with a 3 dB/octave high-pass filter (corner frequency 3.39 kHz) to filter out most low-frequency engine and flow noise. While this particular array was not calibrated, an identical array
(Barlow et al. 2008a) had a reasonably flat frequency response (±3.5 dB) from 5 to 40 kHz.
The signal was recorded with a Roland R44 digital recorder, producing WAV files at high resolution (sampling rate 96.0 kHz, 16 bit). At Kaikōura, sperm whales usually surface alone, typically spaced more than one mile apart (Childerhouse et al. 1995). Therefore, there was usually no ambiguity in attributing the recorded clicks to a particular individual. However, if other whales came close to the target whale, the acoustic recording was stopped to ensure that the target whale was the loudest.
2.2.2. Stable isotope analysis of sperm whale skin
Lipids in animal tissues are depleted in 13C relative to proteins and carbohydrates, and lipid content can vary widely among species, individuals, tissues and samples (DeNiro & Epstein
1977, Melzer & Schmidt 1987, Focken & Becker 1998). To avoid confounding the interpretation of δ13C results, it is generally recommended to extract lipids from samples prior to isotope ratio mass spectrometry (IRMS) analysis (Ruiz-Cooley et al. 2004, 2012,
Sweeting et al. 2006, Post et al. 2007). Lipid extraction, however, can cause small Chapter 2 – Intra-population variability of isotope ratios 25
fractionations in δ¹⁵N leading to biased interpretations (Post et al. 2007), so ideally δ13C is measured in lipid-free subsamples and δ15N in untreated subsamples. Unfortunately, samples of whale skin were often too small to allow subsampling. The error in δ¹⁵N caused by lipid extraction was thus quantified using skin samples that were large enough to measure δ¹⁵N in both untreated and lipid-free skin.
Skin samples were first rinsed with distilled deionized water, oven dried at 60°C for 48 h, and homogenised to a fine powder using a mortar and pestle. Extraction of lipids from skin was performed using accelerated solvent extraction (ASE; Richter et al. 1996, Bodin et al.
2009). This protocol is automated, results in marginal loss of sample material, and has little impact on δ¹⁵N compared to more traditional methods such as the ‘Bligh & Dyer’ or Soxhlet extractions (Bodin et al. 2009). These qualities are particularly advantageous in studies in which small amounts of tissue are available, and there is not enough sample for both lipid- free and bulk treatments.
Samples were individually packaged in pre-combusted GF/F filters and uniquely labelled, then transferred to 34 ml ASE cells. Lipid extraction was carried out on a DIONEX 300 ASE system (Department of Chemistry, University of Otago), using a triple extraction with dichloromethane at 70°C and 1500 psi for a static hold time of 5 min (plus 5 min heating time), 60% flushing volume and a 60 s N2 purge (Bodin et al. 2009). Samples were dried at
50°C for 12 h to evaporate any traces of solvent.
Aliquots of 1 mg (±0.1 mg) of lipid-free skin powder were packed into individual tin capsules. Samples were analysed for δ13C and δ15N by Iso-Trace (University of Otago) on a
Europa Scientific ‘20/20 Hydra’ stable isotope mass spectrometer interfaced to a Carlo Erba
NC2500 elemental analyser in continuous flow mode. Isotope ratios were normalised by three-point calibration to the international scales using two IAEA (International Atomic
Energy Agency) reference materials and an EDTA laboratory standard. Results are expressed per mille (‰) in the standard notation (Peterson & Fry 1987), as δX =
([Rsample/Rstandard]-1) x 1000, where X = the element in question and R = the ratio of the heavy over the light isotope (i.e., 13C/12C or 15N/14N). Vienna Pee Dee Belemnite (V-PDB) and atmospheric N were used as standard for 13C and 15N, respectively (precision: ±0.1‰ for δ13C and ±0.2‰ for δ15N). Chapter 2 – Intra-population variability of isotope ratios 26
It was not possible to know where on the whales’ bodies each piece of skin originated from.
Studies of small-bodied cetacean species have found isotopic homogeneity throughout the skin on the whole body (Arregui et al. 2017), but it is not known if similar homogeneity applies to large cetaceans. To quantify the variability associated with the isotopic signatures of an individual whale at any given time, I measured the differences in δ13C and δ15N among pairs of skin samples collected from the same individual on the same day.
2.2.3. Body length estimation
Sperm whale body length can be estimated acoustically from the pulse structure of their echolocation clicks. The inter-pulse interval (IPI) is the time lag between consecutive pulses being reflected within the whale’s head, representing the time taken for sound to travel the length of the spermaceti sac (Norris & Harvey 1972, Møhl et al. 2003). The IPI is in turn proportional to total body length (Gordon 1991, Rhinelander & Dawson 2004, Growcott et al. 2011), which can be estimated according to the allometric relationship total length = 1.258
IPI + 5.736, established in previous studies at Kaikōura (Growcott et al. 2011).
Recordings were analysed using a Pamguard (v 1.6) IPI plug-in developed by Brian S. Miller
(detailed in Growcott et al. 2011, Miller et al. 2013b), and IPI measurements were calculated as in Miller et al. (2013b), using recordings containing a minimum of 100 clicks. Miller et al.
(2013b) concluded that at least six months between recordings were necessary to detect growth in the length of sperm whales at Kaikōura. Therefore, if an individual whale was recorded multiple times in the same season (i.e., < 3 months apart), the average total length for that season was calculated based on the available recordings.
There were nine cases in which a whale had been sampled for skin but no acoustic recordings were available for that particular season. In these cases, I extrapolated body length from estimates obtained in the previous or subsequent season, using acoustically derived growth rates specific for the whales at Kaikōura (Miller et al. 2013a). Because growth rates vary with body size (Miller et al. 2013a), size-specific growth rates (at 1m intervals) were used for this purpose. To quantify extrapolation error, a random selection of 25 acoustically derived lengths was compared with lengths estimated via extrapolation. The resulting error was ±11 cm (SE = 2), or around 0.7% of body length, which I considered acceptable for this study, given that body length estimates ranged between 12.5 and 16 m. Chapter 2 – Intra-population variability of isotope ratios 27
2.2.4. Individual residency patterns
Sighting frequencies of individual sperm whales were based on encounter histories from photo-identification data, and used as a proxy for how much they utilised Kaikōura as a foraging ground. Seasonal sighting frequencies were calculated as the number of days in which a whale was photographed in a season (summer or winter), standardised by the total number of effort days in that season. Because whale abundance varied among seasons, sighting frequencies were scaled by the number of whales sighted in each season relative to the maximum number of whales sighted in any season during the study. This way, individual sighting frequencies would not be biased high in seasons of low abundance, when each whale was likely to be encountered more often. The proxy of sighting frequency was based on the assumption that the more time whales spend foraging inside the study area, the more frequently they would be encountered.
The time taken for the isotopic signatures of a sperm whale’s skin to reflect those of its prey, known as ‘turnover time’ is not known. In bottlenose dolphins, the closest taxon for which turnover times have been identified in skin, it takes between 2 and 4.5 months for δ¹³C assimilation to reach equilibrium (Giménez et al. 2016). Turnover of δ15N takes longer, with near-complete assimilation taking between 4 and 9 months (Giménez et al. 2016). Given that turnover times lengthen with body mass (Thomas & Crowther 2015, Vander Zanden et al.
2015), and that sperm whales have thicker skin than smaller odontocetes (Rice 1989), it is likely that sperm whales lie in the upper range of these values, and possibly exceed them.
There are no published data on turnover times for δ13C in skin of any whale species, while it has been estimated to be c. 3 – 9 months for δ15N in blue whales (Busquets-Vass et al. 2017).
Therefore, it was not known if a particular skin sample would be better represented by an individual’s sighting frequency from the season in which the skin was collected (i.e., within
3 months), from the previous season (i.e., 5 – 6 months before), or over some other time period. To identify the most appropriate temporal scale at which to calculate sighting frequency for each individual, four covariates were considered in the analysis. These included the sighting frequency in the season of sampling, the sighting frequency in the previous season, the average of the two, and a global mean across the duration of the study
(i.e., 3 years). The global mean was considered for two reasons: (1) potential variability in the turnover times among whales or among samples (e.g., Browning et al. 2014, Giménez et al. Chapter 2 – Intra-population variability of isotope ratios 28
2016) may have prevented any of the other three time-lags from accurately representing the turnover time in the population, (2) although the global mean sighting frequency was not expected to reflect a particular turnover time, its larger sample size (7 seasons) may provide a better representation of the typical residency pattern of each whale. The selection of the most suitable covariate of sighting frequency is described below (statistical analysis).
The association of individual sperm whales with the Kaikōura canyon over a longer time scale was estimated as the year span between the first time they had been photographed and the time of sampling. Although it is not certain that each whale had been photographed on its first arrival in Kaikōura, this measure provided a minimum estimate of its long-term association with the area.
Table 2.1. Summary of explanatory variables used to describe individual traits of sperm whales foraging at Kaikōura.
Individual trait Definition Variants Time scale (abbreviation)
Sighting frequency Number of days SF season Field season of sample (‘SF’) in which an collection (<3 months) individual was SF previous Field season prior to sample photographed in collection (c. 5-6 months prior) a field season, SF ssn + prev Average for season of sample standardised by collection and previous season survey days SF global mean Average over study duration (3 years)
Long-term association Year span from - Length of long-term dataset with Kaikōura time first seen to (28 years) (‘Year span’) time of sampling
Body length Full body length - Estimated for each field (‘BL’) season (<3 months)
2.2.5. Statistical analysis
The variability in stable isotope ratios among samples of sperm whale skin was determined using an information-theoretic approach (Burnham & Anderson 2002, Anderson 2008). This framework allows for the simultaneous consideration of multiple hypotheses (or models), each of which can include several explanatory variables. I used generalised additive models
(GAM; Hastie & Tibshirani 1990), a data-driven approach that enables fitting smoothed non- Chapter 2 – Intra-population variability of isotope ratios 29
linear curves. This flexibility is particularly useful in ecological studies, in which relationships are often nonlinear (Guisan et al. 2002, Pirotta et al. 2011, Shillinger et al. 2011).
Within the GAM framework, I used generalized additive mixed models (GAMMs; Lin &
Zhang 1999, Zuur et al. 2009), which include random effects, in addition to fixed effects.
GAMMs were used to model the whales’ isotope ratios as a function of the explanatory variables. Whale identity was included as the random effect in all GAMMs to account for the dependency among measurements from the same individual (Zuur et al. 2009).
The explanatory covariates considered in the models included month, year, sighting frequency, year span at Kaikōura, total body length and skin type. The variable ‘year’ was included to quantify potential inter-annual variation in diet. The variable ‘month’ was included to account for fine-scale temporal variation in diet, mainly to explore its seasonal variability. ‘Month’ was preferred over ‘season’ (summer vs winter) to account for potential within-season variability. Generalisations from monthly to seasonal trends could then be inferred after analysis. Samples from the same individual collected in the same month and of the same skin type were averaged for further analysis to avoid autocorrelation. There were no cases of analysed samples from the same individual which had been collected in different months but within a month of each other (e.g., 22-Nov and 2-Dec), so this was not a source of autocorrelation. ‘Skin type’ was included as a factor in the models to account for potential variability in isotope ratios associated with using skin of two different qualities. The variables ‘sighting frequency’ and ‘year span’ were used to quantify the influence of residency patterns on isotope ratios.
Prior to model selection, two steps were necessary: selecting the time scale of the explanatory variable ‘sighting frequency’ (SF), and removing explanatory variables that were correlated with each other. (1) To select the most suitable time scale of SF, four measures were considered: SF in the season of sampling, SF in the previous season, an average of the two, and the global mean SF (section 2.2.4). Each of the four covariates was fitted independently to univariate GAMMs for the response in δ¹³C and δ¹⁵N, with whale identity as the random effect. The covariate with the lowest Akaike’s Information Criterion corrected for small sample sizes (AICC; Akaike 1973, Symonds & Moussali 2011) was selected for subsequent use in the full model. (2) Correlation among pairs of explanatory variables was investigated using concurvity tests, which are equivalent to collinearity tests (e.g., Spearman rho), but Chapter 2 – Intra-population variability of isotope ratios 30
better suited to non-linear models (Ramsay et al. 2003, Wood 2006). A threshold index of 0.3
(with 0 indicating no correlation and 1 indicating 100% correlation) was used to identify high concurvity (He et al. 2006). In cases of concurvity, univariate GAMMs were fitted using each of the correlated variables, and compared using AICC. Only the variable from the univariate model with the lowest AICC was retained in the full model. Based on this procedure, the variable ‘year span’ was removed from further analyses.
Once correlated terms had been excluded, the intra-population variability in δ¹³C and δ¹⁵N was modelled with a suite of GAMMs which included all combinations of independent explanatory variables. I modelled the effect of the predictor variables on δ¹³C and δ¹⁵N independently, as they reflect different aspects of foraging ecology. A Gaussian distribution with an identity link function was chosen to model the continuous response in δ¹³C and δ¹⁵N.
The predictor variables ‘year’ and ‘skin type’ were modelled as categorical factors. The variables ‘month’, ‘sighting frequency’ and ‘body length’ were modelled as smoothed terms, derived using thin-plate regression splines, except ‘month’, which was modelled with a cyclic regression spline. All smooths were limited to a maximum of four degrees of freedom to reduce the risk of overfitting (Vaughan & Ormerod 2005, Moore et al. 2010). Smoothing curves derived from the best GAMMs were used to interpret the specific effect of each explanatory variable on the whales’ isotopic signatures. Statistical tests were run in R-Studio v. 1.1.3 (R development Core Team 2012), using the packages ‘mgcv’ and ‘gamm4’ (Wood
2006, 2016, Wood & Scheipl 2017).
Models were ranked according to Akaike’s information criterion, with the best model indicated by the lowest AICC value (Burnham et al. 2011, Symonds & Moussali 2011). AICC combines fit to the data and model simplicity, so that if two models have similar fit, the one with fewer parameters is favoured. Akaike weights (or model probabilities) were computed for each model, calculated as the model likelihood relative to the sum of the likelihoods of all models in the set. Interactions terms were not included to facilitate interpretation of the fitted functions (Suárez-Seoane et al. 2002, Rayment et al. 2014), and because there was no prior reason to do so. Diagnostic plots (histograms of residual distributions and normal Q-
Q plots) were used to verify assumptions of normality for the use of a Gaussian error structure in the GAMMs. Plots of residuals vs fitted values were used to verify homogeneity of variance (Appendix 1, Fig. A1.1). Chapter 2 – Intra-population variability of isotope ratios 31
2.3. Results
A total of 147 skin samples from 37 individual sperm whales were collected over seven seasons between January 2014 and January 2017. Isotope ratios were obtained for 107 samples. The remaining 40 samples were not analysed because they were either collected from unidentified whales (n = 7), too small for IRMS analysis (< 1 mg after sample preparation, n = 19), lost due to problems with ASE or IRMS equipment (n = 4), or archived frozen for future studies (n = 10). Of the 107 analysed samples, 17 were repeated measurements from individual whales within the same month and of the same skin type and, thus, were averaged before further analysis to avoid autocorrelation. This resulted in
90 data points in total, from 37 different whales (Table 2.2), representing 80% of the 46 individuals that were identified during the study period.
Table 2.2. Summary of samples of sloughed skin from sperm whales off Kaikōura analysed for δ¹³C and δ¹⁵N. The year number refers to the category used as explanatory variable in the GAMMs.
# whales # samples # monthly δX Season Year Months covered sampled analysed measurements
summer 2014 1 Jan, Feb 13 14 13 winter 2014 1 Jun, July 9 10 9 summer 2014/2015 2 Nov, Dec, Feb 9 15 13 winter 2015 2 May, June, July 13 17 15 summer 2015/2016 3 Nov, Dec, Feb 11 17 12 winter 2016 3 May, June, July 16 20 19 summer 2016/2017 4 Nov, Dec, Jan 6 14 9
Full study 37 107 90
The number of samples analysed per individual ranged from 1 to 8 (mean = 2.3). Most of the sampled sperm whales were encountered at Kaikōura over two to four seasons, with four individuals encountered in a single season and six individuals in five or more seasons (Fig.
2.3). Individual whales were not sampled in every season in which they were encountered: most were sampled in one or two seasons, with ten whales sampled in three or more seasons (Fig. 2.3). Chapter 2 – Intra-population variability of isotope ratios 32
(a) (b) 10 16 14 8 12 6 10 8 4
frequency 6 frequency 4 2 2 0 0 1 2 3 4 5 6 7 1 2 3 4 5 6 7 seasons encountered seasons sampled
Figure 2.3. Frequency distributions of the number of seasons (summers and winters) that the 37 sperm whales included in the GAMMs were (a) encountered off Kaikōura between Jan-2014 and Jan-2017, and (b) sampled for sloughed skin used in stable isotope analysis.
The average change in δ15N of whale skin after lipid extraction was 0.16‰ (SD = 0.14, n = 34;
Appendix 1, Table A1.1, Fig. A1.2), which was within the expected IRMS error (±0.2‰), and thus considered adequate for the purposes of this study. Therefore, whale δ15N values from lipid-extracted samples were used for further analysis. The increase in δ13C after lipid extraction was reasonably consistent, changing on average +0.38‰ (SD = 0.20, n = 34;
Appendix 1, Table A1.1, Fig. A1.2). The loss of lipids was also reflected in the decrease in
C:N mol ratios (Post et al. 2007), with an average change of -0.26 (SD = 0.12, n = 34). The isotopic difference in duplicate samples of homogenised skin powder was, on average,
0.05‰ for δ13C (SD = 0.05, n = 28) and 0.10‰ for δ15N (SD = 0.10, n = 28), falling within the analytical error of IRMS (±0.10‰ for δ13C and ±0.20‰ for δ15N). The mean isotopic differences between skin samples from the same individual collected on the same day were relatively small, very similar to the differences within the same month, and slightly smaller than the differences within the same season (Table 2.3). Overall, these results indicated consistency in isotope ratios among samples obtained from the same animal over short time scales. This provided reassurance that isotope ratios in sloughed skin were suitable to examine variability in foraging among individuals (in relation to their individual traits) and over seasonal and inter-annual time scales.
There were only four cases in which whale samples of both skin types (1 and 2) were available from the same individual and collected within a month. As an approximate indication of the variability caused by using different skin qualities, the mean difference in Chapter 2 – Intra-population variability of isotope ratios 33
isotope ratios of the four paired samples was 0.32 for δ¹³C and 0.46 for δ¹⁵N. This was larger than the difference from paired samples of the same individual collected in the same month and of the same skin type (0.19 for δ¹³C and 0.25 for δ¹⁵N), justifying the inclusion of skin type as a factor in the models to account for any potential bias.
Table 2.3. Within-individual differences in stable isotope ratios of skin from sperm whales at Kaikōura. The mean difference and standard deviation (SD) are based on paired samples of sloughed skin collected from the same individual on the same day, month or season, and of the same skin type. n = sample size.
Sample Mean difference ± SD n (sample collection δ¹³C (‰) δ¹⁵N (‰) pairs)
same day 0.17 ± 0.10 0.25 ± 0.18 14
same month 0.19 ± 0.12 0.25 ± 0.22 20
same season 0.21 ± 0.14 0.29 ± 0.23 26
2.3.1. Variability in sperm whale isotopic signatures, residency patterns and body length
Overall, values of δ¹³C in whale skin ranged between -18.14 and -15.68‰ (mean ± SD = -16.85
± 0.51, n = 90), and δ¹⁵N values ranged between 12.13 and 17.10‰ (mean ± SD = 15.55 ± 0.91, n = 90; Fig. 2.4). Season-specific SF (days sighted/survey days, scaled by whale abundance) varied between 2 and 43% (mean ± SD = 15 ± 10, n = 77 seasonal estimates), while global mean
SF varied between 1 and 26% (mean ± SD = 8 ± 7, n = 37 whales). The minimum year-span between the first and last time that whales had been present at Kaikōura ranged from 1 to 26 years (mean ± SD = 12.2 ± 9.2, n = 77). IPI estimates ranged between 5.75 ms and 8.09 ms
(mean ± SD = 6.88 ± 0.57, n = 77), and total body length varied between 12.9 and 15.9 m (mean
± SD = 14.4 ± 0.7, n = 77). The number of acoustic recordings for IPI estimation from an individual in a single season ranged between 1 and 8 (mean ± SD = 2.2 ± 1.7, n = 37), and the standard error of a whale’s total length estimated from >1 recording ranged between 0 and
9 cm (mean SD = 2 cm, n = 49).
Chapter 2 – Intra-population variability of isotope ratios 34
(a) (b) (c) 18 Summer gSF 0.0 - 0.05 BL < 14m gSF 0.06 - 0.11 BL 14 - 15m 17 Winter gSF 0.12 - 0.26 BL > 15m
16
) 15
‰ N ( N
δ¹⁵ 14
13
12
11 -18.5 -18.0 -17.5 -17.0 -16.5 -16.0 -15.5 -18.5 -18.0 -17.5 -17.0 -16.5 -16.0 -15.5 -18.5 -18.0 -17.5 -17.0 -16.5 -16.0 -15.5 δ¹³C (‰) δ¹³C (‰) δ¹³C (‰)
Figure 2.4. Variation in stable isotope ratios in sperm whale skin with (a) season, (b) global mean sighting frequency (gSF), and (c) body length (BL). Sample size = 90. For visual clarity, sighting frequency and body length are divided into categories according to approx. 1/3rd quantiles, and summer and winter months are grouped per season.
2.3.2. Temporal scale of whales’ sighting frequencies
Sighting frequencies (SF) were estimated as a proxy for the whales’ residency at Kaikōura.
Of the four temporal scales used to describe the SF, the global mean SF (mean SF across all
seasons) was the covariate with most power to predict the variability of both δ¹³C and δ¹⁵N
in sperm whale skin, with Akaike weights of 47 and 89%, respectively, and an explained
variance of 29 and 36%, respectively (Table 2.4). In the case of δ¹³C values, the whales’ SF in
the season previous to data collection (i.e., 5 – 6 months before) also had high support,
explaining 28% of δ¹³C variance. Based on these results, the global mean SF was used for
further analysis in the full GAMMs.
Chapter 2 – Intra-population variability of isotope ratios 35
Table 2.4. Correlation between isotopic signatures in the skin of sperm whales and sighting frequencies at different temporal scales. Model parameters show the performance of univariate GAMMs using four alternative covariates of whales’ sighting frequency (SF) to model the response of isotope ratios. ‘SF global mean’ = mean SF for all seasons; ‘SF season’ = SF from the season of sample collection; ‘SF previous’ = SF from the previous season; ‘SF ssn+prev’ = average SF of the season of sample collection and the previous season. ∆AICC = delta AICC relative to the best model; wi = Akaike weight (probability of model being the best in the set); Adj. R² = proportion of variance explained by each model (by fixed effects only).
δ¹³C δ¹⁵N
Temporal scale of ∆AICC wi Adj. R² ∆AICC wi (%) Adj. R² sighting frequency
SF global mean 0.0 0.47 0.29 0.0 0.89 0.36 SF previous 0.1 0.45 0.28 6.7 0.03 0.10 SF ssn+prev 10.3 0.00 0.09 4.8 0.08 0.17 SF season 3.7 0.07 0.14 12.3 0.00 0.01
2.3.3. Correlation among explanatory variables
Pairwise combinations of explanatory variables (including global sighting frequency, year span at Kaikōura and total body length) were tested for correlation using concurvity tests
(Table 2.5). The variable ‘year span’ was positively correlated with both sighting frequency and body length (i.e., whales with longer year spans at Kaikōura tended to have higher seasonal sighting frequencies and larger body lengths). ‘Year span’ was therefore excluded from further analysis.
Table 2.5. Correlation among explanatory variables used to model the variability in isotope ratios of sperm whales at Kaikōura. Concurvity values above 0.3 (identified by *) indicate correlation.
Explanatory variables Concurvity
Month vs global mean SF 0.05
Month vs Year span 0.04
Month vs Body length 0.10
Global mean SF vs Year span 0.38 *
Global mean SF vs Body length 0.25
Body length vs Year span 0.52 *
Chapter 2 – Intra-population variability of isotope ratios 36
2.3.4. Intra-population variability in isotope ratios
For explaining δ¹³C of sperm whale skin, the best model had an Akaike weight of 64% and a relatively high explained variance of 41% (Table 2.6). The model included month and mean sighting frequency as significant predictors, excluding year, body length and skin type.
There was little support for models including skin type or body length as explanatory variables (Akaike weight ≤ 5%), and these terms had no statistical significance (Table 2.6).
There was no support for models including inter-annual variability. Seasonal effects on δ¹³C variation were explained by a cyclic function over one year and a total change of c. 0.5‰
(Fig. 2.5a). Whales sampled in winter months (May – July) had lower δ¹³C than those sampled in spring and summer (November – February; Fig. 2.4a, see Appendix 1 Fig. A1.3 for raw data). The difference in δ¹³C between winter and summer months was statistically significant, as indicated by the GAMM smoother for ‘month’ (Table 2.6b, Fig. 4a), while differences between consecutive months were small. Whales with higher sighting frequencies (i.e., that spent more time at Kaikōura) had higher δ¹³C values than whales that were sighted less frequently (Fig. 2.4, 2.5). This effect was non-linear, and most pronounced at the lower range of sighting frequencies, indicating that the whales that were sighted least frequently were the most distinct from the rest of the population. According to the GAMM smooth function, the difference in δ¹³C between whales with the lowest and highest sighting frequencies was up to c. 1‰ (Fig. 2.5).
Chapter 2 – Intra-population variability of isotope ratios 37
Table 2.6. Model selection of GAMMs to explain intra-population variation of δ¹³C of sperm whales at Kaikōura. Metrics of model performance and significance of smooth terms are shown for each model with some support (Akaike weight ≥ 1%). ∆AICC = difference in AICC score relative to best 2 model in the set; wi = Akaike weight; Adjusted R = explained variance (by fixed effects only); edf = estimated degrees of freedom; coef. = parametric coefficient estimate; SF = mean sighting frequency; BL = total body length; (1|ID) = random effect of whale identity included in all models.
Model selection table:
2 Rank Model edf ∆AICc wi (%) Adj. R 1 δ¹³C ~ month + SF + (1|ID) 6 0 64 0.41 2 δ¹³C ~ month + (1|ID) 4 1.98 24 0.17 3 δ¹³C ~ month + SF + skin type + (1|ID) 7 5.22 5 0.40 4 δ¹³C ~ month + SF + BL + (1|ID) 8 5.43 4 0.42 5 δ¹³C ~ month + skin type + (1|ID) 5 7.07 2 0.15
Significance of terms retained in each model:
Skin type Year Model Month SF BL coef. coef. rank edf (p-value) edf (p-value) edf (p-value) (p-value) (p-value) 1 1.70 (p<0.001) 1.82 (p=0.006) - - - 2 1.70 (p<0.001) - - - - 3 1.69 (p<0.001) 1.81 (p=0.006) - -0.03 (p=0.75) - 4 1.70 (p<0.001) 1.52 (p=0.005) 1.59 (p=0.16) - - 5 1.69 (p<0.001) - - -0.02 (p=0.76) -
δ¹³C (‰) δ¹³C (‰)
Figure 2.5. GAMM smoother functions for the variability in δ¹³C of sperm whale skin. The y-axes show the smooth function of each variable that was retained in the best model, with the estimated degrees of freedom in brackets. The shaded area represents the 95% confidence interval of the response. The response of δ¹³C is shown as a function of (a) month and (b) mean sighting frequency. Chapter 2 – Intra-population variability of isotope ratios 38
For explaining δ¹⁵N variability, the best model had an Akaike weight of 62% and explained
39% of the variance in the response (Table 2.7). In this case, only the mean sighting frequency of sperm whales was included as a significant factor, excluding month, body length and skin type. There was some support for the model including month as an explanatory variable
(Akaike weight = 20%) and the model including skin type (Akaike weight = 8%). However, these models did not improve the variance explained (39%), and the effect of the factors was small (month edf = 0.11, skin coefficient = -0.03) and of no statistical significance (p-value =
0.29 and 0.81, respectively). There was therefore little evidence for a difference in δ¹⁵N between whales sampled in winter and whales sampled in summer, or between samples of different skin type. There was very little support for models including body length, and no support for models including inter-annual variation. The relationship between δ¹⁵N and sighting frequency was non-linear; whales that were more frequently sighted at Kaikōura had higher δ¹⁵N signatures, but once sighting frequencies increased above 0.06 (i.e., equivalent of being sighted on average at least twice in a season), δ¹⁵N values did not change much (Fig. 2.4, 2.6, see Appendix 1 Fig. A1.4 for raw data). In addition, whales with sighting frequencies above 0.06 had a narrower range in δ¹⁵N values (15 – 17‰), while individuals with lower sighting frequencies ranged from 12 to 17‰ (Fig. 2.4, 2.6). As in the case of δ¹³C, whales that were sighted least frequently had the most distinct δ¹⁵N values compared to the rest of the sampled population. According to the GAMM smooth function, the difference in
δ¹⁵N between whales with the lowest and highest sighting frequencies was up to c. 2‰ (Fig.
2.4, 2.6).
Overall, the GAMMs for both δ¹³C and δ¹⁵N indicated that inter-annual variability was small, skin quality was not an important source of isotopic variability, and larger whales did not have noticeably different isotopic signatures than smaller whales. In contrast, δ¹³C and δ¹⁵N were correlated with individual sighting frequency, while δ¹³C also varied seasonally.
Chapter 2 – Intra-population variability of isotope ratios 39
Table 2.7. Model selection of GAMMs to explain intra-population variation in δ¹⁵N of sperm whales off Kaikōura. Metrics of model performance and significance of smooth terms are shown for each model with some support (Akaike weight ≥ 1%). ∆AICC = difference in AICC score relative to best 2 model in the set; wi = Akaike weight; Adjusted R = explained variance; edf = estimated degrees of freedom; coef. = parametric coefficient estimate; SF = mean sighting frequency; BL = total body length; (1|ID) = random effect of whale identity included in all models.
Model selection table:
2 Rank Model edf ∆AICc wi (%) Adj. R 1 δ¹⁵N ~ SF + (1|ID) 5 0.00 62 0.39 2 δ¹⁵N ~ SF + month + (1|ID) 6 2.29 20 0.39 3 δ¹⁵N ~ SF + skin type + (1|ID) 6 4.00 8 0.38 4 δ¹⁵N ~ SF + month + skin type + (1|ID) 7 6.35 3 0.38 5 δ¹⁵N ~ SF + BL + (1|ID) 7 6.71 2 0.38
Significance of terms retained in each model:
Skin type Year Model Month SF BL coef. coef. rank edf (p-value) edf (p-value) edf (p-value) (p-value) (p-value) 1 - 2.25 (p<0.001) - - - 2 0.11 (p=0.29) 2.25 (p<0.001) - - - 3 - 2.24 (p<0.001) - -0.03 (p=0.81) - 4 0.06 (p=0.30) 2.24 (p<0.001) - -0.03 (p=0.82) - 5 - 2.17 (p<0.001) 1.00 (p=0.73) - -
δ¹⁵N (‰)
Figure 2.6. GAMM smoother functions for the variability in δ¹⁵N of sperm whale skin. The y-axis shows the smooth function of the variable that was retained in the best model, with the estimated degrees of freedom in brackets. The shaded area represents the 95% confidence interval of the response. The response of δ¹⁵N is shown as a function of mean sighting frequency. Chapter 2 – Intra-population variability of isotope ratios 40
2.4. Discussion
Stable isotope analyses of δ¹³C and δ¹⁵N in sloughed skin from male sperm whales off
Kaikōura were used to investigate intra-population differences in foraging. Isotopic signatures varied seasonally and according to differences in sighting frequencies among individuals, while there was no apparent correlation with body size. This study was the first to assess variability in stable isotope ratios of sperm whales in relation to season or individual characteristics, providing new insights into the foraging ecology of the species.
We found that stable isotope analysis of sloughed skin was a valuable method to investigate the foraging ecology of sperm whales. Skin samples produced relatively consistent isotopic signatures from individual whales within a short period of time (c. 1 month), allowing for comparisons among conspecifics. Furthermore, temporal variability in stable isotope ratios was small within a field season and between years, but was significantly different between field seasons (i.e., summer and winter), allowing for the identification of seasonal patterns in foraging. Lastly, we found that the use of skin samples of apparently different qualities resulted in marginal variability in isotope ratios, providing further assurance for the use of sloughed skin in isotopic studies to investigate trophic dynamics of cetaceans.
2.4.1. Inter-individual variation in foraging in relation to sighting frequency
Whales foraging off Kaikōura had a wide range of carbon and nitrogen isotope ratios, especially considering the small spatial scale of the study. A large proportion of isotopic variability was correlated with the whales’ sighting frequencies – a proxy for the relative time that each whale spent in the Kaikōura Canyon area. In this sense, individuals are not ecologically equivalent: they utilise the food resources of the Kaikōura foraging ground to different extents. Variation in δ¹³C among individual whales can indicate diversity in the organic matter source pools at the base of their food web, while variation in δ¹⁵N can indicate feeding on prey of different trophic levels within the Kaikōura canyon ecosystem. However, variation in isotope ratios among individuals can also reflect differences at the base of the food web from geographically separate ecosystems (Zupcic-Moore et al. 2017). This can happen if some individuals ‘carry’ with them the isotopic signature of a food web from a foraging area different to the one where they are sampled. Because sperm whales are highly mobile and it takes longer than two months for their skin to integrate the isotopic signatures Chapter 2 – Intra-population variability of isotope ratios 41
of their prey (Giménez et al. 2016, Ruiz-Cooley et al. 2004), this is a likely scenario for whales that do not spend much time at Kaikōura.
The sighting frequencies of individual sperm whales were used as a proxy for the relative extent that each whale used the Kaikōura Canyon area for foraging. Although individual sighting frequencies were distributed in a gradient rather than in distinct groups, whales at the two ends of the spectrum can be thought of as ‘occasional visitors’ or ‘frequent visitors’ for simplicity (van der Linde 2009). The variability in both δ¹³C and δ¹⁵N among whales was highly correlated with individual sighting frequencies, with more enriched isotopic signatures in frequent visitors. These whales also tended to have been at Kaikōura for longer year-spans. In addition, occasional visitors had more diverse isotopic signatures than frequent visitors. Occasional visitors are unlikely to have the isotopic signature of a
Kaikōura-based diet. Instead, their isotope ratios probably reflect the integrated diet from the different areas where they had been foraging prior to being sighted at Kaikōura. In contrast, frequent visitors use Kaikōura more extensively for foraging, likely incorporating the isotopic signatures of locally sourced prey.
Because the isotopic signatures of occasional visitors might reflect a different foraging ground than that of frequent visitors, direct comparisons of primary organic sources (based on δ¹³C) or trophic level (based on δ¹⁵N) are not possible. However, isotopic values of both
δ¹³C and δ¹⁵N can provide information about foraging areas at a larger spatial scale.
Latitudinal and spatial gradients in the isotope ratios of marine primary producers are reflected in organisms at higher trophic levels, including squid, fish and cetaceans (Takai et al. 2000, Marcoux et al. 2007, Popp et al. 2007, Zupcic-Moore et al. 2017). Values of δ¹³C and
δ¹⁵N in pelagic phytoplankton typically decrease at higher latitudes within an ocean basin, on a scale of up to 5‰ over 5-10 degrees of latitude (Rau et al. 1982, Wada & Hattori 1991,
Trull & Armand 2001, Lourey et al. 2003). In addition, pelagic and offshore food webs are typically more depleted in δ¹³C than benthic and inshore food webs (France 1995, Burton &
Koch 1999, Cherel & Hobson 2007), a pattern that is reflected in cetaceans feeding at high trophic levels (Díaz-Gamboa et al. 2017). Food webs of pelagic and offshore environments also tend to be shorter, resulting in lower δ¹⁵N at high trophic levels (Iken et al. 2005). Chapter 2 – Intra-population variability of isotope ratios 42
Sperm whales that were occasional visitors to Kaikōura had lower isotopic signatures than frequent visitors, by up to ~1‰ in δ¹³C and ~2‰ in δ¹⁵N (equivalent in magnitude to the isotopic change of one trophic level; Caut et al. 2009, Giménez et al. 2016). This trend would be consistent with occasional visitors coming from foraging grounds further to the south and/or further offshore, but within New Zealand latitudes. Variation in the isotope ratios of sperm whales in relation to spatial and latitudinal gradients have been recognised before
(Ruiz-Cooley et al. 2004, Mendes et al. 2007b, Zupcic-Moore et al. 2017). In sperm whales from the Gulf of California, males coming from higher latitudes had markedly lower δ¹³C and δ¹⁵N in skin than females feeding locally at lower latitudes (Ruiz-Cooley et al. 2004). In sperm whales from the South Eastern Tropical Pacific, individual variation in isotope ratios reflected different feeding regions (Zupcic-Moore et al. 2017). More specific biochemical tools, such as compound specific isotope analysis (McClelland & Montoya 2002, Popp et al.
2007, Zupcic-Moore et al. 2017), would help to distinguish if the isotopic differences among individuals reflect a different food web baseline and/or different prey.
Differences in δ¹³C and δ¹⁵N among whales with higher sighting frequencies (above ca. 0.06) were much less pronounced compared to those of whales with low sighting frequencies. This suggested that the overall isotopic variation among males was most influenced by the distinct isotopic characteristics of occasional visitors, likely coming from foraging areas beyond the canyon. Isotopic variability was less influenced by potential diet specialisation among whales that were frequently foraging within the study area. However, there is evidence that fine-scale habitat segregation might exist at Kaikōura, even if this is not reflected in isotopic signatures (e.g., Thomson et al. 2012). For example, some whales that are very frequent visitors are encountered repeatedly in the upper canyon, whereas some
‘occasional visitors’ tend to only be encountered further offshore (MMRG, unpublished data). Furthermore, in the upper canyon foraging is mostly pelagic (Miller et al. 2013b), while in deeper areas there is also a high rate of demersal foraging (Guerra et al. 2017). This suggests that there is a range of foraging habitats within the Kaikōura canyon system, which might differ in the species they support. Research on individual variation in foraging distribution, and the influence of social interactions (e.g., through competition or associations), would help understand spatial segregation among sperm whales.
Chapter 2 – Intra-population variability of isotope ratios 43
2.4.2. Time scale of sighting frequencies vs isotope ratios: insights into turnover times
The time taken for the isotopic signature of cetacean skin to reflect the signature of assimilated prey (‘turnover time’) is has only been estimated for dolphins in captivity (δ13C and δ15N; Browning et al. 2014, Giménez et al. 2016) and blue whales (δ15N only; Busquets-
Vass et al. 2017). This study’s assessment of the relationship between the whales’ sighting frequencies and their isotopic signatures at different time scales provided new insight into turnover times in sperm whales. In the case of δ¹³C, the global mean sighting frequency was the best predictor of individual δ¹³C values, but was very closely followed by the sighting frequency from the previous season, while the sighting frequency from the season of sampling was a poor predictor. This suggests that δ¹³C turnover time in sperm whale skin is likely to be around 5 – 6 months. In the case of δ¹⁵N, the global mean sighting frequency of each whale (averaged over the duration of the 3-year study) was clearly the best predictor of variability in δ¹⁵N among individuals. This did not necessarily mean that turnover time approximated three years, but rather that the average rate of an individual’s use of the
Kaikōura habitat was a better predictor of its isotopic signature in any given season than the rate corresponding to that particular season. The turnover time for δ¹⁵N in sperm whales remains uncertain, but the fact that isotope ratios did not correlate with the whales’ sighting frequencies during the season of sampling or the previous season, suggests that it is likely longer than 5 – 6 months.
2.4.3. Seasonal variability in foraging
The δ¹³C values of sperm whales varied between seasons, with samples collected during summer months being c. 0.5‰ more enriched than those collected during winter. This variability indicates seasonal differences in the use of food resources, suggesting that the variation in spatial distribution and diving behaviour between summer and winter (Jaquet et al. 2000) is driven by changes in prey or foraging habitat. Specifically, differences in δ13C may be caused by consumption of prey with different isotopic signatures and/or a change in the carbon source supporting the food web (Fry 2006). Given that δ13C turnover time in sperm whale skin is likely to approximate 5 – 6 months, δ13C signatures in whales sampled in summer are more likely to reflect the diet that was integrated over winter, and vice versa.
Chapter 2 – Intra-population variability of isotope ratios 44
It is possible that other processes might have influenced the observed seasonal changes in isotope ratios, such as variability in isotopic enrichment during growth of skin, or bacterial activity on sloughed skin. While the potential contribution from these factors is unknown, it is likely to be small compared to that caused by variability in foraging patterns. For example, based on the similarity in isotope ratios between the two skin types (with presumed different levels of degradation), fractionation of freshly sloughed skin is probably minimal.
I found no changes in the whales’ δ¹⁵N between winter and summer. δ¹⁵N in sperm whales may turn over too slowly to be a useful tool to discern seasonal variability in diet and trophic level. In general, seasonal changes in δ¹⁵N values of odontocetes are more challenging to detect than those in δ¹³C, due to long turnover times potentially swamping the isotopic signatures specific to each season. An ecological interpretation of the seasonal dynamics in the isotope ratios of sperm whales requires some knowledge of the isotopic signatures of the base of the food web and the potential prey items. This is investigated in Chapter 3.
There was not much within-season variability in δ¹³C signatures, and none in δ¹⁵N. These results suggested that measurements from summer months (November-February) or winter months (May-July) can be pooled for the analysis of seasonal trends without a significant loss of information. At a longer time-scale, the isotopic signatures of sperm whales were reasonably consistent from year to year, suggesting that no major changes in the food web occurred within the three-year period. Although the current study might reflect an already shifted baseline for the Kaikōura Canyon ecosystem, it provides a reference for future trophic studies on the sperm whale population (e.g., Guerra et al. 2018).
2.4.4. Sperm whale body size
The isotopic signatures of male sperm whales off Kaikōura were not correlated with body size, suggesting that larger whales did not have distinct diets. It is still possible that segregation in foraging habitat is influenced by size, but that a better habitat reflects higher prey densities or easier access to prey rather than differences in prey types. Therefore, habitat segregation among sperm whales according to body size (if it does occur) might be better assessed through individual patterns in spatial distribution. Although there was no correlation between body size and isotope ratios across the sampled population, the three males with notably high sighting frequencies (whales MLS70, LNL160 and HL160) had some Chapter 2 – Intra-population variability of isotope ratios 45
of the largest body lengths (14.8 – 15.5m). These individuals were also characterised by heavy isotope ratios and long year spans at Kaikōura (23 – 26yr). Interestingly, these individuals are often encountered in the upper canyon, where other whales are sighted less often
(MMRG, unpublished data). Thus, size and long-term residency (or age) could play a role in habitat segregation by providing a competitive advantage to a very few individuals.
2.4.5. Implications for isotope studies of sperm whales
The results from this study have implications for the use of stable isotope signatures in sloughed sperm whale skin. The use of accelerated solvent extraction (ASE) for removal of lipids from sperm whale skin had only a small effect on δ¹⁵N, showing that this method allows for the measurement of both δ¹³C and δ¹⁵N from lipid-extracted samples. The variability in isotopic signatures of skin samples from the same whale taken within the same season was reasonably low, indicating that sloughed skin (even small amounts of ~1 mg) can provide a consistent representation of an individual’s isotopic signatures. The use of skin with two different structural forms (with potentially different properties), did not appear to introduce significant variability to the analysis. This was the first study to address the potential bias of using different types of sloughed skin in cetaceans, so it is not known if this is a general pattern among species. However, this is consistent with the lack of difference in stable isotope ratios between sloughed skin and the upper epidermal layer (‘stratum externum’) of blue whales (Balenoptera musculus; Busquets-Vass et al. 2017). Lastly, this study supports the use of sloughed skin for analysing biochemical tracers in sperm whales. The collection of sloughed skin can yield large sample sizes and is a non-invasive method.
The isotopic patterns found in this study (such as seasonal variability, isotopic similarity between skin types, isotopic turnover for carbon of c. 0.5 years) were derived from mature males. It is uncertain if these results are applicable to young or female sperm whales, given that they are found in very different environments in the tropics and sup-tropics (Whitehead
2003). However, two points are particularly relevant for the study of the foraging ecology of sperm whales at Kaikōura. Firstly, interpretation of seasonal trends in diet should consider that samples collected at a certain time reflect the whales’ foraging habits integrated over several months before (likely 5 – 6), rather than over the season of sample collection itself.
And secondly, sperm whales with very low sighting frequencies (‘occasional visitors’) are Chapter 2 – Intra-population variability of isotope ratios 46
unlikely to reflect a Kaikōura-based diet, thus their trophic position and food web cannot be modelled based on organic matter source pools from the canyon area.
2.4.6 Implications for conservation and future research
This study highlights some issues for the management and conservation of sperm whales at
Kaikōura. Variation in foraging patterns within the population may require management strategies to address threats targeting different subgroups. For example, localised impacts on the canyon’s food resources, such as fishing or environmental variability, may have different influences on the whales, depending on their residency rates at Kaikōura. On the other hand, whales that are occasional visitors to Kaikōura may be exposed to other impacts in their foraging grounds or migration routes beyond the canyon. To date it remains unknown where sperm whales from this population go when not at Kaikōura. Data on their long-term movements and diets would be required to fill this gap and guide conservation efforts. The analysis of more specific biochemical tracers in sloughed skin (e.g., trace elements, compound-specific stable isotopes; Popp et al. 2007, Ramos & González-Solís 2012) could provide valuable insights into the geographical origin of occasional visitors. Isoscape maps of isotope signatures collated from SPOM and cetacean top predators around New
Zealand and the Southern Ocean currently do not exist. These maps would be very valuable for narrowing down the whales’ likely foraging regions. Lastly, while sperm whales strand very occasionally, it is important to maximise what can be learnt from these rare opportunities. For example, samples of various tissues, including skin from multiple parts of the body, can be used to address within-individual variability in stable isotope ratios. In addition, stable isotope analyses of teeth layers (e.g., Mendes et al. 2007a,b) can provide valuable information on long-term movements and variability in foraging.
While the population’s variability in δ¹³C values was comparable to other sperm whale populations (Ruiz-Cooley et al. 2004, 2012, Marcoux et al. 2007), the range in δ¹⁵N was quite high (~5‰, or 2 – 3 trophic levels), especially considering the small spatial scale of the study.
This suggested a wide range of foraging patterns among males. Diversity in foraging strategies can influence population stability by increasing resilience to natural or anthropogenic disturbance (Bolnick et al. 2011). Individual specialisation in foraging habitat Chapter 2 – Intra-population variability of isotope ratios 47
or targeted prey, for example, can alleviate competition among conspecifics (Rossman et al.
2015), particularly during periods of lower food availability.
2.4.7. Conclusions
Stable isotope analyses suggested that sperm whales foraging around the Kaikōura Canyon change their foraging seasonally, consistent with variations in distribution and diving behaviour. Isotope ratios also varied among whales according to their residency rates, suggesting variability in habitat use by different parts of the population. Whales that visited
Kaikōura occasionally were likely to reflect diets from foraging grounds beyond the canyon, probably from further offshore and/or further south. The use of sighting frequencies as a proxy for the extent of foraging at Kaikōura highlights the value of integrating isotope analysis with long-term sighting data in studies of foraging ecology of marine mammals.
Sperm whales are large, highly mobile, and pelagic; they forage at depth and rarely strand.
These attributes make studying their feeding habits particularly challenging, which makes stable isotope analysis of sloughed skin a convenient tool to study their diet and habitat use.
It is important, however, to acknowledge the limitations of this method, especially considering the uncertainty around parameters such as discrimination factors. The results obtained here provide a framework to examine the whales’ trophic ecology in the context of the Kaikōura Canyon food web. To better understand the sperm whales’ diet and its seasonality, information on the isotopic signatures of potential prey and primary producers is required. This is further investigated in Chapter 3.
Chapter 2 – Intra-population variability of isotope ratios 48
Chapter 3.
Trophic interactions in the Kaikōura Canyon: using stable isotope ratios to identify the food resources of sperm whales
3.1. Introduction
An understanding of the food resources used by top predators is essential for identifying their habitat requirements. While prey availability can directly affect the distribution of populations (e.g., Bearzi et al. 2008), changes to the sources of organic matter at the base of food webs can influence apex predators through more indirect pathways (Baker et al. 2007).
Submarine canyons are often hotspots for top predators, including deep-diving cetaceans such as sperm whales (Vetter 1994, MacLeod & Zuur 2005, Moors-Murphy 2014). Although this appears to be driven by high food availability, it remains poorly understood what sources fuel such high productivity, and what makes top predators especially attracted to canyon habitats.
Submarine canyons are situated at the interface between coastal and pelagic deep-sea habitats, so organic matter subsidies can originate from various sources. Complex interactions between topography and hydrography in canyons can greatly enhance pelagic primary productivity by supplying nutrient-rich deep waters to the photic zone (Ryan et al.
2005, Moors-Murphy 2014). Channelling and entrainment of particulate organic materials can also enhance deposition in canyons (Shepard & Dill 1966, McHugh et al. 1992, van
Oevelen et al. 2011). One of the main sources of organic matter is the downward flux of pelagic production as ‘marine snow’ from the photic zone (Sumich 1999), including phytoplankton and zooplankton detritus, faecal pellets, and bacteria (Smith et al. 1996). In addition, submarine canyons can act as conduits for the transport of substantial quantities of macroalgal detritus to deep-sea communities (Harrold et al. 1998, Vetter & Dayton 1998,
1999, Krause-Jensen & Duarte 2016). Chapter 3 – Food web 50
The number of sperm whales in the foraging ground of the Kaikōura Canyon has declined over the last three decades (van der Linde 2009, Somerford 2018). It is possible that this trend reflects a shift in whale distribution, driven by oceanographic or ecological changes that affect food availability. It is therefore important to investigate what sources of organic matter sustain the whales’ food web, and which prey types are important in their diet. The Kaikōura
Canyon has high levels of net primary production in epipelagic waters (Leduc et al. 2014) and is very close to coastal macroalgal communities that line the shore (Shears & Babcock
2007). Photographic surveys have shown macrophyte detritus on the canyon seafloor (De
Leo et al. 2010). Pelagic primary production and subsidies from coastal macroalgae are thus two likely sources contributing to the high productivity of the canyon.
The only published information on the diet of sperm whales in New Zealand comes from an analysis of the stomach contents of 133 whales caught in the wider Cook Strait/Kaikōura region during the 1960s (Gaskin and Cawthorn 1967). Their analysis showed that the whales’ main prey was Onychoteuthid squid (commonly known as warty squid), but that demersal fish made up a high proportion of their diet (about a third of their stomach contents by weight, based on quantitative analysis of nine whales). The analysis also suggested seasonal differences in prey consumption, with some demersal fish being caught mostly in winter.
However, stomach contents analyses can only provide a ‘snapshot’ of diet over the previous hours or days, and the Gaskin & Cawthorn study took place nearly 60 years ago. Considering the decline in the number of sperm whales foraging in the Kaikōura region, it is important to learn about their current diet. Given the extreme benthic productivity of the Kaikōura
Canyon (De Leo et al. 2010), and that demersal foraging is one of the feeding strategies of
Kaikōura sperm whales (Miller et al. 2013a, Guerra et al. 2017), demersal prey may be particularly important for them.
Stable isotope analysis (SIA) is a useful tool to investigate the food resources used by predators. Natural differences in stable isotope ratios of carbon (δ¹³C) and nitrogen (δ¹⁵N) among primary producers can be used to identify the pathways of organic matter through a food web (Fry 2006). Thus, SIA is useful to quantify the relative contributions of different sources of organic matter sustaining a consumer, even when the trophic links between them
(i.e., prey) are not well known. This ecological application has been used across a diverse range of habitats and species, from dolphins in fiords (Lusseau & Wing 2006) to flamingos Chapter 3 – Food web 51
in wetlands (Yohannes et al. 2014), and can produce valuable information for the management and conservation of biological communities. SIA is also useful to investigate predator-prey relationships (Newsome et al. 2010a, Ruiz-Cooley et al. 2012). Although the contribution of particular species to the diet of a predator cannot be resolved, SIA can quantify the importance of different prey groups over a longer time scale than stomach content analyses, as isotope ratios reflect diet integrated over several months (Peterson & Fry
1987, Newsome et al. 2010a). The shift in δ13C and δ15N that takes place as a consumer incorporates its diet into its own tissues is known as ‘trophic discrimination factor’ (TDF;
Peterson & Fry 1987). Because TDF in δ¹³C is small with each trophic transfer (around 0.5 -
1‰), values of δ¹³C in a consumer can be used to identify the organic carbon sources at the base of its food web (DeNiro & Epstein 1978, Peterson & Howarth 1987). Shifts in δ¹⁵N are larger (2 - 4‰ per trophic transfer), and are thus useful to estimate the trophic level of a consumer (DeNiro & Epstein 1981, Post 2002).
In the present study, I investigated the food resources utilised by sperm whales foraging in and around the Kaikōura Canyon. Given the high pelagic primary productivity of the canyon (Leduc et al. 2014) and the proximity of the canyon to coastal macroalgal communities, these carbon sources were considered the two most likely candidates to be at the base of the whales’ food web. Specific questions included:
1) What are the relative contributions of pelagic productivity vs coastal macroalgae
as sources of carbon fuelling the food web that sustains sperm whales?
2) What is the trophic level of sperm whales foraging at Kaikōura?
3) How does the whales’ food web vary between summer and winter?
4) What can be learnt about the relative importance of squid vs demersal fish in the
whales’ diet?
To address these questions, I used stable isotope analyses of sperm whales, potential prey, and candidate source pools of organic matter, and applied isotopic mixing models (Phillips
& Gregg 2001, Phillips 2012; e.g., Markel & Shurin 2015, von Biela et al. 2016) to investigate the relative contribution of primary producers. With these data, I aimed to better understand which biological communities and trophic pathways sustain the sperm whales.
Chapter 3 – Food web 52
3.2. Methods
3.2.1. Sample collection and preparation
Samples of SPOM, macroalgae, krill, fish, squid and sperm whales were collected from the
Kaikōura Canyon and surrounding areas between January 2014 and January 2017. Research was conducted aboard a 6 m outboard-powered boat. I aimed to account for seasonal and spatial variation in isotopic signatures of primary carbon sources (Wing et al. 2008, Guest et al. 2010, Hepburn et al. 2011) by sampling in autumn/winter (May, June, July; hereafter winter) and spring/summer (November, December, January; hereafter summer), and at different locations across the research area (Fig. 3.1).
Marine SPOM (suspended particulate organic matter) predominantly reflects photosynthetic production by marine phytoplankton (Williams & Gordon 1970) but can include organic particulates from other origins such as detritus and recycled microbial production, in addition to microzooplankton (Wakeham & Lee 1993, Middelburg & Nieuwenhuize 1998).
SPOM was therefore used to characterise ‘pelagic production’, without assuming it was exclusively primary production. Coastal macroalgae were used to represent marine macrophyte production.
Pelagic SPOM was obtained from water samples collected in summer and winter of
2015/2016, from a 5 m depth using a Niskin bottle. Collection sites were randomly distributed within the Conway Trough, upper and lower Kaikōura canyon (Fig. 3.1). In addition, deep
SPOM was collected from a depth of 300 m in an attempt to characterise the organic matter below the photic zone, where older and more heavily recycled material is potentially more abundant than newly synthesized primary production (i.e., from phytoplankton) (Wakeham
1995). The aim of collecting deep SPOM was to identify the isotopic signature of microbially recycled pelagic material. SPOM samples were pre-screened by passing through a 200 μm mesh to remove large particles and zooplankton, and then filtered through pre-combusted
GF/F filters (0.7 μm pore size) with the aid of a vacuum pump. Filters with SPOM were oven dried at 40-50°C for 12 h and stored in a desiccator until analysis. Before analysis, filters were acid-fumed with HCl (11.65 M) for 8 h at room temperature to remove carbonates, and oven dried at 50°C for 12 h to remove any traces of acid (Lorrain et al. 2003). Filters were packed into tin capsules for isotope ratio mass spectrometry (IRMS). Chapter 3 – Food web 53
Figure 3.1. Locations of samples collected for stable isotope analysis in the Kaikōura region. Sampling sites of macroalgae, fish, and krill represent multiple specimens collected from the same location, while SPOM, squid and sperm whale locations represent individual samples. The exact locations of some fish and squid provided by fishermen were unknown and therefore excluded from this map. Depth contours show 500, 1000, 1500 and 2000 m isobaths, and the study area is represented by the red dotted line (offshore bounds are 12 nmi navigational limits).
Pelagic SPOM was also visually inspected to aid the interpretation of its origins. Samples of
120 ml were collected and preserved with Lugol solution (1% concentration). Samples were screened through a 200 μm filter and placed in a sedimentation chamber for 24 h. SPOM contents were then examined under a compound microscope (20x magnification) by counting and estimating the area of cells and organic matter particles (such as fragments of faecal pellets). Cells were identified into broad taxa, including diatoms, dinoflagellates, and ciliates.
Macroalgae were collected from three sites via SCUBA and shore collection (Fig. 3.1), during in summer and winter of 2015/2016. The most dominant species in the area were chosen based on highest algal biomass (Shears & Babcock 2007). These included the following large Chapter 3 – Food web 54
macroalgae: Ecklonia radiata, Marginariella boryana, Lessonia variegata, Landsburgia quercifolia and Carpophyllum maschalocarpum (Shears & Babcock 2007), in addition to Durvillaea willana.
All these species of macroalgae are palatable to invertebrates and/or fish (Villouta et al. 2001,
Cornwall et al. 2009, Duarte et al. 2010, Suárez Jiménez et al. 2015, Baker et al. 2016), so have the potential to provide a direct source of organic carbon to pelagic and deep-sea communities. Although D. willana was not recorded as a dominant kelp species in the sites surveyed by Shears & Babcock (2007), it was included in this study due to its high biomass along the coast fringing the study area, further south of the sites they surveyed. Between two and three plants of the four most abundant species were collected at each site. These were gathered from different depths (0-5 m, 5-10 m, 10-15 m) whenever available, to account for potential isotopic variability driven by depth (Hepburn et al. 2011). Macroalgae were rinsed with distilled water and stored frozen until analysis. For analysis, macroalgae samples were gently scraped with a blunt blade to remove epiphytes (Dauby & Poulicek 1995), oven dried at 60°C for 48 hours, and homogenised into fine powder using a ball mill machine. Aliquots of 2 mg (±0.1 mg) were packed into tin capsules for IRMS. The means of δ13C and δ15N across the three sites were used to characterise the macroalgal source pool used in the isotopic mixing models (see below). This average was not weighted according to species biomass because the relative abundance and contribution of each species to the ecosystem was unknown. Although I analysed fresh samples, macroalgae are more likely to enter the food web as detritus (Krumhansl & Scheibling 2012, Krause-Jensen & Duarte 2016). Stable isotope values of macroalgae are fairly robust to decomposition (Fenton & Ritz 1988, Markel &
Shurin 2015), so I assumed minimal influence on isotope values.
Krill were collected to represent low-trophic-level consumers of the pelagic food web and aid with food web interpretation. Euphasiid krill were collected opportunistically from aggregations at the surface (Fig. 3.1), kept on ice packs until the end of each field day, and stored frozen until analysis. Individual krill were dissected under a stereo-microscope to extract the abdominal muscle (third and fourth abdominal segments), excluding the exoskeleton and gut (Schmidt et al. 2003).
Fish and squid were sampled opportunistically to characterise the potential prey targeted by sperm whales. Fish were supplied by commercial fishermen using demersal set nets at depths of 250 – 500 m (Fig. 3.1). Squid were collected from the water surface, fresh beach- Chapter 3 – Food web 55
strandings, and from the stomachs of fish caught in set nets (Fig. 3.1). Squid muscle samples were cut from the edge of the mantle. Fish samples were dissected from the dorsal muscle behind the skull, or from the ventral muscle behind the jaws if the dorsal muscle could not be obtained. Fish and squid samples were rinsed with distilled water and stored frozen until analysis. The fish and squid used in this study included:
- species recorded as frequent prey for sperm whales in the Cook Strait/Kaikōura
region, based on stomach contents analyses (Gaskin & Cawthorn 1967), including
warty squid (Onykia sp.), southern arrow squid (Nototodarus sloanii), ling (Genypterus
blacodes) and hāpuku (also known as groper, Polyprion oxygeneios)
- species known to be occasional prey for sperm whales (Gaskin & Cawthorn 1967),
including giant squid (Architeuthis sp.), kitefin shark (Dalatias sp.) and dory species
(Cyttus sp)
- species that are not confirmed prey items of sperm whales but are common in the
Kaikōura Canyon, have similar body sizes and depth distributions as the confirmed
prey species, and could therefore be potential prey, including hoki (Macruronus
novaezelandiae), bluenose (Hyperoglyphe antarctic), hake (Merluccius australis) and blue
shark (Prionace glauca)
Opportunistic sampling of prey items in faecal plumes was carried out while following sperm whales at the surface. Squid beaks were collected with a dip-net, cleaned with fresh water, dried at air temperature and stored in sealed bags. I also recorded opportunistic observations of dead squid with tooth marks presumed to be from sperm whales.
Sloughed sperm whale skin was collected with a dip-net while following a whale at the surface or entering the slick after it had fluked (Fig. 3.1), as detailed in the methods of
Chapter 2. The identity of each whale was determined via photographic identification using the unique pattern on the trailing edge of the flukes (Childerhouse et al. 1995).
3.2.2. Lipid extraction and stable isotope analysis
Muscle and skin samples from all animal specimens were subsampled to measure δ13C from lipid-extracted material and δ15N from untreated material (Sweeting et al. 2006, Post et al.
2007). In the case of sperm whales, δ13C and δ15N values were obtained from lipid-extracted samples, since the error in δ¹⁵N caused by lipid extraction of skin is within the analytical Chapter 3 – Food web 56
error of isotope ratio mass spectrometry (Chapter 2). The methods for sample preparation, lipid extraction and IRMS were the same as those detailed in Chapter 2. Measurement of stable isotope ratios of δ13C and δ15N were performed by Iso-Trace (University of Otago).
Approximately 20% of the samples were measured in duplicates (i.e., two aliquots of the same sample) to estimate the precision of IRMS from this study. The isotope ratios from duplicate samples were averaged for further analysis.
If more than one skin sample was obtained from the same whale within a season, the average isotope ratios were used for further analysis. This way, a maximum of one isotopic signature per season was available for each whale. Sperm whales at Kaikōura have a wide range of sighting frequencies (Childerhouse et al. 1995), which are assumed to reflect the relative extent to which they use the canyon area for foraging (Chapter 2). To maximise the likelihood of characterising the Kaikōura-based food web, whales that used the canyon infrequently were excluded from this analysis. Thus, samples from whales with an average sighting frequency of fewer than 2 days/season (i.e., 0.06 after standardising for field effort) were excluded. Although there was no clear bimodality in the distribution of sighting frequencies, a cut-off of 2 days/season was chosen to include only whales that were re-sighted at least once within a season (frequency distribution in Appendix 1, Fig. A1.5). This decision rule excluded 22 whales, leaving a total of 45 samples from 15 different whales.
3.2.3. Determination of organic matter sources sustaining sperm whales at Kaikōura
In order to trace the relative contributions of different organic matter sources to a consumer’s diet, those source pools need to be isotopically distinct (Phillips & Gregg 2001, Phillips 2012).
Differences in δ13C and δ15N between the SPOM and macroalgal pools were assessed based on their means and 95% confidence intervals (CI). The statistical significance of the difference in means was tested using two-sample t-tests for unequal variances (Harraway 1993), with significance set at the 5% level (α = 0.05). Prerequisites of normality were evaluated using a
Shapiro-Wilk test (Shapiro & Wilk 1965). Because deep SPOM was isotopically indistinguishable from pelagic surface SPOM (see results), deep SPOM was not used as a proxy for microbially recycled material and was excluded from further analysis.
The isotopic setting for sperm whales was visualised by plotting the individual δ15N and δ13C values from whales sampled in summer and winter in the context of candidate source pools Chapter 3 – Food web 57
and their isotopic projections (i.e., their estimated signature at increasing trophic levels). The relative importance of pelagic surface SPOM and coastal macroalgae in fuelling the whales’ food web was investigated using two different approaches: isotopic mixing models and food web δ13C-δ15N correlations.
3.2.3.1. Isotopic mixing models and trophic level estimation
Isotopic mixing models (Phillips & Gregg 2001, Phillips 2012) estimate the relative contribution (or ‘mixture’) of the primary sources of organic carbon sustaining a consumer.
The models are based on the δ13C values of the candidate source pools and the consumer, and on an estimated value of isotopic shift (or trophic discrimination factor, denoted TDF or
∆X) for each step in the food web. A two-step iterative method was used to calculate the contribution of each organic matter source and the trophic level of individual whales.
- Step 1: a two-source model (following Phillips & Gregg 2001) was used to estimate
the relative contribution of coastal macroalgae and pelagic production to each sperm
whale using δ13C, based on the following mass balance equations:
13 13 13 δ Cwhale = fA δ CA + fB δ CB
fA + fB = 100%
where f is the contribution of each source, and δ13CA and δ13CB are their isotopic values
corrected for isotopic shift. The results of this model were used to estimate the
corresponding δ15N of the ‘mixture’ of organic matter sources supporting each
individual (δ15Nmix base), and the trophic level (TL) of each individual, according to the
following equations:
15 15 15 δ Nmix base = fA δ NA + fB δ NB
15 15 TL = (δ Nwhale − δ Nmix base)/∆N
where δ15NA and δ15NB are the isotopic values of each source, and ΔN is the TDF for
δ15N. The TL of primary producers is designated as 0.
- Step 2: the estimate of trophic level was iterated back into the mass balance model
until a stable solution was reached for the mixture of organic matter sources and
trophic level. Chapter 3 – Food web 58
I used the average TDF for aquatic marine environments of +0.96‰ (SE ±0.18) for δ13C and
+2.39‰ (SE ±0.18) for δ15N (Caut et al. 2009). These values were chosen because a distinction is made between TDFs measured in marine vs freshwater environments2, and marine TDFs are typically higher at around +1‰ (France & Peters 1997, Michener & Lajtha 2007, Caut et al. 2009). For the last trophic step (prey to sperm whale) I used the closest available taxon and tissue-specific values of +1.01‰ (SE ±0.18) for δ13C and +1.57‰ (SE ±0.26) for δ15N, from bottlenose dolphin skin (Giménez et al. 2016, consistent with Browning et al. 2014).
The accuracy of mixing model estimates can be greatly influenced by how well the TDFs reflect the true trophic enrichment in a food web. Due to the high variability in TDFs among taxa, diet, environment and trophic level (McCutchan et al. 2003, Caut et al. 2008), and the difficulty of measuring TDFs in food webs in the wild, this assumption is easily violated in marine food web studies (Vanderklift & Ponsard 2003, Newsome et al. 2010b, von Biela et al.
2016). The results of the mixing model were therefore tested for sensitivity to trophic shift variation by altering the assumed TDF in δ13C and δ15N by ±1 SE (as in McLeod & Wing 2007,
Jack & Wing 2011), and recalculating the relative contributions of organic matter sources and the estimated trophic level of sperm whales.
Another assumption of the mixing models is that the organic matter source pools are well represented by their isotopic signatures. Due to the difficulty of characterising the base of pelagic food webs from SPOM (Bowes & Thorp 2015), this assumption can be a source of error in food web modelling. The mixing models were tested for sensitivity to trophic source variation by altering the δ13C signatures (both macroalgae and SPOM) by ±1 SE of the measured values. In addition, due to the variation in palatability of SPOM to primary consumers (e.g., zooplankton), not all its constituents are grazed in the same proportion
(Meyer & El-Sayed 1983, Haberman et al. 2003). One way to account for the potential effect of this selective filter is to estimate the isotopic signature of the SPOM that is transferred up
2 The study by Caut et al. (2009) proposes a method to estimate values of TDF in cases where these cannot be measured experimentally. This method is based on linear relationships between the stable isotope composition of diet and isotope discrimination factors from a meta-analysis. This application has been criticised (Perga et al. 2010) because such relationships do not necessarily reflect relevant trends that can be extrapolated to the field. In the present study, I did not use the extrapolation method, but simply used the reported average values of TDF for marine environments, with the unavoidable assumption that the TDF values in the food web studied here will fall within the range of those reported by Caut et al. (2009). Chapter 3 – Food web 59
the pelagic food chain (‘grazed SPOM’) based on the signature of pelagic primary consumers:
∆Xgrazed SPOM = δXprimary consumer − TLprimary consumer ∆X
The isotopic signatures of euphasiid krill Nyctiphanes australis and Euphausia lucens were used for this purpose, as they are low trophic level consumers near the base of the food web, and abundant at Kaikōura (Bradford 1972). I assigned krill a trophic level of 1.5, based on a mixed diet of phytoplankton, micro-zooplankton and detritus (Stuart 1986, Dalley &
McClatchie 1989, Ritz et al. 1990, Stuart & Pillar 1990). The estimated isotopic signature of selectively grazed SPOM was included in the sensitivity model as an alternative way to characterise pelagic production. This approach did not account for how SPOM might be transferred up the food web through a benthic pathway (e.g., after sinking to the seafloor), but provided an additional line of analysis to characterise organic matter sources and to test the robustness of the mixing models.
Seasonal differences (summer vs winter) in the organic source contributions and trophic level of sperm whales were evaluated with two-sample t-tests. Based on seasonal differences in the isotope ratios of sperm whales, separate models were created for whales sampled in summer and winter.
3.2.3.2. Food web δ13C-δ15N correlations
High δ13C-δ15N correlations within food web communities typically indicate a single and isotopically homogenous organic source pool, while low correlations point to a wider array of organic sources, or a single but intrinsically heterogeneous source (Polunin et al. 2001,
Fanelli et al. 2011a, 2013). This is based on the principle that the use of multiple sources of organic matter by consumers will be reflected in a wider array of δ13C values at a certain trophic level, weakening the correlation with δ15N among individual consumers (Fannelli et al. 2011a). The δ13C-δ15N correlation between sperm whales and assumed prey was first estimated to assess whether there was more support for a food web fuelled by a dominant carbon source or by a mixture. The strong correlation (see results) suggested a dominant carbon source. Thus, SPOM and macroalgae isotope ratios were added independently to the sperm whale and prey trophic community to compare the strength of the relationships that Chapter 3 – Food web 60
included each alternative source. The relationships were estimated using simple linear regressions, with the regression coefficient (R2) as a measure of strength of each relationship.
This approach was used to compare the strength of evidence supporting a mostly pelagic versus a mostly macrophyte-derived food web, without attempting to estimate their relative contribution. The correlations did not require prior knowledge on the TDF between trophic levels or on the trophic position of sperm whales.
3.2.4. Isotopic position of sperm whales relative to their potential prey
The estimated signature of the sperm whales’ diet was compared to the isotopic signatures of potential prey species. This was done to assess the potential contribution of squid and fish to their diet and explore possible predator-prey relationships. The δ13C and δ15N values of sperm whale diet were projected from the whales’ mean isotopic signature and calculated as
δXdiet = δXsperm whale – TDFX, where X is 13C or 15N. This equation used TDF values of +1.01‰
δ13C (SE ±0.18) and +1.57‰ δ15N (SE ±0.26) from bottlenose dolphin skin (Giménez et al.
2016). Quantifying the relative contribution of different fish and squid groups to the diet of sperm whales was beyond the scope of this study, as it would have required samples of the full suite of prey species. This was not possible because not all the potential prey species within the ecosystem could be sampled or are known.
3.2.5. A note on reporting isotopic variability
As in Chapter 2, summarised data on the isotope ratios of sperm whales and other taxa are presented as means ±1 standard deviation (SD), rather than standard errors (SE). This was to better reflect the variability among samples and to allow for comparisons with other studies on the isotopic signatures of sperm whales (Marcoux et al. 2007, Ruiz-Cooley et al.
2012, 2014). In cases in which the difference in means between groups was being statistically evaluated (based on a t-test), 95% CIs are presented. SE are used, however, in the sensitivity analysis and to express uncertainty around mean TDFs from literature values, as this is the metric used in the literature (McCutchan et al. 2003, Caut et al. 2009, Giménez et al. 2016), and the measure of error typically used in studies employing isotopic mixing models
(Phillips & Gregg 2001, McLeod & Wing 2007, Markel & Shurin 2015).
Chapter 3 – Food web 61
3.3. Results
3.3.1. Isotopic composition of consumers and primary producers from Kaikōura
A total of 213 isotopic signatures were obtained from sperm whales, potential prey (three squid and eight fish species), two krill species, and two organic matter sources (pelagic
SPOM and large coastal macroalgae) (Fig. 3.2a, b, Table 3.1). A total of 45 samples were used from 15 different whales. Of the species previously identified as sperm whale prey (Gaskin
& Cawthorn 1967), a minimum of three samples was obtained for warty squid, giant squid, arrow squid, ling and hāpuku. The variation in δ13C and δ15N measured from duplicate samples of all taxa (Appendix 1, Table A1.1) was, on average, ±0.06‰ for δ13C (SD = 0.07, n
= 51) and ±0.14‰ for δ15N (SD = 0.17, n = 51), falling within the expected error of IRMS
(±0.10‰ for δ13C and ±0.20‰ for δ15N). Lipid extraction resulted in the δ15N values of fish and squid changing on average by ±0.3 and ±0.5‰, respectively (Appendix 1, Table A1.1).
δ15N from untreated samples were therefore used for further analysis, in conjunction with
δ13C from lipid-free samples.
Coastal macroalgae showed a wide range of δ13C values, with a relatively narrow range of
δ15N. Surface SPOM had a narrower δ13C range than macroalgae, but a slightly wider δ15N range (Fig. 3.2b, Table 3.1). SPOM from 300 m depth was not isotopically distinct from surface SPOM (p = 0.34, Fig. 3.2b, Table 3.1), therefore it could not be used to identify a trophic pathway of recycled production distinct from new production and was not used for further analysis.
Chapter 3 – Food web 62
(a) 18
16
14
12
10 Sperm whales )
‰ Sperm whale prey 8
N ( N Krill δ¹⁵ 6 Surface SPOM Deep SPOM (300m) 4 Macroalgae 2
0
-2 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8 δ¹³C (‰)
(b) 18
16 Sperm whales
14 Hāpuku Ling
12 Kitefin shark Dory
10 Arrow squid N (‰)
δ¹⁵ Warty squid 8 Giant squid Krill 6 Surface SPOM Deep SPOM 4 Macroalgae
2 -28 -26 -24 -22 -20 -18 -16 -14 -12 δ¹³C (‰)
Figure 3.2. Isotopic composition of sperm whales, potential prey, krill, and organic matter source pools from Kaikōura. (a) Raw data on isotopic signatures; (b) Isotopic means ± 1 SD. Only prey known to be targeted by sperm whales are shown here; see Fig. 3.5 for all sampled potential prey. The δ¹³C values are from lipid-free samples. Note that plots (a) and (b) have different scales. Chapter 3 – Food web 63
Table 3.1. Measured values of δ15N, δ13C and C:N (mean ± 1 SD) of organisms collected at Kaikōura. δ13C and C:N values of consumers are from lipid-free samples.
C:N mol n δ¹⁵N (‰) δ¹³C (‰) ratio
Top predator
Sperm whale (Physeter macrocephalus) 45 16.0 ± 0.4 -16.7 ± 0.4 4.2 ± 0.2
Secondary/tertiary consumers – species previously identified as sperm whale prey
Giant squid (Architeuthis sp.) 4 13.8 ± 1.6 -17.9 ± 0.5 2.9 ± 0.5
Warty squid (Onykia sp.) 6 13.0 ± 1.6 -19.3 ± 1.2 2.2 ± 0.6
Southern arrow squid (Nototodarus sloanii) 9 13.0 ± 1.1 -18.1 ± 0.8 3.6 ± 0.1
Ling (Genypterus blacodes) 11 15.8 ± 0.7 -16.7 ± 0.5 3.6 ± 0.0
Hāpuku (Polyprion oxygeneios) 13 14.7 ± 1.0 -17.5 ± 0.6 3.6 ± 0.0
Kitefin shark (Dalatias sp.) 2 14.7 ± 0.1 -16.7 ± 0.0 3.2 ± 0.0
Silver dory (Cyttus sp.) 1 15 -16.8 3.6
Secondary/tertiary consumers
Hake (Merluccius australis) 2 14.9 ± 1.8 -17.6 ± 0.8 3.7 ± 0.0
Hoki (Macruronus novaezelandiae) 2 14.7 ± 0.1 -17.3 ± 0.4 3.6 ± 0.1
Bluenose (Hyperoglyphe antarctic) 3 15.8 ± 0.4 -18.5 ± 0.1 3.7 ± 0.0
Blue shark (Prionace glauca) 1 13.3 -17.9 3.1
Pelagic primary consumers
Krill (Nyctiphanes australis) 25 9.4 ± 0.3 -20.8 ± 0.3 3.9 ± 0.0
Krill (Euphausia lucens) 5 9.8 ± 0.7 -20.4 ± 0.4 4.0 ± 0.1
Primary producers
Surface SPOM (5 m) 25 4.3 ± 1.9 -25.1 ± 1.4 8.3 ± 1.3
Deep SPOM (300 m) 7 4.9 ± 1.1 -24.5 ± 1.3 8.7 ± 0.8
Large macroalgae 51 8.3 ± 0.9 -16.0 ± 2.9 30.6 ± 9.6
Chapter 3 – Food web 64
3.3.2. Contents of pelagic SPOM
Six water samples were visually examined to quantify the relative proportions of organic components in SPOM. Plankton made up the largest proportion by surface area (between 77 and 91%), with varying amounts of detrital organic matter (between 8 and 22%). Diatoms
(autotrophic primary producers) were the most abundant plankton group, closely followed by ciliates and dinoflagellates (Appendix 1, Table A1.2, Fig. A1.6). Most of the identified ciliates were heterotrophic micro-zooplankton (mostly tintinnids) and some mixotrophs
(e.g., Laboea sp). The density of ciliates was relatively high at >1/ml. The biomass of the different groups was not estimated, therefore observations of SPOM were used only as an approximate guide to its contents.
3.3.3. Organic matter source pools and seasonal differences in isotopic composition
The mean δ13C and δ15N signatures of pelagic SPOM were significantly lower than those of coastal macroalgae (p < 0.001 for both δ13C and δ15N; Fig. 3.2b, Table 3.2). There were no significant differences between the summer and winter isotopic signatures of either macroalgae (p = 0.16 for δ13C, p = 0.23 for δ15N) or pelagic SPOM (p = 0.45 for δ13C, p = 0.15 for δ15N), indicating low seasonal variability (Appendix 1, Fig. A1.7). Isotopic signatures from both seasons were thus averaged for further analysis, and the means for each organic source were used in the mixing models to estimate their relative contribution to sperm whales. Due to high wave exposure at one site during winter, macroalgae samples were collected from two sites only, resulting in a smaller sample size than summer (Table 3.2).
Based on the small variation in isotopic signatures among sites (Appendix 1, Fig. A1.8a), it was assumed that this would not bias the isotopic characterisation of macroalgae in winter.
After correcting for selective grazing by krill, the isotopic signature for ‘grazed SPOM’ was estimated at δ15N = 5.9‰ (SD = 0.4) and δ13C = -22.2‰ (SD = 0.3). Isotope ratios of corrected
SPOM and coastal macroalgae were significantly different (p < 0.001; Table 3.2). This allowed for their isotopic distinction in the mixing models.
Seasonal comparisons of the isotopic composition of sperm whales indicated a significant difference in δ13C (p < 0.001), with the mean δ13C of whales sampled in summer 0.48‰ higher than those sampled in winter (Table 3.2, Fig. 3.3). The mean δ15N was also higher in whales sampled in summer (by 0.28‰), although the statistical significance of the difference was Chapter 3 – Food web 65
marginal (p = 0.04; Table 3.2, Fig. 3.3). To assess seasonal differences in the relative contribution of carbon source pools to sperm whales, mixing model estimates of these contributions were compared between winter and summer.
Table 3.2. Seasonal differences in measured values of δ15N, δ13C and C:N of sperm whales and primary source pools from Kaikōura. Values are expressed as means ± 95% CI. Significant differences in mean isotope ratios between summer and winter are indicated by ‘*’. δ13C and C:N ratios of sperm whales are from lipid-free samples.
δ¹⁵N (‰) δ¹³C (‰) C:N mol ratio n mean (95% CI) mean (95% CI) mean (95% CI)
Sperm whales
summer 22 16.1 (15.9, 16.3) * -16.4 (-16.5, -16.3) * 4.1 (4.0, 4.2) * winter 23 15.8 (15.6, 16.0) * -16.9 (-17.0, -16.8) * 4.3 (4.2, 4.3) *
Surface SPOM
summer 10 5.0 (3.7, 6.3) -25.4 (-26.2, -24.2) 7.4 (6.7, 8.1) * winter 15 3.8 (2.7, 4.9) -24.9 (-25.5, -24.3) 9.0 (8.4, 9.6) *
overall 25 4.3 (3.5, 5.1) -25.1 (-25.7, -24.5) 8.3 (7.8, 8.8) Estimated grazed SPOM summer 8 6.0 (5.8, 6.2) -22.2 (-22.4, -22.0) n.a. winter 22 5.7 (5.5, 5.9) -22.2 (-22.4, -22.0) n.a. overall 30 5.9 (5.7, 6.1) -22.2 (-22.3, -22.1) n.a. Large macroalgae
summer 35 8.2 (7.9, 8.5) -15.6 (-16.7, -14.5) 29.3 (25.9, 32.8) winter 16 8.5 (8.0, 9.0) -16.9 (-17.9, -15.9) 33.4 (29.0, 37.8)
overall 51 8.3 (8.1, 8.5) -16.0 (-16.8, -15.2) 30.6 (27.9, 33.3)
3.3.4. Sperm whale food resources: organic source pool utilization and trophic level
The isotopic setting for sperm whales was visualised by plotting the individual δ15N and δ13C values from whales sampled in summer and winter in the context of candidate source pools and their estimated signature at increasing trophic levels (Fig. 3.3). The whales’ isotopic signatures fell between the projected signature of pelagic SPOM and coastal macroalgae, sitting closest to the projected signature of grazed SPOM at 4-5 trophic increments. The δ15N span of sperm whale signatures (1.7‰) was of approximately one trophic level (1.57 ± 0.26 SE, Giménez et al. 2016). Chapter 3 – Food web 66
22
20
18
SPOMSPOM projection fractionation 16 MacroalgaeMacroalgae projection 14 fractionation grazedcSPOMSPOM fractionationprojection 12
spermsperm whales whales summer δ¹⁵N (‰) δ¹⁵N 10 summer spermsperm whales whales winter winter
8 Macroalgae 6 grazed SPOM 4 SPOM 2 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8 δ¹³C (‰)
Figure 3.3. Isotope ratios of candidate primary carbon sources and sperm whales off Kaikōura. The δ15N and δ13C values of the carbon sources are represented by filled triangles. The estimated projected signatures at increments of +1 trophic levels (after correcting for TDF) are represented by empty triangles, showing up to 5 trophic levels. Error bars represent accumulated SE (isotopic signature SE + TDF SE). Isotope ratios of sperm whales sampled in summer are shown in red circles, and whales sampled in winter in blue circles.
3.3.4.1. Mixing model estimates
The estimated contributions of pelagic SPOM and coastal macroalgae to the whales’ food web were very sensitive to source pool characterisation and choice of TDF, with the SPOM contributing between 41 and 97% under different scenarios (Table 3.3a). The model estimates based on the measured values of SPOM and macroalgae suggested an important contribution from both sources, but slightly higher from SPOM. SPOM contribution varied from 41 to 71%. Within this range, choice of higher values for TDF resulted in higher estimated contributions of SPOM. When corrected for selective grazing by pelagic zooplankton, SPOM was the dominant source pool. In this scenario, 61 – 97% of the organic matter used by sperm whales originated from SPOM, and 3 – 39% from macroalgae. Again, the contribution of SPOM was estimated to be larger at higher TDF values. Chapter 3 – Food web 67
Table 3.3. Sensitivity analysis in the estimation of (a) relative contribution of SPOM and macroalgae to the sperm whales’ food web at Kaikōura, and (b) trophic level of sperm whales. Mixing model estimates are shown for a set of source pool isotopic signatures and of TDF values (i.e., enrichment per trophic level). The mixing model estimates are expressed as means. For clarity, the SE around each estimate (based on inter-individual differences in sperm whales’ isotopic signatures, n = 45) is not shown in the table, but was within ±1% of SPOM contribution and within ±0.03 trophic levels in all cases. For ease of interpretation, sensitivity to TDF is shown only for the shift that caused the most variation (Δδ¹³C for organic source contribution, and Δδ¹⁵N for sperm whale trophic level). Notes: 1’Grazed SPOM’ isotope ratios are those estimated after correction for a selective grazing filter (see text, section 3.3.4). The isotope ratios of source pools and values of TDF used in the mixing models are shown in Table 3.5.
(a) CONTRIBUTION OF SPOM (%) vs If TDF (Δ δ¹³C) = MACROALGAE: - 1 SE mean + 1 SE Contribution of SPOM (%): - 1 SE 41 51 62
Measured values mean 45 56 67 If SPOM + 1 SE 50 60 71 isotopic - 1 SE 61 77 93 signature = Grazed SPOM1 mean 65 79 95 + 1 SE 68 82 97
If TDF (Δ δ¹⁵N) = (b) SPERM WHALE TROPHIC LEVEL: - 1 SE mean + 1 SE Trophic level: - 1 SE 4.9 4.4 4.0 Measured values mean 5.0 4.5 4.0 If SPOM + 1 SE 5.1 4.6 4.1 isotopic - 1 SE 4.8 4.3 3.9 signatures = Grazed SPOM1 mean 4.8 4.4 4.0 + 1 SE 4.8 4.4 4.0
Chapter 3 – Food web 68
Table 3.4. Seasonal differences in the relative contribution of SPOM vs macroalgae to the diet of sperm whales at Kaikōura, and trophic level of sperm whales. The mixing model estimates are expressed as means. Notes: 1’Grazed SPOM’ isotope ratios are those estimated after correction for a selective grazing filter (see text, section 3.3.4).
SEASONAL DIFFERENCES IN ORGANIC SOURCE POOL CONTRIBUTIONS AND TROPHIC LEVEL:
Scenario 1: measured SPOM, measured macroalgae
Contribution of SPOM vs Season: macroalgae (%) Trophic level
Summer 53 (95% CI = 51 - 55) * 4.5 (95% CI = 4.4 – 4.5) Winter 59 (95% CI = 57 - 61) * 4.5 (95% CI = 4.4 – 4.6)
Scenario 2: estimated grazed SPOM1, measured macroalgae
Contribution of SPOM vs Season: macroalgae (%) Trophic level
Summer 76 (95% CI = 73 - 79) * 4.4 (95% CI = 4.3 – 4.5) Winter 83 (95% CI = 82 - 84) * 4.3 (95% CI = 4.3 – 4.4)
Table 3.5. Isotope ratios of source pools and TDF used in the mixing models to estimate the relative contribution of SPOM and macroalgae to the food web of sperm whales. Values are expressed as means ± 1 SE. Source pools have an assumed TL=0. References: 1 Giménez et al. 2016; 2 Caut et al. 2009 (marine food web).
Organic matter source δ¹³C (‰) δ¹⁵N (‰)
Measured macroalgae -15.99 (±0.41) 8.26 (±0.12)
Measured SPOM -25.10 (±0.28) 4.28 (±0.39)
Grazed SPOM -22.20 (±0.07) 5.89 (±0.07)
TDF Δ δ¹³C (‰) Δ δ¹⁵N (‰)
1.01 (±0.18) 1.57 (±0.26) Top trophic level 1
Remaining trophic levels 2 0.96 (±0.18) 2.39 (±0.18) Chapter 3 – Food web 69
The mean trophic level (TL) of sperm whales was estimated as 4.5, ranging between c. 4 and
5. TL was sensitive to variability in TDF, with differences of up to 1.2 TL (Table 3.3b). The sensitivity of TL to the variability in organic source pools was small, with differences of up to 0.2 TL.
The trophic contribution of pelagic production to sperm whales was estimated to be, on average, 6-7% higher for whales sampled in winter than for those sampled in summer (Table
3.4). This difference was statistically significant (p < 0.001) but relatively small. There were no significant differences in the trophic level of sperm whales between summer and winter
(p = 0.76 if SPOM based on measured values, p = 0.50 if SPOM corrected for grazing filter;
Table 3.4).
3.3.4.2. Food web δ13C:δ15N correlations
The δ13C of sperm whales and their potential prey showed an overall isotopic enrichment with increasing δ15N, with the δ13C:δ15N linear regression indicating a strong positive correlation (R2 = 74%, p < 0.001, Fig. 3.4). The strength of this relationship suggested a food web sustained predominantly by one carbon source pool. Thus, the isotopic signatures of candidate sources of organic matter were added independently to those of sperm whales and their prey. Inclusion of organic matter sources resulted in strong δ13C:δ15N correlations in the case of SPOM and estimated grazed SPOM (93% and 96%, respectively, p < 0.001) and a weak correlation in the case of macroalgae (3%, p = 0.6).
Chapter 3 – Food web 70
18 y = 1.3x + 37.0 16 R² = 0.74, p < 0.001
14 y = -0.3x + 7.9 12 R² = 0.03, p = 0.6
10
) y = 1.3x + 37.9
‰ R² = 0.93, p < 0.001
8
N ( N δ¹⁵ 6 y = 1.8x + 45.4 4 R² = 0.96, p < 0.001
2 Macroalgae + sperm whales and prey gSPOM + sperm whales and prey 0 SPOM + sperm whales and prey Sperm whales and prey -2 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8
δ¹³C (‰)
Figure 3.4. δ13C:δ15N correlations of potential food web scenarios. The simple linear regression equations and regression coefficients (R2) are shown for isotopic signatures of sperm whales and potential prey, as well as for the addition of pelagic production (SPOM), pelagic production corrected for grazing selectivity (gSPOM) and coastal macroalgae, as primary sources of organic matter.
The regression slope of the δ13C:δ15N correlation reflects the relationship between the TDFs of δ13C vs the TDF of δ15N. The value of the regression slope under each scenario was therefore used to assess whether the resulting TDFs would be feasible. Assuming a mean
δ15N TDF between +2.4‰ (McCutchan et al. 2003, Caut et al. 2009) and +1.6‰ (Browning et al. 2014, Giménez et al. 2016) per trophic level, the mean value of δ13C TDF would be between
+0.9 and +1.3‰ in a food web fuelled mostly by SPOM (corrected for grazing selectivity), between +1.2 and +1.8‰ if fuelled mostly by SPOM (based on measured values), and between -5.3 and -8.0‰ if fuelled mostly by exports of macroalgae. Of these δ13C TDF values, the closest to typical marine food web values (0.96 ± 0.18 SE, Caut et al. 2009) were the ones under the scenario of a food web fuelled predominantly by pelagic production, particularly when corrected for grazing selectivity.
Chapter 3 – Food web 71
3.3.5. Potential prey in sperm whale diet: isotope ratios and opportunistic observations
The projected signature of sperm whale diet was compared to the mean isotopic signatures of potential prey3. Compared to species previously recorded as prey items in the Cook
Strait/Kaikōura region (Gaskin & Cawthorn 1967), the estimated signature of sperm whale diet was lower than the mean δ13C and δ15N of fish species and higher than the mean δ13C and δ15N of squid species (Fig. 3.5). This indicated that the diet was likely a mixture of fish and squid. The mean isotopic signatures of prey were depleted in relation to sperm whales, with differences ranging from -0.02‰ δ13C, -0.2‰ δ15N (ling) to -2.6‰ δ13C, -3.0‰ δ15N
(warty squid). The mean isotopic signatures of the sampled prey species spanned ca. 3‰ in
δ13C and 3‰ in δ15N values, an equivalent to 1-2 trophic levels.
Isotopic signatures of four unconfirmed but potential prey species (bluenose, hake, hoki and blue shark) were characterised to investigate the possibility of a predator-prey relationship with sperm whales. Of the four species, hake, hoki and blue shark fell within -1.2‰ δ13C and
-2.7‰ δ15N of sperm whales and within the isotopic space of other previously confirmed prey (Fig. 3.5). Bluenose fell within -1.8‰ δ13C and -0.2‰ δ15N of sperm whales, indicating a less likely predator-prey relationship.
3 Note on prey body sizes. The warty squid collected at Kaikōura had an average mantle length (ML) of 43 cm (range 20-52), and the arrow squid of 25 cm (range 20-31). These sizes were within the range of the squid found in the stomachs of sperm whales from Cook Strait (warty squid: range 16-45 cm; arrow squid: range 21-30 cm, Gaskin & Cawthorn (1967). The isotopic signatures of the collected squid were therefore likely to reflect roughly similar life stages of those known to be targeted by sperm whales. Fish lengths were not recorded by Gaskin & Cawthorn (1967), but the collected fish sizes (c. 70-100 cm) were within the range of prey targeted by sperm whales. Chapter 3 – Food web 72
17.0
16.5 SPERM WHALES 16.0 LI BN 15.5
15.0 HK DO HO KS HA
14.5 DIET N (‰) N
14.0 δ¹⁵
GS 13.5 BS Sperm whale 13.0 WS AS Prey - squid species
12.5 Prey - fish species
12.0 Potential prey - fish species 11.5 -20.5 -20.0 -19.5 -19.0 -18.5 -18.0 -17.5 -17.0 -16.5 -16.0 δ¹³C (‰)
Figure 3.5. Estimated isotopic signature of the diet mixture of sperm whales in relation to their prey. In this figure, ‘Prey’ refers to species that have been previously identified as sperm whale prey in the Cook Strait/Kaikōura region (based on Gaskin & Cawthorn 1967) and ‘Potential prey’ refers to species that could be prey based on their similar size and distribution to previously confirmed prey. SW = sperm whales, DIET = sperm whale diet mixture, LI = ling, DO = silver dory, KS = kitefin shark, HA = hāpuku, BN = bluenose, HK = hake, HO = hoki, BS = blue shark, GS = giant squid, AS = arrow squid, WS = warty squid. Error bars are ±1 SD.
Seasonal differences in isotopic signatures of prey could not be resolved because few samples were available from summer (n = 10 from three squid and three fish species). This was due to commercial fishermen switching to inshore fisheries in summer and not having a reliable catch of the species targeted in this study. Based on this small sample size, however, the δ13C and δ15N values in fish and squid were slightly higher in summer than in winter, with a mean difference of 0.6‰ δ13C (SD = 0.7) and 1.4‰ δ15N (SD = 1.0). Chapter 3 – Food web 73
Six squid beaks were collected from opportunistic sampling of sperm whale faecal plumes.
The beaks belonged to Mastigoteuthis sp. and Nototodarus sloaneii. It was unknown if the squid were eaten directly by the sperm whales or if they were within the stomachs of other prey.
In addition, specimens of Onykia ingens were found dead floating at the surface near foraging sperm whales on five occasions. In all cases, the individuals were freshly dead. Three of these squid had puncture marks 2 – 3 cm in diameter, spaced 8 – 14 cm apart (Fig. 3.6), consistent with the tooth spacing of male sperm whales (Best 1979). The squid had mantle lengths of 46
– 52 cm.
Figure 3.6. Warty squid (Onykia ingens) from Kaikōura. These three specimens were collected from the water surface in proximity to foraging sperm whales between 2014 and 2017. Arrows indicate puncture marks, likely to be caused by sperm whale teeth.
Chapter 3 – Food web 74
3.4. Discussion
This was the first study to characterise the food web of sperm whales by using stable isotope analysis of primary producers and potential prey. The results presented here provide new insights into the diet of sperm whales in the Kaikōura region, and suggest that their food web is most likely fuelled by pelagic production. Stable isotope analysis was a useful tool to examine trophic relations in the Kaikōura Canyon, advancing our understanding of the seasonal dynamics in the use of food resources by an apex predator.
3.4.1. Sources of organic matter
Two of the major source pools of organic matter in the Kaikōura Canyon and coastal area could be differentiated based on stable isotope composition, with pelagic SPOM having a more depleted δ¹³C and δ¹⁵N signature than coastal macroalgae. The measured values of δ¹³C and δ¹⁵N of macroalgae were consistent with those measured from other temperate habitats
(Tallis 2009, von Biela et al. 2016), and specifically from New Zealand (Bode et al. 2006,
Hepburn et al. 2011, Jack & Wing 2011, Stephens & Hepburn 2014). The δ¹³C signature of
SPOM was around 2-7‰ more negative than typical values of marine phytoplankton and
SPOM from temperate latitudes (-22 to -18‰; Fry & Sherr 1989, Boutton 1991, Miller et al.
2008, Carlier et al. 2009) and from New Zealand (-22 to -20‰; Twist 2005, McLeod & Wing
2007, Wing et al. 2008), with the exception of the Otago continental shelf (Van Hale 2003).
The δ¹⁵N values of SPOM were within the range of temperate phytoplankton (4 to 6‰; Fry
& Sherr 1989, Peterson & Fry 1987, McLeod & Wing 2007), but with a higher proportion of light values. The more negative isotopic signatures of SPOM relative to typical values for phytoplankton suggested the presence of an additional source of organic matter in the pelagic waters off Kaikōura. The light δ¹³C signature in SPOM was consistent with a contribution of organic matter derived from microbial recycling, a fundamental feature of many pelagic food webs (Azam et al. 1983, Levin & Michener 2002, Wing et al. 2012).
Bacterial isotopic fractionation results in organic matter with δ¹³C values that are lower than marine phytoplankton (Robinson & Cavanaugh 1995, Levin & Michener 2002). In the process of microbial recycling, bacteria decompose organic matter and re-mineralise nutrients, returning them back into the food web as they are eaten by higher trophic levels (Pomeroy et al. 2007, Fenchel 2008). The organic matter used by bacteria can originate from various Chapter 3 – Food web 75
sources, such as dead phytoplankton and macroalgae, faecal material from marine organisms, and terrestrial material. The presence of microbially recycled production in pelagic SPOM would be consistent with (1) similar isotopic measurements of SPOM from the surface waters over the Otago continental shelf (Van Hale 2003, Van Hale & Frew 2010), where the pelagic food web has a strong component of microbial recycling (Bradford-Grieve et al. 1999, Van Hale & Frew 2010), and (2) a relatively high abundance of microzooplankton in SPOM, including ciliates. Marine ciliates graze directly on bacteria and are known as the
‘top predators of the microbial loop’ (Levinton 2001, Pomeroy et al. 2007). The isotopic character of SPOM and the presence of ciliates suggest that a pelagic microbial loop (Azam et al. 1983, Pomeroy et al. 2007) could play an important role in the productivity of the canyon’s ecosystem, in addition to primary production from phytoplankton.
If detritus from coastal macroalgae were to be microbially recycled (e.g., in the pelagic microbial loop or by benthic bacteria), its δ¹³C signature could become more negative after fractionation, and thus more similar to the values of SPOM. This could have resulted in some underestimation of the contribution of macroalgae to the food web of sperm whales.
Although the identity of the additional source of organic matter present in SPOM was unknown, the low C:N ratios (<6) reflected a pelagic marine origin (Richard et al. 1997,
Maksymowska et al. 2000). This suggested that macroalgal detritus or terrestrial organic matter (C:N ratios >12) were less likely to contribute directly to the SPOM pool than detritus originating from marine phytoplankton and other pelagic organisms. Further research is required to clarify the importance of microbial recycling at Kaikōura.
3.4.2. Contribution of organic matter sources to sperm whales
Isotopic mixing models and food web δ13C:δ15N correlations were used to estimate the contribution of organic matter resources fuelling the food web that supports sperm whales.
Estimates from the mixing models were quite sensitive to source pool characterisation and uncertainty in TDF – varying between 41 and 97% contribution from pelagic production vs coastal macroalgae. Within this range, the upper values of SPOM contribution were more likely, for three reasons. First, food webs dominated by carnivorous consumers, such as that involving sperm whales and their prey, tend to have higher δ¹³C TDFs (Vander Zanden &
Rasmussen 2001). Open-ocean food webs also have higher δ¹³C TDF than inshore ones, with Chapter 3 – Food web 76
values closer to 1.1‰ (France & Peters 1997). Therefore, the value of δ¹³C TDF applicable to the sperm whales’ food web was likely to lie within the upper range of its variability (i.e.
+1SE, see Table 3.2), making the higher estimates of SPOM contribution more likely: between
62 and 97%. Second, the contribution of SPOM to sperm whales was estimated to be 93 – 97% after correcting for pelagic grazing by primary consumers. Although this correction did not account for a benthic pathway for SPOM (e.g., utilised by benthic consumers after sinking to the seafloor), it provided further support for a higher contribution of pelagic production vs coastal macroalgae. Lastly, the high δ13C:δ15N correlations in the isotope ratios of sperm whales and their prey suggested that their food web was more likely to be sustained by a dominant source of organic matter (pelagic production) than by a mixture of two important sources.
Based on two different approaches, this study suggested that the sperm whales’ food web at
Kaikōura is predominantly fuelled by pelagic production. This is consistent with what is known about sperm whale ecology in the area. Previous studies have shown that mesopelagic squid are important prey for sperm whales in the wider Kaikōura/Cook Strait region (Gaskin & Cawthorn 1967), and that sperm whales off Kaikōura do part of their foraging in mid-water (Miller et al. 2013a, Guerra et al. 2017). The main prey of warty squid are pelagic myctophids (Jackson et al. 1998, Phillips et al. 2001), which feed mainly on copepods (Phillips et al. 2001). Copepods, in turn, prey on phytoplankton and microzooplankton, including ciliates (Calbet & Saiz 2005, Pomeroy et al. 2007). The pelagic food chain ‘copepod-myctophid-squid-predator’ has been recognised as an important trophic pathway independent of krill species in the Southern Ocean (Rodhouse et al. 1992,
Phillips et al. 2001). This could also be an important mechanism for the transfer of pelagic organic carbon to sperm whales at Kaikōura.
Coastal macroalgae did not contribute substantially to the organic matter used by sperm whales. Perhaps the macroalgal communities are too narrow to provide a large pool of organic matter, in spite of being so close to the canyon, or the bathymetry and hydrology are not conducive to cross-shelf transfer of macrophyte detritus. It remains uncertain, however, whether the contribution from macroalgae was simply smaller than that of SPOM (but still important), or negligible. Leduc et al. (2014) suggested a low input of refractory macrophyte material to the seafloor of the Kaikōura Canyon, based on the low C:N ratios of sediment Chapter 3 – Food web 77
organic matter (ranging from 6.3 – 9.6). The C:N values of macroalgae measured in this study averaged 30.6 (SD ±9.6), while those of SPOM averaged 8.3 (SD ±1.3). These observations are consistent with the hypothesis made by Leduc et al. (2014), and suggest that: (1) macroalgae detritus does not contribute much to benthic organic matter, and (2) SPOM sinking from the surface is likely to be an important contribution to the sediment organic matter in the canyon.
Previous studies at Kaikōura, however, have shown that coastal macroalgal detritus is a component of deep seafloor sediments (De Leo et al. 2010) and could provide a small carbon subsidy to benthic communities.
Lastly, in this study I collected POM from 300m depth in an attempt to characterise the isotopic signature of microbially recycled pelagic material, but this POM was not isotopically distinct from SPOM collected at the surface. Sampling material from greater depths would help to reveal the magnitude of microbial recycling of ‘marine snow’, and the link between surface productivity and benthic organic matter on the canyon’s seafloor.
3.4.3. Sperm whale diet: potential prey items
The isotopic signatures of sperm whales in relation to their prey were consistent with a diet composed of a mixture of squid and demersal fish. The projected diet mixture was isotopically enriched in relation to the three species of squid, but depleted relative to the demersal fish species, suggesting that both groups contributed significantly to the diet. The importance of demersal fish is consistent with the high rate of demersal foraging by sperm whales at Kaikōura compared to other regions (Guerra et al. 2017), and with past stomach contents analysis (Gaskin & Cawthorn 1967). Based on these data, it is likely that the particularly high densities of benthic invertebrates and demersal fish in the Kaikōura
Canyon (De Leo et al. 2010, Leduc et al. 2014) contribute to the whales’ use of demersal habitat.
Ling and warty squid were isotopically distant from the estimated diet of sperm whales; in the case of ling, by approx. 1 TL above the diet, and in the case of warty squid, by approx. 1
TL below the diet. While this suggests that neither of them was the single most dominant prey, it does not exclude them as important diet components, as the whales’ isotopic signature is an integrated mixture of the signatures of their prey. Given that warty squid has been previously recorded as the predominant prey of sperm whales in the wider Chapter 3 – Food web 78
Kaikōura/Cook Strait region (Gaskin & Cawthorn 1967), and that there have been repeated observations of warty squid targeted by sperm whales at Kaikōura (present study, Roger
Williams pers. comm.), warty squid remains likely to be an important prey item. Thus, a contribution of ling to the whales’ diet would be feasible. Hāpuku was isotopically close to the estimated diet, supporting its contribution to the suite of prey targeted by sperm whales.
Due to the opportunistic sampling of fish and squid it was not possible to collect all the prey species that sperm whales might target at Kaikōura. For example, orange roughy
(Hoplostethus atlanticus) and gemfish (or ‘southern kingfish’, Rexea solandri) were occasional prey items for whales caught in the Cook Strait/Kaikōura region (Gaskin & Cawthorn 1967), but it was not possible to obtain samples. Among squid, the species likely to be targeted by sperm whales most frequently were probably well represented. Of all the squid species known to be eaten by sperm whales in New Zealand (Gaskin & Cawthorn 1967, Clarke &
Roper 1998, Gomez-Villota 2007), the most frequently caught at Kaikōura are arrow, warty and giant squid (MPI data release OTA2015-001, B. Cooper pers. comm.). Warty squid were the predominant squid (71% by occurrence, 88% by weight) in the stomachs of sperm whales from the Cook Strait/Kaikōura region (Gaskin & Cawthorn 1967).
The sperm whales of Kaikōura reflect a food web supported mainly by pelagic production, while demersal prey seem to be an important part of their diet. This suggests that demersal prey are partly sustained by pelagic production. There are at least two mechanisms that could explain this link. Firstly, demersal prey which feed on benthic organisms could be supported by pelagic production through downward flux of POM (as suggested by De Leo et al. 2010, Leduc et al. 2012), and hence act as a trophic link between the productive benthos and sperm whales. Ling, for example, feed primarily on benthic prey, including macrourid fishes (Clark 1985, Dunn et al. 2010), which are very abundant in the Kaikōura Canyon (De
Leo et al. 2010). Secondly, demersal species targeted by sperm whales may feed on pelagic organisms whose vertical distributions are truncated by the canyon topography (e.g., during vertical diel migrations; Mauchline & Gordon 1991, Sutton et al. 2008). In this scenario, the isotopic signatures of demersal prey would reflect a pelagic food web. This is known to occur at Kaikōura, at least for some species. Arrow squid, for example, feed on mesopelagic fish and crustaceans (Dunn 2009, Pethybridge et al. 2012), and are commonly found in the stomachs of demersal hāpuku and bluenose (pers. obs., B. Cooper pers. comm.). In either Chapter 3 – Food web 79
case, the canyon’s steep topography would play a role in concentrating or aggregating food resources. Research on the food resources utilised by benthic and demersal organisms would contribute to identifying bentho-pelagic coupling in the canyon’s food web. This would help to establish whether the extreme productivity of the benthos plays a role in sustaining the whales’ food web.
3.4.4. Trophic level of sperm whales at Kaikōura
Mixing models indicated that the trophic position of sperm whales was between four and five levels above the primary producers. This is consistent with the expected number of trophic steps leading up to the sperm whales. Their likely prey (such as warty squid, ling and hāpuku) feed on fish, squid and crustaceans, that are in turn secondary and tertiary consumers (Jackson et al. 1998, Phillips et al. 2001, Dunn et al. 2010, Nguyen et al. 2015,
Tromp et al. 2016). Although there have been studies on the relative trophic position of sperm whales among populations (Ruiz-Cooley et al. 2012), and versus other cetaceans
(Ostrom et al. 1993), there are no published estimates of their absolute trophic level based on stable isotope ratios. Based on stomach contents, the trophic level for sperm whales globally has been estimated at 4.4 (Pauly et al. 1998b), consistent with the present study.
3.4.5. Seasonal differences in source pool utilisation and diet of sperm whales
The average δ¹³C value of sperm whales was 0.5‰ higher in samples collected in summer than in winter – equivalent to a change of about half a trophic level. This δ¹³C translated into a slightly lower (6 – 7%) contribution from pelagic production to sperm whales sampled in summer. Due to the lack of significant seasonal variation in the δ¹³C signatures of the organic source pools, the seasonal differences observed in sperm whale skin were likely to reflect temporal changes within the food web. The δ¹³C turnover time (i.e., the time lag for the isotopic signature of skin to reflect those of the consumed prey) in sperm whales is likely to be at least two months (Ruiz-Cooley et al. 2004, Giménez et al. 2016), and most likely about half a year (Chapter 2). Thus, δ¹³C signatures in sperm whales sampled in summer are more likely to reflect the diet that was integrated over winter, and vice versa.
Two possible mechanisms could explain the seasonal variation in isotope ratios in sperm whale skin: a seasonal shift in the targeted prey, and/or a seasonal shift in what carbon source pools are utilised by prey. The most likely explanation is a shift in prey, consistent with the Chapter 3 – Food web 80
seasonal changes in the spatial distribution and diving behaviour of foraging whales at
Kaikōura (Jaquet et al. 2000). Based on the whales’ seasonal signatures, the most likely change is a higher contribution from squid in summer (lighter δ¹³C), and a higher contribution from demersal fish in winter (heavier δ¹³C). This is consistent with temporal variations in the stomach contents of sperm whales in the Cook Strait/Kaikōura region
(Gaskin & Cawthorn 1967), and with seasonal pulses in prey species known from fisheries catches (Jaquet et al. 2000, MPI data release OTA2015-001). For example, hāpuku are thought to move inshore during the winter to spawn closer to the continental shelf (Beentjes & Francis
1999). The trend in δ¹³C is also consistent with the whales’ more offshore distribution during summer compared to winter (Jaquet et al. 2000), given the typically lighter δ¹³C of offshore/pelagic food webs (France 1995, Cherel & Hobson 2007). Lastly, although Chapter
2 suggested no seasonal variability in δ15N of whales, when occasional visitors were excluded
(present study) there was some evidence for slight changes in δ15N between summer and winter. Overall, these results suggest that seasonal patterns in the whales’ food web at
Kaikōura are likely driven by variations in the range of prey available.
3.4.6. Caveats and limitations
Although stable isotope analysis is a useful tool for unravelling the complex dynamics of food webs, it has important limitations. The absence of data on trophic discrimination factors for sperm whales and their prey meant relying on values from published literature to develop the mixing models and examine predator-prey relationships. The resulting analyses are therefore susceptible to error due to the variation among ecosystems, species, tissues, and analytical methods (Vander Zanden & Rasmussen 2001, Boecklen et al. 2011). This resulted in uncertainty around the contribution of different organic source pools. The sensitivity analysis used in this study addressed part of this uncertainty explicitly, providing a range of possible values for the relative contribution of organic matter sources to sperm whales. Although information from other marine food webs was useful for identifying the most likely scenarios suggested by the isotopic mixing models, the precise contributions of source pools could not be fully resolved.
The choice of SPOM and coastal macroalgae as the two main candidates of primary carbon sources for sperm whales involved a simplification of the food web. Other potentially Chapter 3 – Food web 81
important sources of organic matter (e.g., terrestrial material from river outputs, benthic chemoautotrophs) were not included in the analysis. Studies on the origin of organic matter in the canyon sediment (Leduc et al. 2018) and benthic consumers (e.g. McLeod & Wing 2009) will be very valuable for understanding the source pools sustaining the benthic communities off Kaikōura. The application of more specific biochemical tools (e.g., amino-acid specific isotope analysis and fatty acid tracers; McClelland & Montoya 2002, Budge et al. 2006) will be particularly useful for clarifying the potential contribution of macroalgae to the canyon’s food web and the role of microbial recycling. Recommendations for future research also include using δ34S to increase the power of identifying bacterial organic sources (e.g.,
MacAvoy et al. 2002, McLeod & Wing 2007), and compound-specific isotope analysis of sperm whale skin to provide more accurate identification of the base of the whales’ food web
(e.g., Ruiz-Cooley et al. 2014, Zupcic-Moore et al. 2017).
3.4.7. Conclusions
This study provides the first insights into the structure of the food web supporting sperm whales at Kaikōura and its basal carbon sources. The stable isotope analysis suggested that most of the organic carbon sustaining the whales is derived from pelagic production, with possibly a small input from the neighbouring macroalgal communities. The pelagic pool of organic matter appeared to be composed of phytoplankton, with a potential contribution of microbially recycled material. Stable isotope signatures of the whales and potential prey were consistent with a diet composed of demersal fish in addition to mesopelagic squid.
There was evidence for seasonal variations in the sperm whales’ food web, likely due to shifts in prey availability. These findings highlight the importance of better understanding the drivers of pelagic productivity at Kaikōura, and what physical, oceanographic and biological processes influence the sperm whales’ food resources. These mechanisms will ultimately influence the whales’ spatial distribution and the temporal fluctuations in their numbers at Kaikōura.
Chapter 3 – Food web 82
Chapter 4.
Habitat preferences of sperm whales in the Kaikōura
Canyon: environmental drivers of foraging distribution.
4.1. Introduction
The loss of top predators from marine ecosystems is a global issue (Myers & Worm 2003,
Estes et al. 2006). For previously exploited populations, reliable access to feeding opportunities is essential for recovery (Werner & Mittelbach 1981, Baker et al. 2007). The effective protection of marine top predators therefore requires knowledge of their foraging habitat: where is it, and what features make it suitable? Sperm whales were heavily hunted in New Zealand during the 19th century (Gaskin 1973, Grady 1982). The recovery of their populations is likely to be influenced by the ecological status of their feeding grounds. The area around the Kaikōura Canyon is a year-round foraging area for male sperm whales
(Childerhouse et al. 1995), and one of the few places in the world where this species is found so close to the coast.
Submarine canyons are often habitats of high productivity, enhancing biomass and biodiversity in the deep sea (Vetter et al. 2010, Van Oevelen et al. 2011). Many canyons are also hotspots for top predators, particularly deep-diving cetaceans (Vetter 1994, MacLeod &
Zuur 2005, Moors-Murphy 2014). The Kaikōura Canyon is one such hotspot, containing extremely high biomass of benthic invertebrates and demersal fish (De Leo et al. 2010, Leduc et al. 2014), and attracting an abundance of pelagic top predators (Childerhouse et al. 1995,
Benoit-Bird et al. 2004, Boren et al. 2006). The decline in the number of sperm whales foraging at Kaikōura (van der Linde 2009, Somerford 2018) may reflect a shift in the whales’ distribution, potentially driven by ecological or oceanographic changes to their habitat. It is therefore important to understand the habitat features that attract these apex predators to
Kaikōura. Chapter 4 – Habitat preferences 84
Multiple studies have investigated the drivers of sperm whale distribution (e.g., Whitehead et al. 1992, Jaquet & Gendron 2002, Pirotta et al. 2011, Wong & Whitehead 2016). Although some general conclusions can be drawn, such as a preference for deep-waters over complex topography, the environmental characteristics of their foraging habitat are highly variable.
Importantly, studies rarely use oceanographic data from the water column, rather than the sea surface, to explain habitat preferences (but see Griffin 1999, Skov et al. 2008, and Torres et al. 2011 for exceptions). There is also very little understanding of the seasonal variation in habitat preferences, and of drivers of distribution at small spatial scales.
At Kaikōura, sperm whales forage throughout the water column and in proximity to the seafloor (Miller et al. 2013a, Guerra et al. 2017), and their diet includes demersal fish as well as pelagic squid (Gaskin & Cawthorn 1967, Chapter 3 of this thesis). Their food web appears to be fuelled mostly by pelagic production (Chapter 3), while topography influences the whales’ spatial distribution (Jaquet et al. 2000, Sagnol et al. 2014b). Therefore, hydrographic features in the water column (e.g., phytoplankton distribution, thermocline depth) as well as seafloor characteristics (e.g., depth, slope), are likely to influence the abundance and distribution of prey in the canyon area. However, the environmental drivers of the whales’ foraging distribution, and in particular the influence of oceanographic factors, are poorly understood. There are significant differences in sperm whale behaviour and spatial distribution between summer and winter (Jaquet et al. 2000, Sagnol et al. 2014b, MMRG unpublished data), and the contribution of different prey types to the whales’ diet varies seasonally (Gaskin & Cawthorn 1967, Chapter 3). It is therefore likely that habitat preferences differ between summer and winter.
Species-distribution models (SDMs) are useful tools for quantifying the habitat preferences of a species in its environment (Guisan & Zimmermann 2000), and are widely used in ecological studies of cetaceans (Redfern et al. 2006, Pirotta et al. 2011, Forney et al. 2012).
SDMs are used to relate the spatial distribution of individuals to the characteristics of their habitat, quantifying the relative importance of measured environmental variables in determining the habitat requirements of the species. SDMs are most effective when they incorporate absence as well as presence data, providing background information on the range of available environmental conditions (Brotons et al. 2004, Barbet-Massin et al. 2012). Chapter 4 – Habitat preferences 85
Evaluating the habitat preferences of a population is also useful for assessing the risk of overlap with anthropogenic activities (Torres et al. 2013), and guiding the development of management strategies (Embling et al. 2010, Becker et al. 2012). A Marine Protected Area
(MPA) was established in October 2014 over the Kaikōura Canyon, including the ‘no-take’
Hikurangi Marine Reserve, but the effectiveness of the reserve in protecting sperm whale habitat is unknown. In the context of the decline in abundance of sperm whales in the area, it is important to understand the value of the MPA for the conservation of the population.
Sperm whales off Kaikōura spend most of their time foraging (Douglas et al. 2005). The absence of females and calves implies that predation risk, shelter requirements and mating opportunities are not factors shaping the whales’ spatial distribution. In addition, the loud echolocation sounds produced in search of prey (Goold & Jones 1995) facilitate easy identification of foraging behaviour. Environmental conditions can thus be related specifically to individuals that are known to be foraging; as their distribution is likely to reflect the most suitable foraging habitat.
In this study, I investigated the environmental factors that drive the fine-scale foraging distribution of sperm whales around the Kaikōura Canyon. Specific questions included:
1) What oceanographic and topographic factors correlate with the presence of foraging
whales?
2) How do habitat preferences vary seasonally?
3) What specific areas provide the most suitable foraging habitat?
4) What is the value of the Marine Reserve in protecting sperm whale habitat?
To address these questions, species-habitat surveys were conducted at Kaikōura over three years during summer and winter. Locations for the presence and absence of foraging whales were recorded, in conjunction with water-column oceanographic data and seafloor topography. Presence-absence SDMs were used to identify the characteristics of the whales’ preferred foraging habitat at a fine spatial scale, and maps of habitat suitability were created to assess overlap with the Hikurangi Marine Reserve. With this study I aimed to better understand what physical and biological processes contribute to attracting top predators to submarine canyons in general, and sperm whales to the Kaikōura Canyon in particular.
Chapter 4 – Habitat preferences 86
4.2. Methods
4.2.1. Study site
The study site was an area of 850 km2, to the south and east of the Kaikōura Peninsula (South
Island, New Zealand), centred over the Kaikōura Canyon (Fig. 4.1). This area was limited to
12 nautical miles (nmi) offshore and 15 nmi south of the peninsula due to navigation and safety restrictions. The research area covers depths of 50 – 1550 m, including parts of the continental shelf, continental slope, the Kaikōura Canyon, Conway Trough and Conway
Ridge.
Figure 4.1. Study site around the Kaikōura Canyon, New Zealand. The research area for sperm whale surveys is indicated by the dashed red line. Survey blocks are indicated by the white grid. The site for boat launching is indicated by the red star. Depth contours show 250, 500, 1000, 1500 and 2000 m isobaths. Chapter 4 – Habitat preferences 87
The oceanographic regime offshore of Kaikōura involves a complex pattern of currents and water masses, although little is known about the hydrography in the canyon. The East Cape
Current transports warm, high-salinity Subtropical surface water (STW) from the northeast, reaching the region offshore of Kaikōura through various eddies (Heath et al. 1975,
Roemmich & Sutton 1998, Chiswell et al. 2015; see Fig. 5.1 for detail). To the south of
Kaikōura lies the highly productive Subtropical Front, where warm STW converges with the colder, less saline Subantarctic surface water (SAW) (Heath 1981, Hamilton 2006). The
Southland Current, on the western side of the Subtropical Front, is a north-eastward flow of
SAW mixed with STW (Sutton 20013). An extension of the Southland Current is the main coastal current flowing past Kaikōura (Heath 1972a, b, 1985, Chiswell & Schiel 2001). The proportion of SAW vs STW of the current as it reaches Kaikōura is unknown, but the dominant water mass in the broad Kaikōura region is the warmer and more saline STW
(Chiswell et al. 2015). In addition, coastally derived low-salinity water feeds into the inshore neritic water mass (NW).
4.2.2. Survey design
Systematic acoustic-visual surveys were carried out during spring/summer (November –
January, hereafter ‘summer’) and autumn/winter (May – July, hereafter ‘winter’), over a three-year period between May 2015 and July 2017. Surveys were carried out from a 6 m research boat, RV Grampus, powered by a 115-hp four-stroke outboard engine. A survey protocol was designed to standardise search effort and cover the research area as uniformly as permitted by the prevailing weather conditions. The study area was divided into 4 x 4 nmi blocks (Fig. 4.1), and a decision in which block to start a survey was made based on weather conditions and previous effort. The start point within a block was a randomly generated location (termed ‘search station’) which was different for every survey. Locations of search stations were restricted to waters deeper than the continental shelf and Conway Rise (> 250 m) because shallow topography can limit the long-range acoustic detection of sperm whales.
Research was conducted during daylight hours, in sea states ≤ Beaufort 3 and maximum swell heights ≤ 2 m.
Each survey started with a 15 min acoustic-visual search for sperm whales, with the research boat stationary and the engine switched off. A custom-built directional hydrophone Chapter 4 – Habitat preferences 88
(Dawson 1990) was used to listen for sperm whale echolocation clicks, which could typically be heard within 4-5 nmi (MMRG, unpublished data). At the same time, one or two observers conducted a visual search for whales at the surface (typical detection range c. 2 nmi, pers. obs.). A 15 min search duration was chosen to account for the time between foraging dives in which sperm whales are typically silent (on average 14.1 min, of which an average of 9.1 min are spent at the surface; Jaquet et al. 2000, Douglas et al. 2005), and therefore not detected acoustically. Following a standardised protocol, the closest whale to the search station
(estimated by relative loudness and directionality of the clicks, or by visual detection) was tracked first, followed by any other whales that could be heard or seen. Once these whales had been tracked, or if no whales were detected, a new search station was selected to continue the survey, excluding blocks which had been surveyed that day or which were adjacent to any blocks in which whales had been encountered. Sperm whales were tracked acoustically with the directional hydrophone, until they were visually located at the surface.
After a whale had surfaced, the research vessel was manoeuvred 50-100 m behind it, maintaining this distance throughout the encounter (i.e., while the whale remained at the surface) by matching the whale’s swimming speed and direction.
4.2.3. Data collection
Data collected at each search station included:
- Time and location at the start of the search.
- Number of whales seen or heard.
- Sea conditions (swell height and wind speed on the Beaufort scale).
- In situ water-column oceanographic data (temperature, salinity and fluorescence) to
characterise potential whale-absence locations (see below). N.B. if any whales were
estimated to be within one nmi of the search station, no oceanographic data were
collected, as this location would not be used in further analyses to characterise whale-
absence locations (see below).
Data collected during a sperm whale encounter included:
- Photographs of the flukes for individual identification, taken at the time of diving.
- Time and location of dive. Chapter 4 – Habitat preferences 89
- Detection of ‘usual echolocation clicks’ (Goold & Jones 1995) from the target whale,
to identify foraging behaviour (Møhl et al. 2000, Jaquet et al. 2001, Miller et al. 2013b).
These clicks are loud, impulsive, broadband sounds, typically made at rates of 1-2
clicks/s (Goold & Jones 1995). This approach allowed the exclusion of resting or
travelling individuals from further analysis.
- In situ water-column oceanographic data to characterise whale-presence habitat. If a
whale was within 1 nmi of where a CTD cast had been previously made (i.e., for a
different whale close by), the data from that cast were used for both whales.
Search effort and encounter data were recorded and stored on a HP-200LX palmtop computer running custom written software interfaced with a GPS, or on a Samsung Galaxy
Tab A tablet running Cybertracker software. The position of the research vessel was logged every 60 seconds.
Oceanographic variables measured included temperature, salinity and fluorescence (as a proxy for chlorophyll a concentration, hereafter chl-a), sampled at a rate of 4Hz. Data were collected using a Seabird Scientific SBE-19 CTD with a WETStar fluorometer during winter
2015, and an RBR-concerto CTD with a Cyclops-7 fluorometer (Turner Designs) during the remaining five field-seasons. Because the CTDs were hauled manually, casts made with the
SBE-19 were dropped to a depth of only 100 m. Casts made with the much smaller RBR- concerto were dropped to a maximum depth of 550 m (fluorometer depth limit = 600 m), or within 100 m of the seafloor.
4.2.4. Comparability of oceanographic data among seasons
As different CTDs were used during the study, it was necessary to quantify the potential differences in temperature, salinity and chl-a measured by the two instruments. Ten casts were done simultaneously with the SBE-19 and the RBR-concerto to a depth of 100 m in
October 2015. While there was no measurable offset in temperature or salinity, there was a small but relatively consistent offset in chl-a (Appendix 2, Fig. A2.1). A correction equation was thus established to transform fluorescence values measured by the SBE-19 in winter 2015
(Chl-a CORRECTED = 1.197 Chl-a MEASURED + 0.108; R2 = 0.92).
Chl-a concentration is a proxy for phytoplankton abundance, and is therefore related to primary productivity – the rate of carbon produced by phytoplankton in a given area and Chapter 4 – Habitat preferences 90
time (Kaiser et al. 2005). However, primary productivity is also influenced by other processes such as grazing, nutrients, light, and stratification. In addition, the fluorometers used in this study were factory-calibrated but not ground-truthed for local conditions. Therefore, chl-a was used as a relative (rather than absolute) measure of primary productivity (e.g., Hazen et al. 2009). To maximise comparability of fluorescence among casts and account for drift in the fluorescence sensor over time, individual casts made with the RBR-concerto were corrected for the deep fluorescence offset (Guinet et al. 2012, Xing et al. 2012). The offset was detected in each profile as the median fluorescence value in deep waters (depth > 300 m), where chl- a should be null. The offset (median = 0.177 μg/l, n = 443) was then subtracted from all the values in the profile.
Monthly means of chl-a obtained in-situ at 1 m depth were compared to monthly means of surface chl-a from Aqua-MODIS satellite data (NASA 2018) for the period of study over the study area (8-day composites, 4 x 4 km2 resolution). Although satellite-derived chl-a data has its own limitations (Gons 1999, Dall’Olmo et al. 2005), this comparison allowed me to verify that there were no issues with instrument drift which could bias comparisons among field seasons. The comparison between the two methods (Appendix 2, Fig. A2.2) indicated that the chl-a values measured in-situ were on the same scale as the satellite data, and that the observed variability among field seasons was not an artefact of instrument drift.
4.2.5. Presence and pseudo-absence locations
Presence and absence information were incorporated in the SDMs in the form of point data
(Redfern et al. 2006). Diving locations of foraging whales were used for presence data. Only the first location per day for each individual was used in the analysis to minimise autocorrelation.
True absences of mobile species are very hard to confirm, as an animal may be absent from an area at the time of sampling but present at another time. I therefore generated ‘pseudo- absence’ points (Zaniweski et al. 2002) within the surveyed areas where whales were not detected at the time of the surveys (e.g., Torres et al. 2008, Rayment et al. 2014; Fig. 4.2). By carrying out multiple surveys, areas where whales rarely forage have higher probabilities of containing more pseudo-absences. The method for generating pseudo-absences involved three steps. Chapter 4 – Habitat preferences 91
Figure 4.2. Extraction of absence data for SDMs of sperm whales foraging off Kaikōura. A pseudo- absence area (in grey) was generated where foraging sperm whales were assumed absent, and pseudo-absence locations were randomly generated within that area. Three typical scenarios are shown, in which different sizes of pseudo-absence areas were created around a search station, depending on the distance to the first whale tracked from the search station.
First, each search station was plotted in ArcGIS 10.2 (ESRI), and a buffer corresponding to the median acoustic detection distance (2.5 nmi4), or the distance to the first whale encountered (whichever was smallest), was constructed around each station to define a search area (Fig. 4.2). Second, a ‘presence buffer’ was constructed around the diving location of the closest whale (if any), and deleted from the search area (Fig. 4.2). This was done to prevent generating a pseudo-absence point where the whale had been, and thus minimise the chances of absence points occupying the same habitat as actual sightings (Torres et al.
2008, Tepsich et al. 2014). The buffer around the whale was based on the whale’s potential movement during the time it was being tracked. This buffer distance was calculated as the time taken to track the whale multiplied by a conservative measure of the typical horizontal displacement speed between consecutive dives (75th percentile = 1.5 nmi/h, based on n = 117).
4 The median acoustic detection distance (2.5 nmi) was based on 173 measurements of the distance between a search station and the diving location of the closest whale tracked from that station, recorded throughout the three years of study. Chapter 4 – Habitat preferences 92
The maximum value of the buffer was the distance from the whale to the search station. The remaining search area was defined as an ‘absence area’ where sperm whales would have been likely to be heard if they had been foraging there. Lastly, pseudo-absence locations were generated randomly within each absence area using the ‘genstratrandpnts’ tool in Geospatial
Modelling Environment 0.7.3 (Beyer 2012).
SDMs that use regression techniques are most accurate when presence and absence data have the same weight and there are at least as many absences as presence points (Barbet-
Massin et al. 2012). This was achieved differently for each season. For the summer models, one pseudo-absence was generated per absence area, resulting in a slightly higher number of pseudo-absences than presences (overall, a ratio of 1.2:1). Therefore, each pseudo-absence location was down-weighted so that the sum of the weights of pseudo-absences was equal to the sum of the weights of presences (e.g., Torres et al. 2013). In winter, sperm whale densities were relatively high, resulting in fewer absence areas than presence locations
(overall, a ratio of 0.3:1). Up to five pseudo-absence locations were generated within each absence area to obtain an equal number of presences and absences per winter. To distribute pseudo-absence locations evenly, the number of locations per absence area was proportional to the size of the area. The minimum distance between pseudo-absence locations was set to
1.8 nmi (the median minimum distance per day between two whales; n = 163 days), so that the density of pseudo-absences was no greater than the presence locations. Since the number of pseudo-absences and presences per winter was the same, there was no need to weight them in the SDMs.
4.2.6. Environmental variables
Previous studies have suggested that habitat selection by sperm whales is influenced by seafloor depth, seafloor slope, surface chlorophyll, surface oceanographic conditions, temperature of deep-water layers, and vertical temperature gradients (Griffin 1999, Rendell et al. 2004, Praca et al. 2009, Torres et al. 2011, Pirotta et al. 2011, Wong & Whitehead 2014,
Johnson et al. 2016). In addition, sub-surface chlorophyll concentration, thermocline characteristics (depth and strength), and salinity of deep-water layers appear to influence habitat preferences of other deep-diving marine mammals, such as beaked whales and elephant seals (Ferguson et al. 2006a, Hazen et al. 2011, Saijo et al. 2017). These habitat Chapter 4 – Habitat preferences 93
features (outlined in Table 4.1) were therefore selected as candidate factors in the models to predict foraging distribution of sperm whales.
Table 4.1. Candidate explanatory variables for the SDMs used to investigate the habitat preferences of sperm whales foraging off Kaikōura.
Habitat characteristic Candidate explanatory variables Abbreviation (units)
seafloor topography depth depth (m) slope slope (%) aspect aspect (°)
temperature of surface water surface temperature (1 m depth) surface T (°) temperature at 10 m depth T 10m (°)
temperature of deep water temperature at 350 m depth T 350m (°)
salinity of surface water surface salinity (1 m depth) surface S (psu) salinity at 10 m depth S 10m (psu)
salinity of deep water salinity at 350 m depth S 350m (psu)
thermal stratification thermocline depth (1) TC depth (m) thermocline strength (1) TC strength (°) temperature gradient between 10 and 350 m (2) T gradient 350m (°)
chlorophyll concentration surface chl-a (1 m depth) surface chl-a (μg/L) sub-surface chl-a maximum (3) max chl-a (μg/L) depth of sub-surface chl-a maximum depth max chl-a (m)
Notes on calculation of variables:
(1) Using temperature data averaged in 20 m bins, the thermocline depth was estimated as the depth with the greatest variance within the bin. The relative strength of the thermocline was expressed as the variance in temperature values in the bin with maximum variance (Hazen & Johnston 2010). (2) ∆ 푇 = 푇10 푚 − 푇350 푚. (3) The sub-surface chl-a maximum was calculated from chl-a data averaged in 5 m bins. The bin with the maximum chl-a average was selected as the sub-surface chl-a maximum (Scott et al. 2010). A bin of 5 m (rather than 20 m) was selected to reflect the higher rate of change in chl-a with depth (Segar 2007).
Chapter 4 – Habitat preferences 94
Bathymetry data at 250 m resolution were obtained from NIWA (Mitchell et al. 2012) and imported into ArcGIS. Seafloor depth, slope and aspect for each presence and pseudo- absence location were extracted using Spatial Analyst tools. Slope was defined as the rate of maximum change in depth in a raster cell, expressed as percent slope. Aspect was defined as the orientation of the downward slope, expressed as 0 - 360° with respect to true north.
Oceanographic data collected with the SBE-19 CTD were processed with the programs SBE
Seaterm and SBE Data Processing (Sea-Bird Scientific). Data collected with the RBR-concerto
CTD were processed with Ruskin (RBR). Each CTD downcast was averaged in 1m bins
(typically 4-5 samples) to reduce the effect of instrument noise on the data. Estimates of the oceanographic variables were extracted from the bin-averaged CTD profiles in Matlab 2014a
(Math-Works). Temperature and salinity at 10 m were considered because surface values can be strongly influenced by daily weather changes, whereas sub-surface values (i.e., at 10 m) are less susceptible while still being useful descriptors of surface layers (de Boyer Montégut et al. 2004). Many of the casts carried out with the RBR-concerto did not reach a depth of 500 m (n = 302 out of 447), due to boat drift or because of safety precautions to avoid hitting the seafloor. To use deep-water hydrographic variables from a consistent depth while maximising the number of usable casts, temperature and salinity values from a depth of 350 m were used for further analysis (n = 435 casts).
Oceanographic data collected at a search station were used to characterise the pseudo- absence locations generated around that station. This approach required the assumption that oceanographic conditions at the pseudo-absence locations were similar to the conditions at the station where the CTD data had been collected (i.e., within a distance of 0 – 2.5 nmi). The validity of this assumption was tested by plotting the distance between CTD casts made in the same day (range = 1 – 17 nmi) against the difference in each oceanographic variable.
These points were fitted with a smoothed spline, with 95% confidence bounds. If the difference in each variable at 2.5 nmi was within the 95% confidence bound of the difference at 1 nmi, it was considered small enough to validate the assumption for the purpose of this study. This was the case for all variables (Appendix 2, Fig. A2.3).
Sea state and swell height can affect the probability of encountering whales by reducing the visual or acoustic detection range (Barlow & Taylor 2005, Ferguson et al. 2006b, Rayment et Chapter 4 – Habitat preferences 95
al. 2018). The distance to the closest whale from each search station was examined in relation to sea state and swell height. Over the ranges observed in this study, the variability in each factor did not affect the detection range (Appendix 2, Fig. A2.4), so it was unnecessary to account for them in the SDMs.
4.2.7. Species-distribution models
There are multiple techniques that can be used to estimate species-habitat relationships (e.g., boosted regression trees, random forest, multivariate adaptive regression splines; Elith et al.
2006, Redfern et al. 2006). I selected a Generalized Additive Model framework (GAM; Hastie
& Tibshirani 1990, Wood 2006), as this approach can be applied to binary presence/absence data, does not require very extensive datasets, and is particularly useful for capturing non- linear cetacean-habitat relationships (Derville et al. 2018; e.g., Pirotta et al. 2011, Forney et al.
2012, Rayment et al. 2014). Specifically, a binomial GAM with a logit link function was used to model the relationship between the probability of whale presence and the environmental variables at each location. To investigate seasonal differences in habitat use by sperm whales,
GAMs were created independently for summer and winter. Due to not having oceanographic data deeper than 100 m in winter 2015, two data sets were used for modelling habitat preferences in winter: (1) using data from 2016 and 2017 and including deep-water variables (i.e., temperature and salinity at 350 m, vertical temperature gradient to 350 m), and (2) using data from 2015, 2016 and 2017 but excluding deep-water variables. The results of the analysis using the first data set suggested that none of the deep-water variables were important predictors of sperm whale habitat preferences during winter. For simplicity, the results from the first data set are only presented in the appendix (Appendix 2, Table A2.1,
Fig. A2.5).
In winter, the pseudo-absence locations generated in the same absence area in a given day had different topography but shared the same oceanographic characteristics (as described above). While the use of data points that are close in space can provide useful information about fine-scale distribution, it can be a source of spatial autocorrelation. In this circumstance, the assumption of independence required for the GAM is not met, leading to underestimation of the uncertainty associated with model estimates (Zuur et al. 2009). One way to deal with this issue is to explicitly account for autocorrelation among data points Chapter 4 – Habitat preferences 96
located in the same area of sampling (or subsite), by incorporating subsite as a ‘random effect’ in the models (Aarts et al. 2008, Zuur et al. 2009, Rhodes et al. 2009). This can be done with a Generalized Additive Mixed Model (GAMM; Lin & Zhang 1999), a variant of a GAM that includes random effects in addition to fixed effects. In winter, each data point was therefore assigned a subsite ID according to the survey block that contained its location.
There was no need to use this approach in the summer models, as there was only one pseudo-absence generated per absence area, and therefore each point had unique CTD data.
Correlation among explanatory habitat variables was assessed with concurvity tests
(Ramsay et al. 2003, Wood 2006), the GAM analogue of collinearity tests. A threshold of 0.3 was used to identify concurvity among pairs of covariates (He et al. 2006). For correlated pairs, each habitat variable was fitted independently to a univariate GAM with presence or absence of whales as the response, and the covariate with highest explained variance was retained in the full model. Multi-collinearity among all the explanatory variables included in the full model was then checked with variance inflation factors (VIF), with values <2 indicating no collinearity (Neter et al. 1990, Montgomery & Peck 1992, Zuur et al. 2009). Once correlated terms had been excluded from the full model, probability of whale presence was modelled with a suite of GAM(M)s which included all first-order combinations of independent predictor variables, and, in the case of winter GAMMs, with subsite as the random effect. No interaction terms were included in the models because there was no biological reason to do so, and because interaction terms can confound the interpretation of fitted functions (Yee & Mitchell 1991, Rayment et al. 2014).
All predictor variables were modelled as smoothed terms, derived using thin-plate regression splines, except ‘aspect’, which was modelled with a cyclic regression spline. All smooths were limited to a maximum of four degrees of freedom to reduce the risk of overfitting (Vaughan & Ormerod 2005). Exploratory trials with higher degrees of freedom
(up to nine) did not result in changes to the smoothers, indicating that a maximum of four degrees of freedom was not restricting model optimisation. Models were created in R studio
1.1.4 (R development Core Team 2012) using the packages ‘mgcv’, ‘gamm4’ and ‘MuMIn’
(Bartoń 2016, Wood 2016, Wood & Scheipl 2017). Models were ranked according to Akaike’s
Information Criterion (AICC; Akaike 1973), with the best models indicated by the lowest
AICC values (Burnham & Anderson 2002, Symonds & Moussali 2011). Smoothing curves Chapter 4 – Habitat preferences 97
derived from the best GAMs in each set were used to interpret the effect of each explanatory variable on habitat selection. The relative importance of each predictor variable was examined by summing the Akaike weights for each model that included the predictor variable (Burnham & Anderson 2002, Twist et al. 2016).
4.2.8. Model checking
Spatial autocorrelation was examined by plotting correlograms of the residuals of the best fitting models (Bjørnstad & Falck 2001, Zuur et al. 2009), using the package ‘ncf’ in R
(Bjørnstad 2008). Correlograms are graphical representations of the spatial correlation between locations at a range of lag distances, depicted as a spline with 95% pointwise bootstrap confidence intervals. Spatial autocorrelation was not a concern in the final models for summer or winter (Appendix 2, Fig. A2.6).
Diagnostic plots of binomial regression models are difficult to interpret due to the discreteness of the binary response data (Landwehr et al. 1984, Zuur et al. 2009). The adequacy of the best models was therefore examined using a simulation technique which constructs a logistic regression quantile-quantile plot (Landwehr et al. 1984, Zuur et al. 2009; e.g., Rayment et al. 2018). With this method, the residuals from the fitted model are plotted against the residuals from a model based on a simulated distribution, which assumes that the model is correct. Confidence intervals are calculated from the 2.5 and 97.5% quantiles of the simulated data. If the model fits the data well, the plotted points lie close to a 1:1 line and within the simulated 95% confidence intervals. This was the case for the summer and winter models (Appendix 2, Fig. A2.7).
4.2.9. Model validation
Model validation using independent data is key for assessing the robustness of cetacean- habitat models (Forney 2000, Redfern et al. 2006). Model performance was evaluated through multi-fold validation (Forney et al. 2012, Rayment et al. 2014), carried out separately for summer and winter using the highest-ranking model for each season. First, each of the three possible pairs of years (2015 & 2016, 2016 & 2017, 2015 & 2017) was used as a training dataset to fit the predictive GAM(M)s, with the remaining year used as an independent dataset to validate the model. For each validation test, every location (sperm whale presences and pseudo-absences) was assigned a predicted value of whale presence probability according Chapter 4 – Habitat preferences 98
to the fitted functions from the best GAM(M) and the values of the habitat variables measured at that location. Predictive accuracy was then assessed using two indices. (1) The
‘true skill statistic’ (TSS, Allouche et al. 2006) measures model performance in a simple way, and is calculated as:
푇푆푆 = 푠푒푛푠푖푡푖푣푖푡푦 + 푠푝푒푐푖푓푖푐푖푡푦 − 1 where sensitivity and specificity are the proportion of correctly identified presences and absences, respectively. For validation purposes, the probability threshold was set at 0.5 (with values ≥ 0.5 predicted as presences and values < 0.5 as absences). This threshold value is suitable for cases in which the training data contain equal numbers of presences and absences (as in the winter dataset) or the total weight of the presences and absences is the same (as in the summer dataset; Lobo et al. 2008, Barbet-Massin et al. 2012). TSS ranges from
-1 to +1, with values ≤ 0 indicating poor performance (no better than random). (2) The area under the receiver operating characteristic curve, known as AUC, was calculated using the
R package ‘ROCR’ (Sing et al. 2015). AUC estimates range from 0.5 (no predictive power) to
1 (perfect prediction). AUC is a widely used threshold-independent method (Elith et al. 2006,
Derville et al. 2018), but it has been criticised for producing unreliable metrics of model performance (Manel et al. 2001, Lobo et al. 2007). I therefore used specificity and sensitivity as the main way to evaluate the models, along with AUC for comparative purposes. Model validation is also useful to check for potential overfitting, with high values of TSS and AUC indicating a robust relationship between species presence and the selected predictor variables (Pirotta et al. 2011).
There is a possibility that model validation was biased by using the best GAM(M) from the full data set in all the validation tests, rather than the best GAM(M) from each set of training data (e.g., Forney 2000, Rayment et al. 2014, Gilles et al. 2016). To ensure that this was not the case, model validation tests were repeated using the best GAM(M)s from each set of training data (i.e., three model selection exercises per season, based on data from each pair of years) and tested on the data from the remaining year. Although there were some slight variations in the best model selected for each data set, the variables with high scores of relative importance were consistently selected, and the resulting TSS and AUC values were similar to those obtained using the same best GAM(M) across all validation tests. For Chapter 4 – Habitat preferences 99
simplicity, the results from this additional analysis are only summarised in the appendix
(Appendix 2, Table A2.2).
Maps of habitat suitability were created for each summer and winter, based on the probabilities of sperm whale presence modelled by the best GAM(M). The best model from each training data set was used to predict the presence probability at each location of the remaining year (i.e., each presence and pseudo-absence). Maps were created via spatial interpolation using inverse distance weighting to the first power (Ferguson et al. 2006a, b) using the ‘IDW’ tool in ArcGIS. Lastly, the Hikurangi marine reserve was overlaid on the habitat suitability maps to assess the extent of protection of sperm whale habitat.
4.3. Results
4.3.1. Survey effort
Sperm whale surveys were conducted over 169 days from May 2015 to January 2017, resulting in 899 h of effort (covering 6915 km), 287 search stations, and 643 sperm whale encounters (Table 4.2; Appendix 2, Fig. A2.8). The number of individual whales sighted per day ranged between 0 and 6 in summer (median = 2), and between 1 and 9 in winter (median
= 4). Whales were foraging in the vast majority of encounters (95%), as identified by the detection of foraging clicks immediately before or after a surface interval. After filtering out repeated encounters with the same individual in a day, whales that were not foraging, and encounters for which a CTD cast was not available, a total of 334 presences were available for the SDMs. All sightings were of solitary males, except for six encounters with groups of two males. All these encounters were treated as a single presence for the purposes of the
SDMs. A total of 360 pseudo-absence locations were generated in the areas surrounding the search stations within which no sperm whales were detected.
A total of 513 CTD casts were carried out. Chl-a data could not be collected during eight days in winter 2017 due to malfunction of the sensor (27 CTD casts in total). Presence and absence data collected during these days were excluded from the SDMs.
Chapter 4 – Habitat preferences 100
Table 4.2. Summary of survey effort and sperm whale presences and pseudo-absences. ‘pres.’ = presence, ‘abs.’ = absence. *During summer 2016/17, data collection was discontinued over the period of 14th Nov – 9th Dec due to an earthquake affecting Kaikōura.
Research Effort Effort Effort # # whale # pseudo- Ratio Season period (days) (hrs) (km) CTD pres. abs. abs:pres
Winter 2015 28 May – 30 Jul 26 149 1205 66 48 48 1.00
Winter 2016 24 May – 31 Jul 24 147 968 72 83 83 1.00
Winter 2017 23 May – 18 Jul 27 155 980 63 67 67 1.00
Summer 2015/16 20 Oct – 30 Dec 37 145 1068 98 42 57 1.36
Summer 2016/17 * 31 Oct – 18 Jan 23 102 797 65 25 39 1.56
Summer 2017/18 2 Nov – 5 Jan 32 201 1896 122 69 66 0.96
All winters 77 451 3153 201 198 198 1.00
All summers 92 448 3761 285 136 162 1.19
Total 169 899 6915 486 334 360 1.08
4.3.2. Species-distribution models
After excluding correlated terms among candidate explanatory variables (Appendix 2, Table
A2.3), a total of nine and seven variables were included in the full models for summer and winter, respectively. VIFs among the resulting covariates were <2, indicating no issues with multi-collinearity.
For summer, there was clear support for models that featured depth, slope, vertical temperature gradient (which was highly correlated with thermocline strength), and surface salinity as explanatory variables of sperm whale distribution (Table 4.3a, Fig. 4.3a). There was some uncertainty around which was the best model in the candidate set, with the highest-ranking model also including the variables salinity at 350m and surface chl-a, but closely followed by the two models that excluded each of these variables. The exclusion of salinity at 350m or surface chl-a reduced the explained deviance by only 1%, indicating a small contribution to explaining the total variability in sperm whale presence. There was very marginal support for the inclusion of thermocline depth or maximum chl-a, and no support for the inclusion of seafloor aspect (Table 4.3a). The highest-ranking models explained 27-28% of the variance. The presence of foraging whales in summer was correlated Chapter 4 – Habitat preferences 101
with depths of 900 – 1300 m, relatively steep slopes (> 15%), strong vertical temperature gradients (> 4°C across the top 350 m), and surface salinities of 34.6 – 34.7 psu (typical of
Subtropical Water at Kaikōura; Fig. 4.4). There was some evidence for a correlation of sperm whale presence with deep-water salinities > 34.65 psu (more saline Subtropical Water) and low surface chl-a (< 0.75 μg/l; Fig. 4.4), although the statistical significance of these correlations was low, especially in the case of surface chl-a (Table 4.4a).
For winter, there was some uncertainty around model selection, but less so than for summer.
The best model in the set had a weight of 39%, twice the support than for the second-best model. The best model explained 30% of the variance, and featured depth, aspect, maximum chl-a and depth of the chl-a maximum as explanatory variables of sperm whale distribution
(Table 4.3b). There was marginal support for the inclusion of temperature in the surface layer
(Table 4.3b), with low statistical significance for its correlation with whale distribution (Table
4.4b). There was no support for inclusion of slope or surface salinity (Table 4.3b). In winter, whales foraged more often over depths of 500 – 800 m, north/northeast orientated slopes (0
– 65°), and habitat where the sub-surface chl-a maximum was relatively low (0.5 – 0.9 μg/l) and at depths of 20 – 50 m (Fig. 4.5).
(a) summer (b) winter 100% 100%
75% 75%
50% 50%
25% 25%
Relative Relative importanceindex 0% 0%
Figure 4.3. Relative importance indices of habitat variables included in the models to explain presence of sperm whales foraging off Kaikōura. The index reflects the summed model probabilities of all the models containing a particular habitat variable. Chapter 4 – Habitat preferences 102
Table 4.3. Model selection of (a) summer GAMs and (b) winter GAMMs used to explain presence of sperm whales. Model formulae and metrics of model performance are shown for models with Akaike weights ≥5%. SW = probability of sperm whale presence/absence; T = temperature; S = salinity; TC = thermocline; 2 edf = estimated degrees of freedom; ∆AICC = difference in AICC score relative to best model in set; wi = Akaike weight (i.e., model probability); Adjusted R = explained variance; (1|subsite) = random effect included in all winter GAMMs. Note that the proportion of explained deviance cannot be computed for GAMMs.
(a) Summer:
Adjusted Deviance Rank Model edf ∆AICC wi R2 explained (%)
1 SW ~ depth + slope + T gradient 350m + surface S + S 350m + surface chl-a 12 0.00 0.21 0.28 26.2
2 SW ~ depth + slope + T gradient 350m + surface S + S 350m 11 0.82 0.14 0.27 25.5
3 SW ~ depth + slope + T gradient 350m + surface S + surface chl-a 11 1.11 0.12 0.27 25.3
4 SW ~ depth + slope + T gradient 350m + surface S + S 350m + surface chl-a + max chl-a 13 2.11 0.07 0.27 26.2
5 SW ~ depth + slope + T gradient 350m + surface S + S 350m + surface chl-a + TC depth 13 2.31 0.07 0.27 26.2
6 SW ~ depth + slope + T gradient 350m + surface S 10 2.33 0.07 0.26 24.6
(b) Winter:
Adjusted Rank Model edf ∆AICC wi R2
1 SW ~ depth + aspect + max chl-a + depth max chl-a + (1|subsite) 9 0.00 0.39 0.30
2 SW ~ depth + max chl-a + depth max chl-a + (1|subsite) 8 1.49 0.19 0.25
3 SW ~ depth + aspect + max chl-a + depth max chl-a + T 10m + (1|subsite) 11 2.12 0.14 0.30
4 SW ~ depth + max chl-a + depth max chl-a + T 10m + (1|subsite) 10 3.75 0.06 0.26 Chapter 4 – Habitat preferences 103
Table 4.4. Smooth terms from (a) summer GAMs and (b) winter GAMMs used to explain presence of sperm whales. Estimated degrees of freedom (edf) and significance (p-value, in brackets) of smooth terms retained in each model are shown for models with Akaike weights ≥5%. The formula of each model (according to its rank) is shown in table 4.3. T = temperature; S = salinity; TC = thermocline.
(a) Summer:
Model depth slope T gradient surface S S 350m surface chl-a max chl-a TC depth rank
3.69 1.00 1.94 2.57 1.00 1.00 1 - - (p<0.001) (p=0.004) (p<0.001) (p=0.03) (p=0.05) (p=0.15)
3.71 1.00 1.92 2.48 1.00 2 - - - (p<0.001) (p=0.002) (p<0.001) (p=0.03) (p=0.04)
3.70 1.00 2.10 2.57 1.00 3 - - - (p<0.001) (p=0.004) (p=0.003) (p=0.03) (p=0.12)
3.69 1.00 1.93 2.56 1.00 1.00 1.00 4 - (p<0.001) (p=0.004) (p<0.001) (p=0.03) (p=0.05) (p=0.14) (p=0.76)
3.69 1.00 1.95 2.57 1.00 1.00 1.00 5 - (p<0.001) (p=0.004) (p<0.001) (p=0.03) (p=0.05) (p=0.15) (p=0.90)
3.72 1.00 2.07 2.50 6 - - - - (p<0.001) (p=0.002) (p<0.001) (p=0.03)
(b) Winter:
Model depth max depth aspect max chl-a T 10m rank chl-a
3.68 1.84 3.67 3.40 1 - (p=0.006) (p=0.01) (p<0.001) (p<0.001)
3.67 3.71 3.40 2 - - (p=0.006) (p<0.001) (p<0.001)
3.71 1.84 3.64 3.42 1.00 3 (p=0.006) (p=0.01) (p=0.001) (p<0.001) (p=0.13)
3.68 3.68 3.39 1.00 4 - (p=0.007) (p<0.001) (p<0.001) (p=0.14) Chapter 4 – Habitat preferences 104
Figure 4.4. Effect of habitat variables on sperm whale distribution during SUMMER, estimated with the highest-ranking generalised additive model. (a) Seafloor depth, (b) slope, (c) vertical temperature gradient between the surface and 350 m, (d) surface salinity, (e) salinity at 350 m, (f) surface chl-a. The y-axes show the smooth function of each variable, with the estimated degrees of freedom in brackets. Shaded areas represent 95% confidence intervals of the response. Rug plots (vertical black lines above the x-axis) represent the x-value of each data point. Chapter 4 – Habitat preferences 105
Figure 4.5. Effect of habitat variables on sperm whale distribution during WINTER, estimated with the highest-ranking generalised additive mixed model. (a) Seafloor depth, (b) slope aspect, (c) sub- surface chl-a maximum, (d) depth of sub-surface chl-a maximum. The y-axes show the smooth function of each variable, with the estimated degrees of freedom in brackets. Shaded areas represent 95% confidence intervals of the response. Rug plots (vertical black lines above the x-axis) represent the x-value of each data point.
Chapter 4 – Habitat preferences 106
4.3.3. Model evaluation
The highest-ranking model in each season was used to create maps of habitat suitability for each year, based on the predictions of probability of sperm whale presence. In summer, the model consistently predicted the most suitable habitat to be in the Kaikōura Canyon, especially along the slopes, and over the smaller gullies situated over the 1000 m depth contour further offshore (Fig. 4.6a-c). The summer habitat predicted to be most suitable for foraging was generally consistent with the areas of higher sperm whale densities (Fig. 4.6d).
In winter, the Conway Trough and the sides of the Conway Ridge were consistently predicted as areas of high habitat suitability (Fig. 4.7a-c). There was some inter-annual variability in the predictions at the head of the Kaikōura Canyon (with low whale probability in 2015, high in 2016, and intermediate in 2017), as well as on the northeast part of the study area (with high probability in 2015 only). Overall, the winter habitat predicted to be most suitable for foraging was generally consistent with the areas of highest sperm whale densities (Fig. 4.7d), but less so than in summer. The winter models correctly predicted a lower density of sperm whales in the eastern half of the study area (including the Kaikōura
Canyon), as illustrated by the low-to-intermediate probabilities in this region (Fig. 4.7d).
The overlap of the Hikurangi Marine Reserve with preferred sperm whale habitat strongly differed between summer and winter. In summer, a large portion of the reserve contained habitat predicted to be most suitable for foraging, although there were still large areas of high habitat suitability outside the reserve (Fig. 4.6e). In contrast, there was very little overlap with the whales’ preferred habitat in winter (Fig. 4.7e).
Chapter 4 – Habitat preferences 107
Figure 4.6. Habitat preferences of sperm whales foraging off Kaikōura during SUMMER. Values of habitat suitability are based on generalized additive model predictions of sperm whale probability, according to the habitat variables that characterise each presence or absence location of that particular year. Each year was mapped independently, using training data from (a) 2016/17 and 2017/18, applied to 2015/16, (b) 2015/16 and 2017/18, applied to 2016/17, and (c) 2015/16 and 2016/17, applied to 2017/18, with all summer predictions combined in (d) and (e). The actual sperm whale presences and pseudo- absences are shown in (d) to visualise the model validation results, and the Hikurangi Marine Reserve is overlaid in (e).
Chapter 4 – Habitat preferences 108
Figure 4.7. Habitat preferences of sperm whales foraging off Kaikōura during WINTER. Values of habitat suitability are based on generalized additive model predictions of sperm whale probability, according to the habitat variables that characterise each presence or absence location of that particular year. Each year was mapped independently, using training data from (a) 2016 and 2017, applied to 2015, (b) 2015 and 2017, applied to 2016, and (c) 2015 and 2016, applied to 2017, with all winter predictions combined in (d) and (e). The actual sperm whale presences and pseudo- absences are shown in (d) to visualise the model validation results, and the Hikurangi Marine Reserve is overlaid in (e).
Chapter 4 – Habitat preferences 109
The predictive accuracy of the best models for summer and winter were assessed via multi- fold validation (Table 4.5). In the three possible validation tests for summer, the best model correctly predicted between 67 and 80% of the presences and between 59 and 82% of the absences, resulting in a mean TSS of 0.43. The estimated AUC values ranged between 0.76 and 0.82. These results indicated consistency in foraging habitat among years and good model performance, providing assurance that it was not overfitted. The results of the validation tests were more variable in winter, with the best model correctly predicting between 50 and 83% of the presences and between 61 and 70% of the absences, resulting in a mean TSS of 0.34. AUC values ranged between 0.65 and 0.76. These results indicated that the model had useful predictive power, but with slightly worse performance than in summer. This was mostly due to the low proportion of predicted presences in winter 2015, for which year the model appears to be slightly overfitted, with a relatively low TSS value in spite of a high level of explained variance.
Table 4.5. Metrics of model performance for the (a) summer GAMs and (b) winter GAMMs used to predict habitat suitability for sperm whales foraging off Kaikōura. TSS = true skill statistic; AUC = area under the receiver operating curve; adjusted R2 = explained variance. Sensitivity = proportion of correctly predicted presences; specificity = proportion of correctly predicted pseudo-absences. Note that the proportion of explained deviance cannot be computed for GAMMs.
(a) Summer Validation Adjusted Deviance Training data Sensitivity Specificity TSS AUC data R2 explained (%)
2016/17, 2017/18 2015/16 0.67 0.82 0.49 0.82 0.25 24.9 2015/16, 2017/18 2016/17 0.80 0.64 0.44 0.77 0.29 28.6 2015/16, 2016/17 2017/18 0.75 0.59 0.34 0.76 0.32 31.0
Mean values 0.74 0.69 0.43 0.78 0.29 28.2
(b) Winter Validation Adjusted Training data Sensitivity Specificity TSS AUC data R2
2016, 2017 2015 0.50 0.69 0.19 0.65 0.41 2015, 2017 2016 0.83 0.61 0.45 0.76 0.24 2015, 2016 2017 0.67 0.70 0.37 0.71 0.33
Mean values 0.67 0.67 0.34 0.71 0.33 Chapter 4 – Habitat preferences 110
4.4. Discussion
Sperm whales in and around the Kaikōura Canyon are selective about where they forage.
Their preferred habitat is characterised by specific topographic and hydrographic features, which are likely to reflect optimal foraging owing to higher prey abundance and/or ease of capture. The data obtained over a three-year period showed that whale distribution could be partially explained by seafloor depth, slope features, thermal stratification of the water column, and low sub-surface peaks in chl-a concentration. The inclusion of water column oceanographic data in season-specific SDMs advanced our understanding of how sperm whales utilise a dynamic environment to meet their high energy requirements.
4.4.1. Foraging habitat preferences in summer
In summer, foraging occurred more often over depths of 900 – 1300 m. Seafloor depth is typically a good predictor of sperm whale presence in SDMs (Praca et al. 2009, Pirotta et al.
2011, Torres et al. 2011, Tepsich et al. 2014), although the preferred depths vary with region.
Off Kaikōura in summer, foraging occurs in mid-water (at around 500 and 850 m) and in proximity to the seafloor at 900 – 1300 m (Miller et al. 2013a, Guerra et al. 2017), consistent with the depth preferences indicated by the GAMs. The environmental drivers that aggregate prey in pelagic and demersal layers are likely to differ, but the preferred depths may reflect areas where prey are available in more than one layer. The preferred range is similar to the depths over which the benthopelagic warty squid (Onykia ingens) are thought to be most abundant (800 – 1100 m, Anderson et al. 1998, Jackson et al. 2000), consistent with the suggestion that this is one of the main prey items in the Kaikōura/Cook Strait region
(Gaskin & Cawthorn 1967).
Sperm whales also favoured habitat with steep slopes and strong vertical temperature gradients. Strongly stratified waters can concentrate nutrients, organic matter and small planktonic organisms (Franks 1992, Puskaric et al. 1992, Papiol et al. 2012). Slopes can interact with currents to aggregate zooplankton and small nekton, in addition to truncating the vertical distribution of pelagic organisms (Mauchline & Gordon 1991, Nesis 1993, Sutton et al. 2008). These habitat features can locally enhance the availability of food resources for higher trophic level consumers (Hansen et al. 2001, Azzellino et al. 2008, Sutton et al. 2008,
Hazen & Johnston 2010), potentially increasing the densities of sperm whale prey. Indeed, Chapter 4 – Habitat preferences 111
strong evidence suggests that thermally stratified waters and steep slopes provide favourable conditions for deep-water squid (Nesis 1993, Röpke et al. 1993, Quetglás et al.
2000, Vidal et al. 2010, Crespi-April et al. 2013), including warty squid, which tend to aggregate over steep slopes (Jackson et al. 2000). The association between sperm whales and steep seafloor gradients is consistent with previous studies (Gregr & Trites 2001, Torres et al.
2011, Tepsich et al. 2014, Sagnol et al. 2014b). Although this is the first study to find a direct correlation between sperm whale presence and vertical temperature gradients, previous studies have shown higher occurrences of whales in regions with typically stronger stratification (Gaskin 1973, Griffin 1999). This hydrographic feature also enhances prey availability for other deep-diving predators, such as beaked whales (Ferguson et al. 2006a,
Hazen et al. 2011) and fur seals (Nordstrom et al. 2013). The vertical temperature gradient was measured over the top 350 m of the water column, which is shallower than where sperm whales typically forage (Miller et al. 2013a, Guerra et al. 2017). The influence of thermal stratification on foraging is therefore clearly not direct, but there may be an indirect mechanism that contributes to higher prey availability. A likely candidate to mediate this link is the deep scattering layer (DSL), composed of mesopelagic organisms and situated between 200 and 600 m deep (Johnson 1948, Hays 2003). The DSL responds to vertical thermal gradients (Hazen & Johnston 2010, Fennell & Rose 2015) and is thought to contain prey for species that sperm whales feed on (Guerra et al. 2017).
While steep slopes are a permanent habitat feature, vertical temperature gradients are dynamic, changing spatially and over time according to seasonal cycles of thermal stratification and the influence of other mixing processes (Segar 2007). As SDMs respond to both temporal and spatial changes in predictor variables, the selection of important oceanographic factors might reflect, to some extent, temporal changes in the quality of foraging habitat. Abundance of sperm whales off Kaikōura is typically low at the start of spring/summer, increasing gradually as the season progresses (Appendix 2, Fig. A2.9a).
Thermal stratification strengthened throughout November and December (Appendix 2, Fig.
A2.9b, A2.10b), increasing at a similar rate to the number of sperm whales present in the area. Although no causation can be established from this concurrent trend, it is possible that the increase in stratification influences the arrival of sperm whales by facilitating prey aggregations. This progression could be driven by the thermal gradient itself or by other Chapter 4 – Habitat preferences 112
associated environmental processes. In any case, the synchronicity between the arrival of whales and the increase in thermal gradient is likely to be an expression of oceanographic changes modifying the trophic dynamics in the canyon. This process is worthy of further investigation.
Sperm whale foraging during summer also occurred more often in areas of higher salinities, where subtropical water (STW) was the dominant water mass. STW is relatively warm and saline, poor in macro-nutrients and relatively rich in iron (Heath 1981, Boyd et al. 1999,
Hamilton 2006). It is possible that STW indirectly contributes to a more favourable habitat for prey, for example through a preference for more saline waters by oceanic species.
However, a more likely explanation for the observed correlation is that whales forage most often over offshore waters, where STW is typically found, rather than closer inshore where neritic water is present.
The areas of highest habitat suitability during summer were inside the canyon and the gullies further offshore, consistent with distribution patterns observed by Jaquet et al. (2000).
Previous studies have suggested that the high productivity and habitat diversity of submarine canyons can lead to localised high densities of fish and squid (Vetter & Dayton
1998, Bosley et al. 2004, De Leo et al. 2010, Papiol et al. 2012). The results from this study suggest that, during summer, the Kaikōura Canyon provides a favourable combination of oceanographic and topographic features that facilitate prey aggregations, and/or improve their catchability, creating high quality foraging habitat for sperm whales.
4.4.2. Foraging habitat preferences in winter
In winter, whales foraged most often over seafloor depths of 500 – 800 m. A preference for shallower habitats than in summer is consistent with the seasonal variation in diving and echolocation behaviour of sperm whales at Kaikōura. In winter, whales perform shorter dives with shorter surface intervals (Jaquet et al. 2000), start foraging at shallower depths at the start of their dives (Guerra et al. 2018), and their shorter initial inter-click intervals are indicative of shallower search ranges (Isojunno et al. 2011, MMRG unpublished data). A preference for shallower depths suggests that sperm whales hunt different prey in winter and summer, with potentially smaller energetic demands during winter. Chapter 4 – Habitat preferences 113
The distribution of sperm whales during winter was correlated with north/northeast orientated slopes. Although the fundamental reasons behind this association cannot be determined, it is most likely to reflect some form of interaction between the local circulation patterns and the topography. The dominant hydrological feature off Kaikōura is the
Southland Current, flowing from the southwest to the northeast (Heath 1972a, b, 1985,
Chiswell & Schiel 2001). Elsewhere, organic matter and planktonic organisms have been shown to accumulate in the lee (downstream) of slopes (Allen et al. 2001, Palanques et al.
2006). Enriched deposition of organic matter, in turn, can enhance food availability for benthic faunal communities (Nodder et al. 2003, De Leo et al. 2010, Papiol et al. 2012). The interaction between the Southland Current and the local bathymetry may mean that prey aggregate over northeast facing slopes, providing enhanced foraging opportunities for sperm whales. Slope orientation influences the spatial distribution of sperm whales around the Balearic Islands (Pirotta et al. 2011) and of Blainville’s beaked whales in the Abaco canyons off the Bahamas (MacLeod & Zuur 2005). The orientation of submarine canyons, known to influence local hydrographic properties and zooplankton distributions (Macquart-
Moulin & Patriti 1996, Allen et al. 2001), appears to also determine how top predators use their habitat.
The winter GAMs indicated that sperm whales foraged more often in areas where the sub- surface peak in chl-a was relatively low, and where the peak was at depths between 20 and
50 m. It is important to note, however, that the general area off Kaikōura is characterised by high pelagic primary productivity (Leduc et al. 2014). Although the preferred chl-a values
(0.5 – 0.9 μg/l) are low compared to the available range within the study area, they are still indicative of relatively high density of phytoplankton. The observed correlation might be an artefact of the small spatiotemporal scale of the study. While the strength of this approach is to detect fine-scale spatial patterns in habitat use, it is not well set up to account for the large time lags needed for top predators to respond to changes in primary productivity at the base of the food web (Jaquet 1996, Croll et al. 2005), unless the spatial patterns are stable over time. Studies at larger spatial scales have identified a relationship between lagged chl-a surface concentrations and the distribution of top predators weeks or months later (Jaquet &
Whitehead 1996, Wong & Whitehead 2014, Giorli et al. 2016). Nevertheless, the distribution of top predators can correlate with sub-surface chl-a peaks in real time (Scott et al. 2010, Saijo Chapter 4 – Habitat preferences 114
et al. 2017), possibly driven by a more temporally stable relationship with increased densities of mesopelagic prey (Hazen & Johnston 2010, Saijo et al. 2017). The importance of sub-surface chl-a peaks for top predators remains poorly understood, but could have an important role in shaping their distributions. Alternatively, the preference for habitat with low sub-surface chl-a might reflect correlation with an unmeasured environmental variable, or with the as yet poorly understood habitat preferences of sperm whale prey.
The areas predicted by the GAMs to have the highest habitat suitability during winter were the Conway Trough and the slopes of the Conway Ridge. These areas overlap with spots where commercial fishermen find high abundances of demersal and benthopelagic fish during winter, including hāpuku and ling (Ben Cooper, pers. comm.), and sperm whales sometimes forage in proximity to fishing net locations (pers. obs.). This supports the hypothesis that the whales’ habitat preferences are partly reflecting the habitat requirements of demersal fish. It is consistent with stable isotope analyses of sperm whales and potential prey (Chapter 3), which suggest that demersal fish make up a larger proportion of the whales’ diet in winter than in summer.
4.4.3. Habitat preferences in the context of seasonal and inter-annual variability
Overall, differences in foraging habitat preferences of sperm whales off Kaikōura between summer and winter are consistent with seasonal patterns in diving and echolocation behaviour (Jaquet et al. 2000, Guerra et al. 2018) and stable isotope signatures (Chapters 2 &
3). This synchrony leaves little doubt that sperm whales modify their behaviour, spatial distribution and diet in response to temporal fluctuations of prey availability off Kaikōura.
The differences between habitat preferences in summer and winter indicate that many of the processes described vary over seasonal time scales. Therefore, it is important to consider the dynamic nature of the explanatory variables and the lag times which might influence the distribution of high trophic level predators.
There was some inter-annual variability in the distribution of preferred habitat. For example, the head of the Kaikōura Canyon was not predicted as particularly good habitat in winter
2015, but very favourable in winter 2016. Natural fluctuations in climate conditions among years are likely to be partially responsible for this – for example, during the three years of the study, New Zealand experienced a La Niña, a normal, and an El Niño stage of the Chapter 4 – Habitat preferences 115
Southern Oscillation Index (NIWA 2018). In addition, a strong earthquake affected Kaikōura in November 2016, resulting in substantial changes to the canyon seabed (Mountjoy et al.
2018), and in the spatial distribution and behaviour of sperm whales (Guerra et al. 2018).
Differences in habitat use before vs after the earthquake (i.e., three field seasons pre- and three seasons post-earthquake) probably introduced unusual variability in the distribution of optimal foraging habitat. Such variability was not easily accounted for by the models, probably lowering the proportion of explained variance.
4.4.4. Strengths and limitations of species-distribution models
Linking environmental conditions to specific behaviours by organisms (e.g., breeding, foraging, calving) can greatly enhance the accuracy of species-habitat models (Shillinger et al. 2011, Doniol-Valcroze et al. 2012, Rayment et al. 2014). Modelling habitat preferences at a feeding ground, and the use of echolocation clicks to define the presence or absence of foraging, was a convenient way to relate optimal habitat specifically to foraging requirements. The models performed well in predicting sperm whale presence off Kaikōura, particularly in summer. The nature of cetacean-habitat relationships strongly depends on the resolution at which the variables involved are examined (Jaquet 1996, Pirotta et al. 2011), with larger spatial scales generally resulting in more accurate model predictions (Karl et al.
2000). In this study, however, metrics of model performance (i.e., sensitivity, specificity and
AUC values) were within the range of other sperm whale distribution models (e.g., Praca et al. 2009, Torres et al. 2011, Pirotta et al. 2011, Tepsich et al. 2014), in spite of the data being gathered over a much smaller area. The SDMs successfully characterised variations in foraging distribution over relatively small distances (< 5 km), providing insights into fine scale habitat selection within a canyon system. Building habitat models at a fine spatial resolution, however, presented some challenges. The performance of the SDM was probably reduced by the mobility of sperm whales and the mismatch between foraging depths (> 400 m) and measurement of oceanographic features (< 400 m). In addition, the naturally low densities at which whales are found meant that considerable research effort was required to collect enough sightings data within the study area. For future research, alternative analytical techniques (such as boosted regression trees or Bayesian occupancy models) may improve the power to detect important species-habitat relationships (Redfern et al. 2006). Chapter 4 – Habitat preferences 116
While including oceanographic data from the water column is an advancement in the habitat modelling for sperm whales, hand hauling a CTD imposed some limitations. These included the depth restrictions of an instrument small enough to be hauled from a small vessel, and the long time required to carry out each cast (c. 25-30 mins), which meant that more intensive sampling within pseudo-absence areas was not practical. Consequently, I was necessarily constrained to making inferences about deep foraging habitat from observations in the top
350 m, and to assume that oceanographic characteristics did not vary substantially within
2.5 nmi. More complete water column data would remove the need to make these assumptions and possibly improve model performance, but would come at the expense of requiring a larger vessel and more expensive instruments.
4.4.5. Overlap with marine reserve
Marine protected areas (MPAs) have the potential to be a powerful tool for the conservation of pelagic ecosystems and top predators (Hooker et al. 1999, Hyrenbach et al. 2000).
However, designating an MPA that is actually effective is challenging (Hooker et al. 2011).
The value of the Hikurangi Marine Reserve for the protection of sperm whales and their foraging habitat depends mainly on the benefit of: (1) reducing the risk of entanglement with fishing gear, and (2) reducing the effects of fishing pressure on sperm whale prey. The main deep-water commercial fisheries that operate off Kaikōura use demersal gillnets and long- lines. There is some risk of entanglement (Reeves et al. 2013), given the spatial overlap between fishing and sperm whale distribution, however the rate of interaction with fishing gear is unknown. There is also overlap between the commercially targeted species and sperm whale prey – mainly demersal fish (e.g., hāpuku, ling), which are targeted more often in winter by fisheries and possibly whales. In this context, the marine reserve offers marginal protection during winter, given the small spatial overlap with the whales’ foraging areas. In summer there is more overlap, with potentially some benefit from reducing the chances of entanglement and prey extraction. However, since most fishing used to take place outside the reserve boundaries anyway, the reserve has done little to remove or displace fishing effort. Therefore, it seems that the current reserve is not a particularly useful conservation measure for Kaikōura sperm whales. Modifying the boundaries to cover more foraging habitat would be beneficial by reducing entanglement risk. Better knowledge on the indirect Chapter 4 – Habitat preferences 117
impacts of fisheries on the population would be necessary to make a well-informed judgement on the potential protection value of the reserve in terms of prey competition.
4.4.6. Future research
Information on the temporal and spatial variation of foraging distribution is key to understanding how populations of top predators may respond to changes in their environment (Arnould et al. 2011). The correlation of vertical gradients of temperature with sperm whale foraging habitat suggests that inter-annual variability in ocean conditions could influence prey abundance and whale distribution off Kaikōura. Potential linkages between oceanographic patterns and sperm whale distribution are therefore worthy of investigation and addressed in Chapter 5.
While SDMs based on environmental variables provide important insights into the habitat use of sperm whales, they are clearly a simplification of the intricate relationships between foraging and the environment. Incorporating information on other relevant topographic and oceanographic variables (e.g., sediment type, habitat rugosity, topographic enclosure, water flow, dissolved oxygen, nutrients) and from greater depths (i.e., to sperm whales diving depths) could improve our understanding of what constitutes quality habitat. There is still much to learn about the specific mechanisms that cause aggregations of sperm whale prey.
This is reflected in the moderate variability explained by sperm whale habitat models globally (typically less than 40%), which is good but could be improved. To date, what is missing from sperm whale habitat models is information on prey availability and distribution. Simultaneous sampling of foraging depths, environmental variables and prey fields (e.g., using hydro-acoustic surveys; Hazen et al. 2009, 2011, Benoit-Bird et al. 2013) is the next step to unravelling the foraging ecology of sperm whales and understanding their close association with submarine canyons.
4.4.7. Conclusions
The heterogenous distribution of sperm whales off Kaikōura reflected a selectivity for habitats with favourable foraging conditions, which were relatively consistent over a three- year period. A GAM approach to presence-absence SDMs indicated that the whales’ spatial distribution was shaped by oceanographic and topographic features. Physical factors that resulted in concentrating and aggregating processes were particularly important, Chapter 4 – Habitat preferences 118
highlighting the role of submarine canyons in enhancing prey availability and attracting top predators (Moors-Murphy 2014). These factors included strong vertical thermal gradients, steep slopes, and slope orientations that are likely to interact with local currents and concentrate food resources. An inverse correlation with the sub-surface peak in chl-a suggested a complex and indirect relationship with primary productivity, stressing the need for further study of the oceanographic and trophic dynamics in the area. There was strong evidence for seasonality in habitat use, suggesting that the Kaikōura feeding ground provides a temporally dynamic range of foraging opportunities for sperm whales. The seasonal variation in distribution also resulted in temporally varying overlap with the
Hikurangi Marine Reserve, with marginal protection of foraging habitat during winter. This study advanced our understanding of the environmental features that are important for male sperm whales foraging off Kaikōura. The results establish a framework for investigating the possible relationship between long-term changes in ocean conditions and the abundance of whales in the area.
Chapter 5.
Long-term variability in abundance of sperm whales at
Kaikōura in relation to ocean conditions.
5.1. Introduction
Multiple studies worldwide have documented changes in the population dynamics of marine predators concurrent with environmental variability (e.g., Baker et al. 2007, Trathan et al. 2007, Carroll et al. 2015). Although the underlying causes are often uncertain, it is well understood that oceanographic fluctuations, such as El Niño Southern Oscillation (ENSO), can strongly alter food webs (Barber & Chavez 1983, Ainley et al. 1995, Markaida 2006).
Climate-driven changes in sea temperature and primary productivity can change the distribution and abundance of zooplankton and other organisms at low trophic levels (Baier et al. 2003, Trathan et al. 2003). In turn, variations in prey densities can affect the distributions and foraging behaviour of predators (Lea et al. 2006, Worm & Tittensor 2011, Bost et al. 2015).
In severe cases, if food becomes limited, reproduction and survival can be impacted (Forcada et al. 2005, Soto et al. 2006, Baker et al. 2007). In the context of rapidly changing climate
(Hoegh-Guldberg & Bruno 2010), long-term studies of population dynamics are important for understanding how marine predators respond to variability in ocean conditions.
Based on time-series data spanning 28 years, the number of sperm whales foraging at
Kaikōura is in decline (Somerford 2018). The trend is driven by fewer whales visiting
Kaikōura during spring/summer, however the ultimate reasons for this are unknown. In addition, abundance in winter has been relatively consistent over the years, while in summer it has been more variable (Somerford 2018). The most likely explanation for the decline is a change in distribution away from the area, potentially driven by ecological or oceanographic variability affecting prey availability during summer. It is therefore important to investigate changes in whale abundance in relation to environmental conditions. The long-term series of abundance data (Marine Mammal Research Group, University of Otago; Somerford 2018), Chapter 5 – Time-series analysis 120
combined with satellite-derived oceanographic records, provide a unique opportunity to address this question. While additional factors could contribute to a change in distribution
(e.g., whale-watching tourism, fishing), assessing their influence was beyond the scope of this thesis.
There are several reasons why oceanographic variability might affect sperm whales and their food web. Studies elsewhere have suggested that changes in population size may be related to environmental shifts (Cantor et al. 2017), and that warm temperature anomalies decrease feeding success (Whitehead et al. 1989, Smith & Whitehead 1993) and modify residency patterns (Whitehead 1996). In addition, it has already been demonstrated that climate variations can affect the marine food web at Kaikōura. The abundance of euphausiid krill is correlated with ENSO variability, the occurrence of north-east winds, and the influx of colder waters from areas to the south of Kaikōura (Mills et al. 2008). In turn, variability in krill abundance influences the reproductive performance of red-billed gulls (Larus novahollandiae;
Mills et al. 2008). Lastly, squid, which are major prey for sperm whales, are very sensitive to environmental variability, particularly in sea temperatures (Waluda et al. 1999, Sims et al.
2001, Pierce & Boyle 2003, Markaida 2006, Pecl & Jackson 2008). This is of particular interest, given the warming in coastal waters around New Zealand over the last few decades, including off Kaikōura (Schiel et al. 2016, Shears & Bowen 2017).
The Kaikōura Canyon is in a very dynamic oceanographic setting (Larivière 2001). It lies near the highly productive Subtropical Front (STF; Murphy et al. 2001, Sutton 2001), which passes about 150 km to the southeast (Fig. 5.1). The STF is a global oceanic boundary where relatively warm, saline, macronutrient-poor (e.g., nitrate) and micronutrient-rich (e.g., iron)
Subtropical Surface Waters (STW) converge and mix with cooler, fresher, macronutrient-rich yet micronutrient-poor Subantarctic Surface Waters (SAW) (Heath 1981, Sutton 2001,
Hamilton 2006, Chiswell et al. 2015). East of New Zealand, the STF extends eastward along the southern edge of the Chatham Rise. However, on the western side of the Rise, ’wisps’ of
SAW regularly flow through the Mernoo Saddle, intruding northwards across the STF into the region south of Kaikōura (Shaw & Vennell 2000) as part of an extension of the Southland
Current (eSC). This eSC is the main coastal current flowing past Kaikōura (Chiswell 1996).
Further offshore, the eSC interacts with subtropical eddies shed by the warmer East Cape
Current (Greig & Gilmour 1992). Chapter 5 – Time-series analysis 121
Figure 5.1. Schematic representation of the oceanographic setting of the Kaikōura Canyon. The main ocean currents include the Southland Current (SC), extensions of the Southland Current (eSC), the d’Urville Current (dUC), and East Cape Current (ECC). The main eddies are the Wairarapa Eddy (WE), Hikurangi Eddy (HE) and E3. The water masses are Subtropical Water (STW) and Subantarctic Water (SAW). The Subtropical Front (STF) is indicated in light green. Depth contours show 200, 1000 and 2000 m isobaths. MS = Mernoo Saddle, BP = Banks Peninsula. Locations of currents and eddies are based on Heath (1972b, 1975), Greig & Gilmour (1992), Roemmich & Sutton (1998), Shaw & Vennell (2000) and Chiswell et al. (2015).
Thermal stratification is an important factor shaping the foraging distribution of sperm whales during summer (Chapter 4), and pelagic primary production is an important source of carbon sustaining the whales’ food web (Chapter 3). It is therefore relevant to examine how whale abundance correlates with processes that affect the temperature regime and phytoplankton productivity. Of particular interest is the wisp of SAW that typically flows Chapter 5 – Time-series analysis 122
northward through the Mernoo Saddle toward Kaikōura (Shaw & Vennell 2000). This wisp weakens from June to October, due to southward extensions of STW which block its passage
(Shaw & Vennell 2000). Variations in intrusions of SAW could affect primary production, heat fluxes or circulation patterns off Kaikōura, with flow-on effects for the canyon food web.
Indeed, the seasonal cycle of the SAW wisp broadly coincides with changes in abundance, distribution, and diving behaviour of the whales, potentially caused by changes in prey availability, as suggested by Jaquet et al. (2000). However, the effect of the SAW wisp on sperm whales or other marine mammals has never been systematically investigated.
Temperature and productivity can also be influenced by ENSO and the Southern Annular
Mode (SAM), the main climate modes affecting the marine environment around New
Zealand (Bhaskaran & Mullan 2003, Renwick & Thompson 2006). ENSO results from ocean- atmosphere interactions in the tropical Pacific Ocean, and is associated with global climatic anomalies (Barber & Chavez 1983, Wang et al. 2017). ENSO influences sea temperatures and thermal stratification, among other processes, and is thought to impact the feeding success of sperm whales (Smith & Whitehead 1993, Whitehead & Rendell 2004), and the distribution of other top predators (Neumann 2001, Bost et al. 2015, Sprogis et al. 2017). The SAM modulates the latitude and strength of the belt of westerly winds around the Southern Ocean
(Gong & Wang 1999, Lovenduski & Gruber 2005). Fluctuations in the SAM cause changes in sea temperatures, stratification and primary production through the Southern Ocean
(Lovenduski & Gruber 2005, Gillet et al. 2006, Sallée et al. 2010), but its effect on marine predators is not well understood (Forcada & Trathan 2009).
Before changes in oceanographic conditions affect the food webs and distribution of marine predators, lags of weeks to months can occur (Jaquet 1996, Croll et al. 2005). For example, previous studies have suggested that primary productivity influences sperm whales at lags of 0 to 6 months (Jaquet et al. 1996, Praca et al. 2009, Pirotta et al. 2011, Wong & Whitehead
2014, Giorli et al. 2016). Correlations between sea temperature and sperm whale distribution have been recorded at lags of 0 to 1 months (Pirotta et al. 2011, Torres et al. 2011, Wong &
Whitehead 2014), while squid can respond to changing temperatures after weeks, months, and over a year (Waluda et al. 1999, Sims et al. 2001, Pierce & Boyle 2003). Time lags for the effect of ENSO on marine predators range from no lag to more than one year (Smith &
Whitehead 1993, Forcada et al. 2005, Soto et al. 2006, Sprogis et al. 2016). Chapter 5 – Time-series analysis 123
In this study, I investigated the environmental factors that might explain the inter-annual variability in abundance of sperm whales at Kaikōura during summer. Specific questions included:
1) Are fluctuations in ocean conditions correlated with the abundance of sperm whales?
2) What are the temporal lags for the effect of environmental variables on the whales?
3) Does environmental variability help to explain the long-term decline in whale
abundance?
To address these questions, I used data on relative abundance of sperm whales at a monthly scale and total abundance at a seasonal scale during summer from 1990 to 2017. In combination with satellite-derived oceanographic data and climatic indices, I explored associations between whale abundance and environmental variables at different time lags.
Due to the study area covering only part of the sperm whales’ distribution, abundance at
Kaikōura does not reflect population size (number of whales alive) but rather the number of whales present in the area during a sampling period.
5.2. Methods
5.2.1. Sperm whale photo-identification surveys
The study site is an area of 850 km2, to the south and east of the Kaikōura Peninsula, centred over the Kaikōura Canyon (Fig. 5.2). Photo-identification surveys for sperm whales were carried out by the Marine Mammal Research Group (MMRG, University of Otago) from 1990 to 2017, for a total of 18 summers and 21 winters (detailed in Somerford 2018). Field seasons ranged from two weeks to three months in duration, typically between November and
January (spring-summer, hereafter ‘summer’) and between June and July (hereafter,
‘winter’). Surveys were carried out from small (c. 6 m) research boats, powered by outboard engines. Despite subtle differences in methodology since 1990, survey protocol was fairly consistent, and aimed to search the study area as uniformly as permitted by the weather conditions.
Chapter 5 – Time-series analysis 124
Figure 5.2. Locations of data collection for the analysis of correlations between abundance of sperm whales off Kaikōura and satellite-derived oceanographic variables. Locations include the study area where sperm whale photo-identification surveys were conducted from 1990 to 2017 (dashed blue line), area for characterising SST and SSC off Kaikōura (red and green squares, respectively), and areas for characterising SST of Subantarctic water (SAW), Subtropical water (STW) and at the Mernoo Saddle (orange squares), required to calculate the Mernoo Saddle mixing ratio.
Whales were tracked acoustically with a custom-built directional hydrophone (Dawson
1990) until visually located at the surface. Data collected during a sperm whale encounter included the time and location of dive, and a photograph of the flukes for individual identification, taken at the time of diving (Childerhouse et al. 1995). While camera equipment changed over the 28 years of study, photographic methods remained consistent. Search effort Chapter 5 – Time-series analysis 125
data were available since 1994, with the position of the research vessel logged at least once every 120 s. Search effort and encounter data were recorded on a HP-200LX palmtop computer running custom written software and interfaced with a GPS, or on a Samsung
Galaxy Tab A tablet running Cybertracker software. Research was conducted during daylight hours, in sea states ≤ Beaufort 3 and swell heights ≤ 2 m.
Photo-identification methods are detailed in Childerhouse et al. (1995, 1996). Briefly, the identity of each whale was determined by matching its photograph to the existing Kaikōura photo-ID catalogue. The use of poor-quality images can result in mis-identifications and heterogeneity of capture probability (Urian et al. 2015). Therefore, to be included in the analysis, photographs had to meet the following criteria: the entire trailing edge and notch of the flukes must be included in the photo, the flukes must be vertical and in focus, and the photograph must be taken at a perpendicular angle with the flukes. The photo-identification data set from 1990 to 2017 included 6231 sightings, resulting in 2111 sightings after excluding repeated encounters with the same individual in a day.
5.2.2. Monthly averages of relative abundance
To quantify the relative abundance of sperm whales within the study area in summer, I calculated the average ‘abundance index’ for each month. The abundance index was defined as the number of individuals encountered per 25 nmi in a day, standardised by survey effort.
Only data from November, December and January were used to maximise consistency among summers. For each day, the relative abundance index (AI) was calculated as:
푁푖 퐴퐼 = 푥 25 푠푢푟푣푒푦 푒푓푓표푟푡푖 where Ni = the number of unique individuals encountered on day i, and survey efforti = the distance covered searching for whales on day i, in nautical miles (nmi). The multiplier 25 was included to obtain an intuitive measure of relative whale abundance (e.g., Jaquet &
Gendron 2002, De Stephanis et al. 2008), based on 25 nmi being the average distance covered on effort per day (n = 316, CV = 0.45). Daily values of relative abundance were averaged per month. Days with ≤ 5 nmi of survey effort (n = 18) were excluded from further analysis because they were unlikely to provide a robust estimate of relative abundance. Likewise, only months that had at least 5 days of effort were included in the analysis. Relative Chapter 5 – Time-series analysis 126
abundance could not be calculated for summers prior to 1994 (nor for summer 2008) because
GPS effort data were not available.
5.2.3. Summer estimates of sperm whale abundance
Abundance estimates for each summer were calculated by Somerford (2018). Details of the methods are presented in Somerford (2018) and summarised in Guerra et al. (2018). Briefly, abundance estimates were based on capture-recapture analyses, using high-quality identification photos from each summer. Abundance estimates were calculated using
Pollock’s ‘robust design’ capture-recapture models (Pollock 1982) constructed in program
MARK (White & Burnham 1999). This methodology allows whales to be temporarily absent from the study area and is therefore well suited to populations open to migration (Peñaloza et al. 2014). The robust design combines open and closed population models, and is structured to have secondary sampling periods within primary sampling periods. The population is assumed to be closed within primary periods but open between primary periods.
In Somerford (2018), primary periods were defined as summers, each of which contained two secondary periods of approximately equal length (± 1 days). Summers were included in the analysis if they consisted of at least 12 survey days over a period of 3-5 weeks. For consistency, summers included the months November-February. This resulted in 16 primary periods between 1990 and 2017. Individual sightings data were summarised in an encounter matrix of presences and absences for each secondary period. Somerford (2018) constructed a suite of models allowing for different combinations of constant or time-varying survival rate, capture probability and temporary emigration rate. Models were ranked according to AICC
(Akaike 1973) and model-averaged abundance estimates were generated using the Akaike weights of the competing models (Burnham & Anderson 2002).
I used summer abundance estimates and monthly averages of relative abundance to investigate whether climate variability correlated with changes in the extent to which sperm whales visited Kaikōura. Each of the two metrics offered different strengths. Abundance estimates account for varying rates in capture probabilities, apparent survival and temporary emigration rates (detailed in Somerford 2018), representing the total number of whales visiting the study area during a season. Changes in abundance can thus provide useful insights into population dynamics in relation to climate variability (Seyboth et al. Chapter 5 – Time-series analysis 127
2016, Sprogis et al. 2017). Conversely, monthly estimates of relative abundance provide a measure of whale densities at a finer temporal scale. In addition, the relative abundance index indirectly incorporates information on how long individual whales remain in the study area. For example, a small number of whales which remain in the study area for an entire season may result in a greater mean relative abundance index than a larger number of individuals which each visit only briefly.
5.2.4. Environmental data
To characterise oceanographic conditions off Kaikōura from 1990 to 2017 I used data on sea surface chlorophyll-a concentration (SSC), sea surface temperature (SST), the relative influence of SAW intrusions through the Mernoo Saddle, ENSO and the SAM.
Monthly means of SSC data were obtained from the European Space Agency Climate
Change Initiative (ESA CCI), available from the Ocean Colour website at https://rsg.pml.ac.uk/thredds/ncss/grid/CCI ALL-v3.1-MONTHLY/dataset.html. I used data at 0.042° resolution (ca. 4.7 km) collected by SeaWiFS, MODIS-Aqua, MERIS and VIIRS satellite sensors (Sathyendranath et al. 2018). Since satellite-derived surface chlorophyll records did not become available until September 1997, the data set for SSC was limited to
1997-2017. Monthly SSC anomalies (SSCA) for the area offshore of Kaikōura (Fig. 5.2) were calculated as the deviation of each month’s datapoint from the mean for that month between
1997 and 2017. Anomalies were used instead of raw values to remove the influence of seasonality on the analysis (e.g., Lea et al. 2006, Bost et al. 2015). The inshore portion of the study area was not included for data extraction (Chiswell et al. 2017) as suspended sediments and organic matter in turbid coastal waters can influence radiance and bias SSC measurements (Gons 1999, Dall’Olmo et al. 2005).
Daily SST data were obtained from NOAA National Centers for Environmental Information
(NCEI). I used Reynolds’ optimally interpolated SST data at 0.25° resolution (ca. 20 km), v.2, collected by Advanced Very High Resolution Radiometers (AVHRR) on board NOAA satellites blended with ship and buoy data (Reynolds et al. 2007). Monthly means of SST anomalies (SSTA) for the area off Kaikōura (Fig. 5.2) were calculated as the deviation of each month’s datapoint from the mean for that month between 1990 and 2017. Data were available from https://coastwatch.pfeg.noaa.gov/erddap/griddap/ncdcOisst2Agg.html. Chapter 5 – Time-series analysis 128
Daily SST data were used to estimate the relative influence of cool SAW on the waters of the
Mernoo Saddle (Fig. 5.2), where the SAW wisp is typically located (Shaw & Vennell 2000).
Shaw & Vennell used high-resolution (1.1 km) SST images over a period of three years to determine the daily presence or absence of the northward intrusion of cool SAW (the SAW wisp). They established that the SAW wisp could be identified based on the similarity of SST to SAW further south, and its difference in SST to STW further east. Following this principle,
I used a simplified method to estimate the relative influence of each water mass at the
Mernoo Saddle (similar to Mills et al. 2008), rather than the presence or absence of the SAW wisp. The SST at the mean position of the wisp (based on Shaw & Vennell 2000) was compared to the SST in two reference areas representative of SAW and STW (Fig. 5.2).
Reynold’s optimally interpolated SST data were obtained from NOAA NCEI (Reynolds et al. 2007, detailed above). Daily mixing ratios of SAW versus STW (e.g., Chiswell & Sutton
1998, Sutton 2003) were used to estimate the relative contribution of each water mass, and were calculated as:
푆푆푇푀푆 − 푆푆푇푆푇푊 푀푅푀푆 = 푆푆푇푆퐴푊 − 푆푆푇푆푇푊 where MRMS = mixing ratio of water at the Mernoo Saddle, SSTMS = SST at the Mernoo Saddle,
SSTSTW = reference SST for STW, and SSTSAW = reference SST for SAW. Due to the dynamic nature of the region, the reference areas may not be exclusively SAW or STW. The values of the MRMS were therefore considered as a relative rather than absolute representation of SAW vs STW content. MRMS values closer to 0 indicate a greater influence of STW, whereas values closer to 1 indicate a greater influence of SAW. Monthly averages of MRMS were calculated based on daily ratios.
ENSO fluctuates between three phases: La Niña, Neutral and El Niño, each of which is associated with different patterns of winds and SST. The Southern Oscillation Index (SOI) measures the strength of ENSO phases (Trenberth 1997). Monthly SOI values (Ropelewski
& Jones 1987) were obtained from the Climatic Research Unit (CRU), available at KNMI
Climate Explorer (https://climexp.knmi.nl/start.cgi).
The SAM index describes the north-south movement of the westerly wind belt that circles
Antarctica (Gong & Wang 1999, Lovenduski & Gruber 2005). In a positive SAM event, the belt of strong westerly winds contracts towards Antarctica, with lighter winds over New Chapter 5 – Time-series analysis 129
Zealand latitudes. In the negative phase, an expansion of the belt of westerly winds towards the equator results in enhanced winds over New Zealand (Renwick & Thompson 2006).
Monthly SAM values were obtained from the Natural Environment Research Council
(NERC) of the British Antarctic Survey (Marshall 2003), available at KNMI Climate Explorer
(https://climexp.knmi.nl/start.cgi).
5.2.5. Temporal lags in environmental variables
While associations between climate indices and marine mammal reproduction have been identified at scales of several years (e.g., Leaper et al. 2006, Seyboth et al. 2016), the effects of climate on food webs are typically at shorter time scales (e.g., Ainley et al. 1995, Le Bohec et al. 2008). For this reason, and because of the limited sample size of whale abundance values,
I constrained examination of lagged effects to a maximum of one year.
Cross-correlation analyses, which require continuous data sequences, are often used to detect the best time scale for climate or oceanographic lags (e.g., Leaper et al. 2006, Seyboth et al. 2016). Due to the missing data caused by months or years when abundance estimates were unavailable, cross-correlation analyses were not well suited for this study. Instead, a suite of ecologically reasonable lags was considered for each environmental variable to investigate its potential correlation with the abundance of sperm whales (Table 5.1). To investigate the lagged response of monthly abundance indices (which included data from
November, December and January), monthly estimates of temperature-related predictor variables from 0 to 12 months prior were calculated. For the analysis of summer abundance
(typically spanning from November to January), 3-month averages of predictor variables were calculated at 3-month intervals, including lags from 0 (i.e., average value for
November, December and January) to one year before (i.e., average value for November,
December and January from the year before). In the case of SSCA, lags of only up to six months were considered, given the existing evidence concerning the link between primary productivity and sperm whales (Jaquet et al. 1996, Praca et al. 2009, Pirotta et al. 2011, Wong
& Whitehead 2014). The correlation of SSCA and whale abundance was only examined in the analysis of the monthly abundance index; it was excluded from the analysis of summer abundance because: (1) the sample size of the response was further reduced by the lack of Chapter 5 – Time-series analysis 130
SSCA data prior to 1997 (from n = 16 to n = 10), and (2) there was no evidence for correlation between SSCA and relative whale abundance at a monthly scale (see results).
Table 5.1. Temporal scale of response and explanatory variables considered in the analysis. SSCA = sea surface chlorophyll-a anomaly, SSTA = sea surface temperature anomaly, MSMR = Mernoo Saddle mixing ratio, SOI = southern oscillation index, SAM = southern annular mode.
Time scale of Explanatory variable Response variable Candidate time lags response (environmental factor)
SSCA 0, 1, 2, 4, 6 mo Average of daily Monthly abundance index (Nov, Dec, Jan) SSTA, MRMS, SOI, SAM 0, 1, 2, 4, 6, 12 mo
0 mo (Nov-Jan), 3 mo (Aug-Oct), Summer Seasonal SSTA, MRMS, SOI, SAM 6 mo (May-Jul), 9 mo (Feb-Apr), abundance estimate (Nov to Feb) 12 mo (Nov-Jan 1 yr before)
5.2.6. Statistical analysis
Trend in monthly relative abundance. The long-term trend in monthly relative abundance index was investigated using a linear mixed model (LMM; McCulloch et al. 2008, Bolker et al. 2009) as a function of time. Each summer (e.g., summer 1998/99) was included as a random effect to account for temporal auto-correlation in the relative abundance index among months of the same season.
Environment vs whales. Potential correlations between environmental variability and sperm whale abundance (for both monthly and summer estimates) were investigated using linear models within an information-theoretic approach (Burnham & Anderson 2002). While ecological relationships are often complex and non-linear, the detection of non-linear correlations requires relatively large data sets (Zuur et al. 2009). Given the nature of the response variable, the sample size of this study was small (< 30), making linear models a more suitable approach. The response variable of monthly abundance index was modelled with LMM, with the environmental covariates SSCA, SSTA, MRMS, SOI and SAM as fixed effects, and each summer as a random effect. The abundance index was based on a count
(number of whales per day), but was turned into a rate after averaging per month and Chapter 5 – Time-series analysis 131
standardising by survey effort. Abundance index was therefore a continuous variable
(whales/distance), so models were fitted with a Gaussian distribution and identity link function. In the case of summer abundance, I used generalized linear models (GLM;
McCullagh & Nelder 1989), with the environmental covariates SSCA, SSTA, MRMS, SOI and
SAM as explanatory variables. Models were fitted with a Poisson distribution with a log link function. In both analyses, the response was weighted in proportion to the precision around each estimate, using the inverse variance (Barlow & Taylor 2005, Kutner et al. 2005,
Somerford 2018).
Selecting time lags. Including all the temporal scales for each environmental covariate would have made the analysis unnecessarily complex, even after excluding collinear variables. Therefore, an ad hoc procedure was performed to select the most appropriate temporal lags to use in a reduced full model (e.g., Pirotta et al. 2011, Wong & Whitehead
2014). This was done independently for the analyses of monthly abundance indices and summer abundance estimates. First, a suite of models was created with all possible combinations of lagged variables, restricting models to one lag per variable (e.g., SSTA could only be included at 0, 1, 2, 4, 6 or 12-month lag, in addition to other environmental variables).
This was done using the ‘dredge’ and ‘subset’ functions within the package ‘MuMIn’ in R
(Bartoń 2016). Collinear variables were also restricted from being combined in the same model. Collinearity among explanatory variables was assessed with Spearman rank correlation tests, with values greater than ±0.5 indicating collinearity (Booth et al. 1994, Zuur et al. 2009). Models were ranked according to Akaike’s Information Criterion corrected for small sample sizes (AICc; Akaike 1973, Burnham & Anderson 2002). The relative importance index of each lagged variable was calculated by summing the Akaike weights (model probabilities) of all models that included the predictor variable (Burnham & Anderson 2002,
Twist et al. 2016). Lastly, the variables with relative importance indices ≥ 15% were selected and included in a ‘reduced’ full model. The threshold of 15% was chosen based on a visual assessment of the distribution of importance indices (see Fig. 5.4); it could be adjusted if the analysis was found to be overly restrictive.
Final models and coefficients. Once the variables in the ‘reduced’ full model had been selected, the response was modelled with a suite of linear models including all combinations of the chosen variables, and, in the case of monthly relative abundance, with the factor Chapter 5 – Time-series analysis 132
‘summer’ as the random effect. Models were ranked according to AICC, with the best models indicated by the lowest AICC values (Burnham & Anderson 2002, Symonds & Moussali
2011). Akaike weights were used to evaluate the support for the selected models. To estimate the magnitude and precision of the correlation between each environmental factor and the response I used two types of estimates: (1) the coefficients and 95% CI of the variables included in the best model, and (2) model-averaged coefficients (Burnham & Anderson 2002,
Burnham et al. 2011). Model-averaging produces parameter and error estimates whereby each model contributes to a weighted mean in proportion to its model probability (Burnham
& Anderson 2001, Burnham et al. 2011). This method is easily applied to linear models, and is particularly useful when there is uncertainty around which is the best model in the set
(Nakagawa & Freckleton 2011, Symonds & Moussalli 2011). Models were included in model- averaging if their ∆AICC value was <6 (Richards et al. 2011, Symonds & Moussalli 2011). The
95% CI of model-averaged coefficients were calculated based on unconditional standard errors, which incorporate model selection uncertainty (Burnham & Anderson 2002). Models were created in R studio 1.1.4 (R development Core Team 2012), using the packages ‘lme4’
(Bates et al. 2016) and ‘MuMIn’ (Bartoń 2016).
Model evaluation. The goodness-of-fit of the best models was assessed by calculating R2 values, which describe the amount of variance explained. In the case of LMM, two values were produced for each model: marginal R2 (variance explained by fixed effects only) and conditional R2 (variance explained by fixed and random effects (Orelien & Edwards 2018).
R2 values were calculated using the ‘sem.model.fits’ function within the package
‘piecewiseSEM’ in R (Lefcheck 2015). Diagnostic plots (Appendix 3, Fig. A3.1) were used to verify the assumptions of normality (via histograms of residual distributions and normal Q-
Q plots) and homogeneity of variance (via plots of residuals vs fitted values). Temporal autocorrelation in the full and best models of each analysis was examined using the ‘acf’ function in R, and was not a concern (Appendix 3, Fig. A3.2).
Additional considerations. (1) Due to the restricted sample size in this study, two measures were taken to prevent model overfitting and to simplify interpretation of results: (a) no interaction terms were included, (b) models were limited to a maximum of three predictors, as having a low ratio of predictors to observations (e.g., lower than 1:5) can cause spurious results (Zuur et al. 2009). (2) For the selection of variables in the analysis of monthly relative Chapter 5 – Time-series analysis 133
abundance, the data set was first limited to months when SSCA data were available (n = 21).
Because there was no strong evidence to select SSCA as an important predictor of whale abundance (see results), the rest of the analysis was carried out using the full data set (n =
30). (3) Standardizing input variables is sometimes recommended so that model coefficients are on a common scale (Gelman 2008, Grueber et al. 2011). The required data-transformation, however, can distort coefficients, especially if the input variables do not fit an ideal Gaussian distribution (Baguley 2008, Grace et al. 2018). For simplicity, and given the small sample size of the study, I did not use standardization methods. Model coefficients were therefore not used to establish the relative effect size of environmental variables on sperm whale abundance.
5.2.7. Long-term trend in modelled outputs of summer abundance
To visualise the trend in summer abundance of sperm whales predicted by the best model, model predictions were produced for every summer from 1990 to 2017, based on the environmental conditions of each year. The long-term trend was investigated using a weighted linear regression. Pre-requisites of normality were checked using a Shapiro-Wilk test (Shapiro & Wilk 1965). Modelled estimates were weighted according to the precision of each estimate, using the inverse of the standard error, as this was the measure of uncertainty produced by the prediction function. Lastly, I examined the long-term trend in the environmental variables that were selected as important predictors of summer abundance.
This was done to better understand the potential changes in ocean conditions concurrent with the decline in the number of whales at Kaikōura.
5.3. Results
5.3.1. Abundance of sperm whales at Kaikōura
After data filtering there were 30 values of monthly relative abundance of sperm whales during summer from 1994 to 2017. The analysis used 933 sightings of unique individuals/day, recorded over 276 days in a total of 14 summers, covering 690 nmi (1278 km) of effort. Monthly averages of daily relative abundance ranged from 1.2 to 7.7 whales/25 nmi (Fig. 5.3a; Appendix 3, Table A3.1). The long-term trend in monthly relative abundance Chapter 5 – Time-series analysis 134
index indicated a significant decline (-1.26/decade, R2 = 0.40, p < 0.001). Robust design analysis resulted in 16 summer abundance estimates from 1990 to 2017, based on 846 photo-
ID encounters recorded over 245 days (Somerford 2018). Abundance estimates ranged from
5 to 43 whales per summer (Fig. 5.3b), with a long-term trend of significant decline (-
7.40/decade, R2 = 0.48, p = 0.01; Somerford 2018).
(a) 12 November 10 December January 8
6
4 # whales/25 whales/25 # nmi 2
0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 Year
(b) 60 Summer 50
40
30 # whales # 20
10
0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 Year
Figure 5.3. Temporal variability in abundance of sperm whales off Kaikōura, 1990-2017. (a) Monthly averages of daily relative abundance, with 95% CI error bars; (b) ‘capture-recapture’ estimates of summer abundance, with log-normal 95% CI error bars (from Somerford 2018). The upper limit of the 95% CI of the abundance estimate for summer 1990 is 148 (axis is cropped for visual clarity).
Chapter 5 – Time-series analysis 135
5.3.2. Correlation of environmental variables with monthly abundance index
Among lagged environmental variables, MRMS was significantly correlated with SSTA
(Spearman r = -0.59, p < 0.001) and SSCA (Spearman r = -0.57, p = 0.004) at the same time lag.
These variables were therefore prevented from being combined in the same model for the calculation of relative importance indices. All climatic and oceanographic variables had indices of < 50%, indicating that none of them had exceptionally high support (Fig 5.4).
However, in the cases of MRMS, SAM and SOI, correlations with monthly abundance index of sperm whales were stronger at certain time lags. For SSTA, there was some uncertainty around which time lag was the better predictor of monthly abundance index. SSCA had relatively low importance scores at all the examined time lags and was excluded from further analysis. Overall, the variables with highest relative importance scores included: MRMS at no lag, SAM at no lag, SOI at a 4-month lag, and SSTA at three time-scales: no lag, 4 and 12 months. Since MRMS at no lag (MRMS_0) was correlated with SSTA at no lag (SSTA_0) and
MRMS was the covariate with greater support, SSTA_0 was excluded from the full model in further analysis. The other five variables (MRMS_0, SAM_0, SOI_4, SSTA_4 and SSTA_12) were included in the reduced full model and further investigated.
45%
30%
15% Relative Relative importanceindex
0%
SOI_0 SOI_1 SOI_2 SOI_4 SOI_6
SOI_12
SAM_0 SAM_1 SAM_2 SAM_4 SAM_6
SSTA_0 SSTA_1 SSTA_2 SSTA_4 SSTA_6
SSCA_0 SSCA_1 SSCA_2 SSCA_4 SSCA_6
SAM_12
SSTA_12
MR.MS_0 MR.MS_1 MR.MS_2 MR.MS_4 MR.MS_6 MR.MS_12 Candidate environmental variable
Figure 5.4. Relative importance indices of environmental variables at different time lags, derived from models predicting the monthly abundance index of sperm whales off Kaikōura. MR.MS = Mixing Ratio at the Mernoo Saddle, SAM = Southern Annular Mode, SOI = Southern Oscillation Index, SSTA = sea surface temperature anomaly, SSCA = sea surface chlorophyll anomaly. The lag of each variable (in months) is indicated by the number at the end of each name. The dashed red line indicates a relative importance index of 15%. Chapter 5 – Time-series analysis 136
The next step in the analysis was to investigate potential correlations between the selected environmental variables and the relative abundance index of sperm whales at a monthly scale. There was consistent support for models that featured the mixing ratio at the Mernoo
Saddle (at no time lag) as an explanatory variable (Table 5.2), with this variable achieving the highest index of relative importance (Table 5.3). There was some uncertainty around which was the best model in the candidate set (i.e., which variables other than MRMS were important), with the highest-ranking model scoring an Akaike weight of 35%. The best model also included the variables SSTA at a 12-month lag and SAM at no lag, and explained
27% of variance in the response. The correlation coefficients from the best model and from the model-averaged analysis indicated that whales were more abundant at Kaikōura during months when the MRMS had a higher contribution of SAW (i.e., was more positive) (Fig. 5.5a, b; Table 5.3). There was some evidence for a positive correlation between relative abundance and SAM, and for a negative correlation with lagged SSTA, with fewer whales when the previous summer had warm sea temperature anomalies (Fig. 5.5a). While the inclusion of both these factors improved the variance explained by the models, there was no statistical confidence that their effect was different from 0 when model-selection uncertainty was considered (Table 5.3; Fig. 5.5b). Lastly, there was very little support for a correlation between relative abundance of whales at Kaikōura and either SOI or SSTA at a lag of 4 months (Fig. 5.5a, b).
The low support for some of the factors included in the analysis provided assurance that a cut-off of 15% weight to select which lagged variables were included in the analysis was not overly strict (Fig. 5.4), and that no important predictors were being left out. The inclusion of
‘summer’ as a random effect improved the variance explained in all models, indicating some temporal dependency among monthly abundance indices during the same season. The inclusion of the random effect accounted sufficiently for this dependency, as indicated by the lack of significant temporal autocorrelation in the model (Appendix 3, Fig. A3.2). Overall, and taking a conservative approach to suit the available sample size, the only relationship that was consistently supported by the models was a positive correlation between relative abundance of sperm whales and MRMS at no time lag.
Chapter 5 – Time-series analysis 137
Table 5.2. Ranking of LMMs used to explain the monthly relative abundance of sperm whales off
Kaikōura from 1994 to 2017. Metrics of model performance are shown for models with ∆AICC <6. AI = relative abundance index (# whales/25 nmi); (1|summer) = random effect included in all LMMs; df
= degrees of freedom; ∆AICC = difference in AICC score relative to best model; wi = Akaike weight (model probability); Mar. R2 = marginal R2 (variance explained by fixed effects only); Con. R2 = conditional R2 (variance explained by fixed and random effects). The number at the end of each variable represents the time lag (in months) for the correlation with the response.
Rank Model df ∆AICC wi Mar. R2 Con. R2
1 AI ~ MRMS_0 + SSTA_12 + SAM_0 + (1|summer) 6 0.00 0.35 0.27 0.32
2 AI ~ MRMS_0 + SSTA_4 + (1|summer) 5 2.21 0.12 0.11 0.20
3 AI ~ MRMS_0 + SSTA_12 + (1|summer) 5 2.52 0.10 0.13 0.27
4 AI ~ MRMS_0 + SSTA_12 + SOI_4 + (1|summer) 6 2.78 0.09 0.21 0.36
5 AI ~ MRMS_0 + SSTA_4 + SOI_4 + (1|summer) 6 3.76 0.05 0.16 0.29
6 AI ~ MRMS_0 + (1|summer) 4 3.83 0.05 0.05 0.24
7 AI ~ MRMS_0 + SOI_4 + (1|summer) 5 3.99 0.04 0.13 0.37
8 AI ~ MRMS_0 + SSTA_4 + SAM_0 + (1|summer) 6 4.18 0.04 0.15 0.22
9 AI ~ MRMS_0 + SAM_0 + SOI_4 + (1|summer) 6 4.58 0.03 0.19 0.42
10 AI ~ MRMS_0 + SAM_0 + (1|summer) 5 4.63 0.03 0.10 0.28
11 AI ~ SSTA_12 + (1|summer) 4 5.78 0.02 0.09 0.24
(a) (b) 12 12
10 * 10 *
8 8
6 6
4 4
2 2 *
0 0
model model correlationcoefficient
-
averaged averaged correlation coefficient - -2 -2 Best *
-4 Model -4
SOI_4
SAM_0 SAM_0
SSTA_4
SSTA_12 SSTA_12 MR.MS_0 MR.MS_0 Figure 5.5. Correlation coefficients of environmental variables used as predictors of monthly relative abundance of sperm whales. Coefficients are from the best model in the set (a) and from model- averaged estimates (b). Error bars are 95% CI. The symbol ‘*’ indicates effects that are significantly different from 0. Note that environmental variables have different ranges, therefore comparisons of effect size cannot be made based on the magnitude of the correlation coefficient. Chapter 5 – Time-series analysis 138
Table 5.3. Parameter estimates from LMMs used to model the relationship between monthly relative abundance of sperm whales at Kaikōura and environmental variables. Estimate = correlation coefficient; SE = standard error; na = not applicable (variable not present in the best model).
Best-model parameters Model-averaged parameters Relative importance Variable Estimate SE p-value Estimate SE p-value index
MRMS_0 5.92 1.91 0.003 4.91 2.35 0.04 0.91 SAM_0 0.36 0.11 0.003 0.17 0.19 0.39 0.48 SSTA_12 -1.16 0.36 0.002 -0.63 0.61 0.31 0.58 SSTA_4 na na na -0.25 0.53 0.62 0.23 SOI_4 na na na -0.12 0.25 0.64 0.26
5.3.3. Correlation of environmental variables with summertime whale abundance
No collinearity was found among environmental variables in the analysis of summer whale abundance. Relative importance indices showed very clear support for some of the lagged variables (Fig. 5.6), with particularly high indices for MRMS at one-year lag (MRMS_12), and for SSTA during the previous winter (6-month lag, or May-July; SSTA_6). There was no support for the inclusion of climatic indices (SOI or SAM) at any of the time lags considered.
Variables with higher relative importance (index > 15%) included: MRMS_12, SSTA_6 and
SSTA_12. These three variables were included in the reduced full model and further investigated.
90%
75%
60%
45%
30%
15% Relative Relative importanceindex
0%
SOI_0 SOI_3 SOI_6 SOI_9
SOI_12
SAM_0 SAM_3 SAM_6 SAM_9
SSTA_0 SSTA_3 SSTA_6 SSTA_9
SAM_12
SSTA_12
MR.MS_0 MR.MS_3 MR.MS_6 MR.MS_9 MR.MS_12
Figure 5.6. Relative importance indices of environmental variables at different time lags, derived from models predicting the summer abundance of sperm whales off Kaikōura. The lag of each variable (at 3-month periods) is indicated by the number at the end of each name (see Table 5.1 for definitions of each lag). The dashed red line indicates a relative importance index of 15%. Chapter 5 – Time-series analysis 139
There was clear support for a correlation between whale abundance and MRMS in the previous summer (i.e., 12-month lag) and SSTA at a time lag of 6 months (Table 5.4). These two variables featured in the best model, which had a weight of 94% and explained 59% of the variance in the response. Both variables explained a similar amount of variance (30 –
35%) and had similar importance indices (0.94 – 1.00). The correlation coefficients from the best model and the model-averaged estimates indicated that summer abundance of whales at Kaikōura was lower when: (1) the MRMS in the previous summer had had a lower contribution of SAW (Fig. 5.7a, b; Table 5.5), and (2) there had been warmer temperatures 6 months before (i.e., May-July). There was very little support for inclusion of SSTA at one- year lag: the model that included this variable had 15 times less weight than the best model
(Table 5.4), and the model-averaged coefficient for SSTA_12 was very small (-0.04) and of no statistical significance (p-value = 0.81; Table 5.5). Overall, there was evidence to suggest a positive correlation with the SAW wisp in the previous summer, and a negative correlation with SSTA during the previous winter. The goodness of fit of models explaining summer abundance (Table 5.4) was at least twice as high as the models describing relative abundance index at a monthly scale (Table 5.2).
Table 5.4. Ranking of GLMs used to explain the summer abundance of sperm whales off Kaikōura from 1990 to 2017. Metrics of model performance are shown for all models in the candidate set. Abundance = abundance (# whales) estimated with capture-recapture robust design analysis; df = degrees of freedom; ∆AICC = difference in AICC score relative to the best model; wi = Akaike weight (i.e., model probability); R2 = explained variance. The number at the end of each variable represents the time lag (at 3-month periods) for the correlation with the response (see Table 5.1 for definitions of lags).
Rank Model df ∆AICC wi R2
1 Abundance ~ MRMS_12 + SSTA_6 3 0.00 0.94 0.59
2 Abundance ~ MRMS_12 + SSTA_12 3 5.58 0.06 0.56 3 Abundance ~ SSTA_12 2 30.04 0.00 0.41
4 Abundance ~ MRMS_12 2 41.12 0.00 0.35 5 Abundance ~ SSTA_6 2 50.64 0.00 0.30 6 null model 1 103.76 0.00 0.00
Chapter 5 – Time-series analysis 140
(a) (b) 6 6 * * 5 5
4 4
3 3
2 2
1 1
0 0
model model correlationcoefficient
averaged averaged correlation coefficient
- - -1 -1 Best * *
-2 Model -2
SSTA_6 SSTA_6
SSTA_12 MR.MS_12 MR.MS_12
Fig. 5.7. Correlation coefficients of environmental variables used as predictors of summer abundance of sperm whales at Kaikōura. Coefficients are from the best model in the set (a) and from model- averaged estimates (b). Error bars are 95% CI. The symbol ‘*’ indicates effects significantly different from 0. Note that environmental variables have different ranges, therefore comparisons of effect size cannot be made based on the magnitude of the correlation coefficient.
Table 5.5. Parameter estimates from GLMs used to model the relationship between summer abundance of sperm whales and environmental variables. Estimate = correlation coefficient; SE = standard error; na = not applicable (variable not present in the best model).
Best-model parameters Model-averaged parameters Relative importance Variable Estimate SE p-value Estimate SE p-value index
MRMS_12 3.99 0.57 < 0.001 3.94 0.60 < 0.001 1.00 SSTA_6 -0.58 0.09 < 0.001 -0.54 0.16 0.001 0.94 SSTA_12 na na na -0.04 0.16 0.81 0.06
Chapter 5 – Time-series analysis 141
5.3.4. Long-term trend in modelled abundance and environmental variables
Model predictions of sperm whale abundance were produced for every summer from 1990 to 2017 (n = 28), based on the highest-ranking model according to AICC (‘GLM-1’; Abundance
~ MRMS_12 + SSTA_6) and on the environmental conditions for each year (Fig. 5.8a). The weighted linear regression showed a significant decline in the modelled number of whales present at Kaikōura (-0.28/year, F = 5.10, p = 0.03), however there were relatively large deviations from the abundance estimates derived from capture-recapture analyses
(Somerford 2018). The decline from the environmental model (-0.28/year) was smaller than the decline estimated with the robust design analysis (-0.74/year, F = 16.23, p <0.001;
Somerford 2018). Therefore, while the model was the best one from the available candidate set and accounted for 59% of the inter-annual variability in whale abundance, there was a large component of the decline that was not successfully explained by fluctuations in MRMS or SSTA.
60 capture-recapture estimate 50 model predictions
40
30
20
sperm sperm whale abundance 10
0 1988 1992 1996 2000 2004 2008 2012 2016 2020
Figure 5.8. Summertime abundance of sperm whales at Kaikōura 1990-2017. Actual abundance estimates based on photo-ID capture-recapture methods and robust design modelling are shown in black (from Somerford 2018), while predicted abundance based on GLM-1 and environmental conditions are shown in green. Dashed lines represent weighted linear regressions. The time-series of model predictions (green) is slightly shifted to the right for visual clarity, so that data points and error bars are visible for all years; this shift conveys no information about time lags.
Chapter 5 – Time-series analysis 142
The long-term trend for the two environmental variables that were correlated with summer abundance of sperm whales was examined with linear regression (Fig. 5.9a, b). Linear regressions indicated a significant increase in SSTA in May-July (+0.20°C/decade, R2 = 0.16, p = 0.03), and a negative but not significant trend in MRMS in November-January (-
0.01/decade, R2 = 0.02, p = 0.43).
(a) 1.00
0.75
C) 0.50 °
0.25
July( - 0.00
-0.25
SSTA SSTA inMay -0.50
-0.75 SSTA_6
-1.00 1988 1992 1996 2000 2004 2008 2012 2016 2020
(b) 0.70 0.65 0.60
Jan 0.55 - 0.50 0.45 0.40
MR.MS MR.MS inNov 0.35
0.30 MR.MS_12 0.25 1988 1992 1996 2000 2004 2008 2012 2016 2020 Year
Figure 5.9. Time-series data for the duration of the study (1990-2017) for: (a) SSTA off Kaikōura in
May-July (at a 6-month lag before summer), and (b) the Mixing Ratio at the Mernoo Saddle (MRMS) during Nov-Jan with a 12-month lag. Error bars are 95% CI; red dashed lines represent linear regressions.
Chapter 5 – Time-series analysis 143
5.4. Discussion
This study was the first to examine long-term correlations between climate variability and abundance of sperm whales. While the analysis was based on a limited sample size, it spanned a period of almost three decades, and captured a wide range of oceanographic conditions and whale abundance. The results provided new insights into the environmental factors potentially driving variability in the number of whales foraging at Kaikōura in summer, and on the possible causes contributing to the long-term decline. Linear models indicated that two environmental variables were particularly correlated with whale abundance: the relative influence of SAW in waters over the Mernoo Saddle south of
Kaikōura, and local sea surface temperatures. Given the wide range in ocean conditions that sperm whales are exposed to during their migratory movements (Whitehead et al. 2008), direct effects on the whales seem unlikely. Environmental variability is more likely to influence sperm whales via effects on their prey.
5.4.1. Oceanographic fluctuations and relative abundance of whales at a monthly scale
The relative abundance of whales using the Kaikōura area during summer months (Nov,
Dec, Jan) declined significantly over the duration of the study. This is consistent with the long-term decline in summer abundance (Somerford 2018), suggesting that, not only are fewer whales visiting Kaikōura during summer, but they occur at lower densities.
Relative abundance of whales during summer was higher, to some extent, in months with a greater proportion of SAW in the Mernoo Saddle, presumably reflecting a stronger influence of the SAW wisp (Shaw & Vennell 2000). This effect was not lagged, suggesting a link with concurrent conditions in spite of the distance from the Kaikōura Canyon (c. 100 km).
Ecological succession from enhanced productivity is therefore unlikely to explain this correlation, as a lag longer than one month is typically required for changes in primary production to be reflected in prey abundance (Jaquet 1996, Croll et al. 2005), and SSCA was not an important predictor. On the other hand, prey-aggregating phenomena, such as thermoclines (Pelletier et al. 2012), mesoscale eddies (Sabarros et al. 2009) or oceanic fronts
(Bost et al. 2009), are more likely to mediate this link, as they act much more quickly.
Off Kaikōura, thermal stratification of the water column increases habitat suitability for sperm whales during summer (Chapter 4). A higher influx of SAW into the Kaikōura region Chapter 5 – Time-series analysis 144
is anticipated to result in cooler surface water, reducing thermal stratification. Thus, any influence of the SAW wisp on whale abundance is unlikely to be mediated via a control on stratification. An increase in the extension of the SAW wisp into STW may result in more numerous or persistent oceanic fronts between the two water masses in the Kaikōura region.
Oceanic fronts are thermal boundaries that can act as convergence zones of nutrients and plankton (Sabarros et al. 2009, Woodson & Litvin 2015), increasing the aggregation of mid- trophic consumers and creating foraging hotspots for marine predators (Scales et al. 2014,
Miller et al. 2015). The region between Kaikōura and the Mernoo Saddle has a complex configuration of mesoscale eddies (Heath 1975, Vincent et al. 1991, Gibbs & Shaw 2002, Shaw
& Vennell 2000). It is possible that strength of the SAW wisp affects the vorticity or location of eddies (Larivière 2001), potentially intensifying eddy-associated convergence zones. The presence of SAW at the Mernoo Saddle may also reflect stronger northward advection (Shaw
& Vennell 2000), which could lead to current-assisted movement of nekton (e.g., Pierce &
Boyle 2003). These processes may facilitate aggregation of food for sperm whale prey, or aggregate prey directly, increasing the occurrence of whales. The results are consistent with
Mills et al. (2008), who found that the abundance of krill at Kaikōura in late spring was correlated with the influence of SAW at the Mernoo Saddle.
Higher ratios of SAW vs STW at the Mernoo Saddle were correlated with colder SST at
Kaikōura in summer. Extensions of cold SAW may therefore influence the spatial distribution of sperm whale prey through temperature-mediated effects. The spatial distributions of squid and fish can shift due to increasing temperatures (Perry et al. 2005,
Dulvy et al. 2008, Rodhouse 2013). For example, warty squid (Onykia ingens) is likely to be a major prey for sperm whales in summer (Gaskin & Cawthorn 1967) and tends to be more abundant in colder water masses (Jackson et al. 2000). Its abundance at Kaikōura may be increased by cooler temperatures in months with higher influx of SAW. Alternatively, the association between the SAW wisp and whale abundance may be driven by other large-scale processes to which both variables respond. Clearly, the biological mechanisms linking water mass dynamics at the Mernoo Saddle and the abundance of whales in the canyon are complex.
Chapter 5 – Time-series analysis 145
5.4.2. Inter-annual variability in summertime whale abundance
The number of sperm whales using the study area in summer was higher when the proportion of SAW at the Mernoo Saddle had been relatively high the previous summer, and when sea surface temperatures had been colder during the previous winter. Unlike the analysis of monthly relative abundance, the analysis of summertime abundance indicated a lagged relationship between the proportion of SAW and the numbers of sperm whales. This result suggests ecological succession between physical variables and prey availability. The difference in lag times between the two analyses may have resulted from examining relationships at different scales. Monthly estimates of relative abundance of whales could reflect links with more immediate changes in the marine environment, while variability in summertime abundance may signal mechanisms operating at longer time scales. The association between the SAW wisp at the Mernoo Saddle and whale abundance at Kaikōura highlights the connectivity between circulation processes and top predators at large spatio- temporal scales.
While the effect of environmental variability on sperm whale abundance during summer may act through any component of their food web, it is most likely to be mediated by squid.
Squid are typically short-lived and fast-growing (Anderson & Rodhouse 2001, Pierce & Boyle
2003), take advantage of intermittent food resources (Jackson et al. 2005), and respond particularly strongly to environmental variability (Dawe et al. 2000, Rodhouse 2001). For example, changes in sea temperature can influence their growth, recruitment, migratory movements and abundance (McMahon & Summers 1971, Collins et al. 1997, Waluda et al.
1999, Sims et al. 2001, Pecl & Jackson 2008). Deep-water demersal fish, in contrast, have slower growth and longer life spans, and are comparatively more resilient to changes in environmental conditions (Leaman & Beamish 1984, King & McFarlane 2003). Additionally, squid are prey that are likely to be particularly important for sperm whales during summer
(Gaskin & Cawthorn 1967, Chapter 3).
To consider how lagged oceanographic variables may influence sperm whales, it is useful to consider the life cycles of squid. Warty and arrow squid are annual species, with spawning peaks during winter (Jackson 1997, Cherel & Weimerskirch 1999, Jackson 2001, Jackson et al.
2005, McKinnon 2007). Given this annual cycle, variations in the hydrology of the Mernoo Chapter 5 – Time-series analysis 146
Saddle could influence abundance or availability of squid the following year at Kaikōura.
For example, favourable foraging conditions may lead to increased recruitment later on.
Additionally, higher influx of SAW may promote the occurrence of convergence zones that concentrate nutrients and food resources (as explained in section 5.4.1), or enhance current- assisted transport for passive organisms. These processes may enhance the availability of prey for squid in the region.
There are at least two possible mechanisms to explain the correlation between warmer temperatures in winter with lower abundance of whales the following summer. Firstly, the relationship could be driven by direct effects of temperature on squid recruitment, given the similarity between the time of the correlation lag (winter) and the peak time for hatching of warty and arrow squid (Jackson 1997, McKinnon 2007). Warm temperature anomalies at the time of hatching can reduce squid recruitment (Waluda et al. 1999, Chen et al. 2007), and reduce the size of hatchlings (Villanueva 2000, Vidal et al. 2002, Steer et al. 2003). Favourable oceanographic conditions during winter spawning can strongly increase the abundance of adult squid (Dawe et al. 2000). If recruitment of squid was reduced during winter, a lower abundance of squid during summer would probably decrease the availability of prey for sperm whales, potentially resulting in fewer individuals foraging in the canyon area.
Secondly, warm temperature anomalies might be a proxy for other oceanographic conditions to which the food web is responding, such as stratification or thermocline depth (Soto et al.
2006, Bost et al. 2015). Warmer surface temperatures during winter could reflect stronger stratification in the water column, which may reduce nutrient supply to the photic zone
(Segar 2007). This, in turn, can limit plankton growth and reduce food resources available to higher trophic levels (Scott et al. 2006, Carroll et al. 2015). Previous studies have reported the impact of warm temperatures on pelagic top predators through changes in food availability, such as reduced biomass of krill and fish, and changes in species composition of fish assemblages (Forcada et al. 2005, Lea et al. 2006).
5.4.3. Long-term trends in environmental variability
Summertime and monthly analyses of sperm whale abundance provided consistent support for the importance of the SAW wisp during summer, regardless of the lag (0 vs 12 months).
Although the negative long-term trend in MRMS during summer was not statistically Chapter 5 – Time-series analysis 147
significant, the measure of MRMS used in this study was unlikely to capture the full variability of the SAW wisp. Significant long-term trends in its strength or structure should not be discarded based on these results.
The inverse correlation between winter temperatures and whale abundance is a cause for concern, given the increase in winter temperatures at Kaikōura over the last three decades
(Schiel et al. 2016; this study), and the global projections for further warming over the next century (IPCC 2014). A causal relationship between SST and the whales’ food web cannot be established without further investigation, however this study identifies sea temperature anomalies as a potential driver of the long-term decline in the abundance of sperm whales at Kaikōura.
The best model from the summertime analysis indicated a long-term decline in whale abundance, with a total explained variance of 59%. The modelled decline, however, was much less pronounced than the observed trend (Somerford 2018). While variability in sea temperature and SAW wisp dynamics may influence whale abundance from year to year, they do not completely explain the long-term decline. This implies that additional factors contribute to the observed trend. These factors could include other environmental features, fishing pressure, tourism, noise, ship strike, pollution, entanglement with fishing gear, or complex ecosystem interactions.
5.4.4. Caveats and limitations
Most studies on the variability of population size in relation to oceanographic conditions share one important limitation: a small sample size in spite of extended periods of survey effort (e.g., Seyboth et al. 2016, Sprogis et al. 2017). This limits the power to detect meaningful relationships between abundance and climate, and increases the risk of finding significant but spurious correlations due to concurrent trends (Zuur et al. 2009). It is therefore important to be cautious when interpreting the ecological significance of modelled relationships. In the present study, a conservative approach was applied through the use of multi-model inference and model-averaged estimates. The detected correlations identify potential ecological relationships, which can be further investigated to better understand the underlying mechanisms. Nevertheless, the population at Kaikōura is the subject of one of the two longest-running studies of sperm whales in the world, with the other one based at Chapter 5 – Time-series analysis 148
the Galápagos (Whitehead et al. 1997, Cantor et al. 2017). Even if the sample size is ‘small’, it provides the best available opportunity to examine long-term ecological relationships.
Applying other analytical methods, such as a Bayesian Hierarchical analysis framework
(King et al. 2009; e.g., Moore & Barlow 2013, Schick et al. 2013), could improve power and would be worth investigating.
A caveat particular to this study was the simplified method to characterise the SAW wisp at the Mernoo Saddle. Mixing ratios were useful as a first approach to examine trends in the relative contribution of SAW, however it had limitations. Firstly, large movements of the wisp could not be accounted for, due to the fixed position of the SST ‘box’ used to calculate the mixing ratio. This may have caused the ratio to underestimate the presence of the wisp at certain times. Future studies should consider more flexible methods to allow for wisp movements, such as front-detection algorithms within a larger area (e.g., Ullman & Cornillon
1999, Belkin & O’Reilly 2009). Secondly, the detection of the wisp based on SST could not capture dynamics in surface currents. This component of the wisp is of particular biological interest, therefore future studies might benefit from investigating variations in flow strength with current-detection methods (Bonjean & Lagerloef 2002).
Based on this study, there was no evidence for a correlation between phytoplankton productivity and sperm whale abundance at any lag. However, this may have been due to low statistical power due to a reduced time-series. In addition, the sea surface values used could not account for variability in sub-surface peaks in chlorophyll-a concentration, which could be more relevant for deep-diving predators (Hazen & Johnston 2010, Scott et al. 2010,
Saijo et al. 2017, Chapter 4 of this thesis). Lastly, chlorophyll-a concentration may not have fully captured temporal variability in phytoplankton production; other proxies such as net primary productivity may be more informative (e.g., Barlow et al. 2008b, Leduc et al. 2014), but are analytically complex (Behrenfeld & Falkowski 1997).
Finally, the modelling exercise presented here did not account for other external variables that influence sperm whale abundance. For example, the Kaikōura earthquake (November
2016) may have contributed to the particularly low abundance of summer 2016/17 (Guerra et al. 2018). In addition, because sperm whales range so widely, they are undoubtedly influenced by environmental conditions in regions other than Kaikōura. While it was not Chapter 5 – Time-series analysis 149
possible to incorporate these aspects in the analysis, they may account for part of the unexplained variance in the models.
5.4.5. Knowledge gaps and future research
The key constraints to understanding the biological mechanisms behind the observed relationships are a lack of knowledge on: (1) prey response to the environment, and (2) hydrology in the Kaikōura Canyon and wider Kaikōura/Mernoo Saddle region. For example, very little is known about the habitat preferences of warty and arrow squid, their seasonal movements, or how they respond to environmental variability. Warty squid are an important species for sperm whales and a key mid-trophic level species in New Zealand and the Southern Ocean (Jackson et al. 1998, 2000, Phillips et al. 2001, Rodhouse 2013). Better knowledge of their ecology is needed to understand trophic links in the Kaikōura Canyon and how squid and their predators might respond to a changing climate. As for the oceanographic regime off Kaikōura, multi-scale inter-disciplinary studies (e.g., Allen et al.
2001, Bosley et al. 2004) are needed to better understand circulation patterns in the canyon and connectivity with the hydrology in the wider region.
Long-term monitoring studies are crucial for assessing drivers of population change. Given the ongoing decline in the number of sperm whales foraging at Kaikōura, it is crucial to continue population monitoring. One way to advance our understanding of how sperm whales respond to inter-annual variability in ocean conditions is to examine long-term correlations with foraging patterns. For this purpose, acoustic recordings and sighting locations available for 1990-2017 (MMRG) could be used to quantify temporal variability in echolocation behaviour and spatial distribution (e.g., Pirotta et al. 2014, Brough et al. 2018).
5.4.6. Conclusions
The long-term variability in summer abundance of sperm whales at Kaikōura is partially correlated with fluctuations in ocean conditions. Lower proportions of SAW crossing the
Mernoo Saddle in summer and higher sea surface temperatures at Kaikōura in winter were associated with fewer whales foraging in the area. Although no causation was established from this exploratory analysis, the observed correlations are probably mediated through indirect effects on prey availability. The majority of proposed mechanisms involve processes that aggregate prey, and delayed effects of temperature on squid. The long-term increase of Chapter 5 – Time-series analysis 150
sea temperatures was identified as a potential factor contributing to the decline in sperm whales in the region, possibly through changes in the food web. This study identified associations between environmental conditions and sperm whales at time lags of 0 to 12 months, suggesting the importance of both contemporary processes and ecological succession in facilitating prey aggregations (Soldevilla et al. 2011). The results also highlight the value of accounting for time lags when modelling the occurrence of top predators. This research serves as an impetus to address important knowledge gaps in the hydrology of
Kaikōura and the ecology of sperm whale prey. A better knowledge of ecosystem structure and food web interactions will help to better understand and predict how sperm whales respond to climatic variability. The Kaikōura Canyon is a unique ecosystem with huge biodiversity value; it is imperative to understand its susceptibility to climate change in order to predict the response to changing conditions, and attempt to mitigate negative impacts.
Chapter 6.
Synthesis and conclusions.
The main motivation for this thesis was the unexplained decline in the number of sperm whales feeding in the Kaikōura Canyon over the last three decades. I sought to better understand the foraging ecology of the population in order to identify potential causes of the decline – a first step in the development of mitigation measures. The research provided new insights into the whales’ food and habitat requirements, and how these vary between summer and winter. This information was used to investigate how changes in the whales’ environment may have led to a decreased use of the Kaikōura feeding ground over time. In a wider context, this research contributes to our understanding of the ecological relationships between submarine canyons and top predators.
6.1. Summary of findings
In Chapters 2 and 3 I investigated the sperm whales’ food web using stable isotope analyses.
Stable isotope ratios in sloughed skin varied among individuals according to their sighting rates, probably reflecting the extent to which they used Kaikōura for foraging. Occasional visitors had lower and more diverse isotope ratios than frequent visitors, likely reflecting a range of food webs from areas beyond the canyon. I used stable isotope signatures of whales with higher sighting frequencies (i.e., those more likely to reflect a Kaikōura-based diet) to investigate the primary organic sources sustaining their food web. While the precise contributions of pelagic productivity vs coastal macroalgae could not be determined, there was strong support for a food web sustained mostly by pelagic production, with a possible input from microbially-recycled material in addition to phytoplankton. Differences in the whales’ isotope ratios between summer and winter suggested variability in diet. In contrast to the cephalopod-based diet recorded in most populations, the diet at Kaikōura appeared to be a mixture of demersal fish in addition to squid. This may reflect the extremely high benthic productivity of the canyon (De Leo et al. 2010, Leduc et al. 2014). Chapter 6 – Synthesis and conclusions 152
In Chapter 4, I investigated what characteristics make good foraging habitat. Species distribution models were used to identify topographic and oceanographic features that were correlated with the whales’ habitat use at a small spatial scale. Physical factors that contribute to aggregating prey (such as steep slopes, slope-current interactions and strong thermal stratification) were particularly important. Habitat preferences differed between summer and winter, and were consistent with temporal patterns in foraging behaviour (Jaquet et al.
2000) and stable isotope ratios (Chapters 2 and 3). In Chapter 5, the long-term variability in abundance of sperm whales in relation to oceanographic conditions was assessed for the first time. While further study is needed to identify causative ecological relationships, the results suggested that whale abundance in summer is correlated with winter sea temperatures and influx of Subantarctic water through the Mernoo Saddle during summer. I hypothesised mechanisms to explain these correlations, mostly through indirect effects on prey. The long- term increase in winter temperatures at Kaikōura was identified as a possible factor contributing to the decline.
6.2. Ecological implications
6.2.1. Sperm whales have seasonal patterns in habitat use
Based on differences in distribution and diving behaviour between summer and winter,
Jaquet et al. (2000) hypothesised that sperm whales off Kaikōura change their feeding habits in response to fluctuations in prey availability. My research supports that hypothesis and provides insights into what the ecological drivers might be. The seasonal changes in isotope ratios and habitat preferences were consistent with an increase in the contribution of squid to their diet in summer, and an increase in demersal fish in winter. While adaptability in habitat use increases resilience to changes in food availability (Whitehead et al. 2008), it may also be reflecting variations in food web stability. Whale abundance in summer is highly variable from year to year (Somerford 2018), suggesting that prey targeted in this season may not be a very stable resource. Squid, due to their fast growth and short lives, are particularly sensitive to environmental conditions, and can exhibit large variations in biomass from year to year (Rodhouse 2001, Pecl & Jackson 2008). A higher reliance on squid, which are a highly variable food resource, could result in years with very low whale abundance as a consequence of low food supply (e.g., Jaquet et al. 2003). Chapter 6 – Synthesis and conclusions 153
Figure 6.1. Schematic representation of habitat use by sperm whales foraging at Kaikōura, as suggested by the results from this thesis. Information on sperm whale abundance is derived from
Somerford (2018). Chapter 6 – Synthesis and conclusions 154
Temporal patterns in the distribution of top predators can influence the strength of trophic interactions in their ecosystem (Andrews & Harvey 2013). Differences in habitat use and abundance of sperm whales are therefore likely to influence their role in the food web in each season. For example, varying rates of feeding on demersal vs pelagic prey, or in the
Kaikōura Canyon vs the Conway Trough, would clearly exert different pressures on different faunal communities, which may have important implications for the deep-sea ecosystem at Kaikōura.
6.2.2. The number of sperm whales at Kaikōura is declining
While the reasons for the decline in sperm whales remain uncertain, it is likely to reflect a shift in distribution away from Kaikōura. Given the large movements of sperm whales, and that the study area is only a small part of their range, it is possible that reduced abundance at Kaikōura simply reflects a shift to other, perhaps more productive, feeding grounds. So why is the decline relevant? There are three main reasons. Firstly, the local decline could be a symptom of a more widespread phenomenon impacting food availability, with consequences for population fitness. Secondly, the local decline of an apex predator is likely to have consequences for the canyon ecosystem. Thirdly, sperm whales are a key attraction for tourism at Kaikōura, and whale-watching has played a major role in the economic and social development of the region (Curtin 2003).
Given that sperm whales off Kaikōura comprise the only monitored population in New
Zealand, it would be surprising if this was the only foraging ground experiencing a decline in whales or food resources, especially considering the global effects of climate change on cetacean distributions (Simmonds & Elliot 2009, Hazen et al. 2013b). Prey responses to changing oceanographic conditions may result in shifted or more variable distributions
(Perry et al. 2005, Rodhouse 2008). Prey distributions that are less predictable likely increase the whales’ travelling time while searching for feeding areas, reducing time spent foraging.
In addition, the whales may not adapt so easily to new patterns in prey abundance. As a species, sperm whales are highly mobile and have diverse feeding strategies, however, clans and individuals are thought to develop particular foraging habits (Whitehead & Rendell
2004). For example, clans tend to maintain their ‘traditional’ movement patterns in spite of changing environmental conditions, which can cause them to have lower feeding success Chapter 6 – Synthesis and conclusions 155
(Whitehead & Rendell 2004). This suggests that the whales have limited ability to adapt, and that changes in prey distribution could be costly.
The loss of top predators from marine communities can influence their structure and function through top-down interactions (Estes & Duggins 1995, Heithaus et al. 2008, Markel
& Shurin 2015). Although their natural abundance is low, sperm whales have large energy requirements and forage at a high trophic level, on a variety of species, and in different habitats throughout the whole water column. Therefore, they are likely to play an important role in regulating the canyon’s food web. It is possible that the decrease in their numbers could change the dynamics of prey populations, with potential consequences for community structure. In addition, by feeding at depth and releasing fecal plumes close to the surface
(Kooyman et al. 1981, Whitehead 1996), whales act as an ‘upward biological pump’ (Lavery et al. 2010, Roman & McCarthy 2010). Nutrients are recycled from the deep and brought to the photic zone, where they become available to phytoplankton and bacteria (Roman &
McCarthy 2010, Roman et al. 2014). The recycling of iron, a limiting micro-nutrient, has been considered of particular importance due to its substantial stimulation of carbon production by phytoplankton (Lavery et al. 2010, 2014, Smith et al. 2013). This process may be an important contribution to the canyon’s productivity.
6.2.3. Submarine canyons as foraging hotspots for sperm whales
This study expanded our knowledge on how canyons enhance food availability to deep- diving top predators. Sperm whales are specialist deep-water hunters. As air-breathers, they must balance access to food at depth and oxygen at the surface, which results in regular vertical migrations that reduce the time spent foraging (Kramer 1988). Habitats that aggregate prey over small spatial scales and provide reliable food resources must be advantageous for foraging efficiency. In the Kaikōura Canyon, high levels of phytoplankton production (Leduc et al. 2014), possibly enhanced by microbial recycling, appear to sustain high pelagic productivity in the ecosystem (Chapter 3). The steep topography of the canyon may help to concentrate pelagic organic matter sinking to the seafloor, potentially supplying benthic communities with enhanced food resources. In addition, steep slopes appear to aggregate mid-trophic-level consumers, and indirectly concentrate organisms through the interaction with local currents (Chapter 4). These processes probably help sustain a rich and Chapter 6 – Synthesis and conclusions 156
complex food web, which sperm whales tap into. Lastly, the potential role of distant oceanographic features in determining local prey abundance (such as the SAW wisp through the Mernoo Saddle; Chapter 5) highlights the importance of large-scale oceanographic context in driving canyon productivity.
Submarine canyons can experience strong seasonal variations in their faunal communities, driven partly by changes in thermal stratification, influx of productivity to the seafloor, and the particular life cycles of the species that inhabit them (Papiol et al. 2012, 2013, De Leo et al. 2018). Seasonal variability in ecosystem structure is certainly the case in the Kaikōura
Canyon. This appears to provide a temporally dynamic range of foraging opportunities for sperm whales, but also food resources that are sensitive to different natural and anthropogenic pressures.
6.3. Implications for management
The protection of species with wide ranges is especially challenging. The geographic extent of the population of sperm whales visiting Kaikōura is unknown, making it difficult to define management areas and identify potential threats. This lack of knowledge should not preclude protection, even if Kaikōura is only a small part of the whales’ range. In fact, some individuals spend months at a time there, suggesting that the area may be core habitat for at least part of the population. Given that whaling of sperm whales has ceased, long-term preservation of their populations depends partly on the health of prey populations and integrity of important habitat. This study identifies oceanographic variability as a potential factor contributing to the decline, and highlights the importance of demersal fish, in addition to squid, in the whales’ diet. While climate-driven changes in oceanographic conditions cannot be managed at a local or short-term scale, it is possible to manage other pressures to increase the resilience of marine mammal populations (Simmonds & Elliot 2009). I therefore recommend the application of a precautionary approach to managing other potential impacts, especially those that could compromise prey, foraging, or important habitat.
The main local anthropogenic pressures are probably fishing, whale-watching tourism (boat and aircraft), and noise pollution from boats and shipping. Commercial fisheries target species that are important prey (e.g., hāpuku, ling) and pose a mortality risk to sperm whales Chapter 6 – Synthesis and conclusions 157
through entanglement in fishing gear. Whale-watching tourism targets the whales directly, and is more intensive in summer than winter. A moratorium on new whale watching permits was established in 2002 by the Department of Conservation and is currently in place. Studies on the short-term effects of tourism on sperm whales at Kaikōura have detected changes in behaviour, which have been considered to be of little biological significance (Richter et al.
2006, Markowitz et al. 2011). However, individuals that may have left Kaikōura due to disturbance are not there to be studied, so the above research has inevitably focussed on the more ‘tolerant’ individuals. Additionally, the cumulative effect of multiple and repeated interactions with the same individual – a common occurrence when whale densities are low
– is not easily assessed. Lastly, noise pollution in general can cause displacement of cetaceans and reduce foraging efficiency by interfering with echolocation (Bowles et al. 1994, Aguilar
Soto et al. 2006, Forney et al. 2017). Given these potential impacts, I suggest the following management recommendations:
(1) Regarding whale-watching tourism:
- Continuation of the moratorium on whale watching permits for boat and aircraft
tours, to prevent an increase in the impact on diving and foraging behaviour.
- Establishment of measures to avoid multiple consecutive encounters with the same
individual, especially in spring and summer. Voluntary guidelines could be
developed in conjunction with tour operators (e.g., Whale Watch Kaikoura) and
management groups (e.g., DOC, Te Korowai).
- Investigation of the long-term effects of tourism on sperm whales. Given that the
decline in abundance appears to be occurring during summer, it is important to
investigate whether the increase in tourism activity is contributing to a displacement
of whales away from the area. This could be done by exploring the correlation
between tourism intensity (e.g., number of trips per month, extracted from vessel
logs) vs relative abundance and/or spatial distribution of whales.
(2) Regarding commercial fishing:
- The Hikurangi Marine Reserve provides little meaningful protection for sperm
whales, with considerable overlap between their distribution and areas where fishing Chapter 6 – Synthesis and conclusions 158
occurs (Chapter 4). Increasing the reserve’s size to cover more of their habitat would
be beneficial for reducing entanglement risk and possibly competition for prey.
- Increased observer coverage and more stringent reporting to produce reliable access
to fisheries data and better understand the distribution and seasonality of targeted
and by-caught species.
(3) Regarding noise pollution:
- Whale Watch Kaikoura, the only boat-based operation targeting sperm whales,
currently uses waterjet-propelled vessels, which are remarkably quiet underwater
(Buckstaff 2004, Markowitz et al. 2011). This strategy for minimising noise should be
maintained for future boat upgrades. It would be beneficial if this option was
considered by other companies that work in the canyon area (e.g., dolphin-watching
operations).
- Re-evaluation of the Kaikōura Marine Management Area could include shifting of
shipping lanes outside of the Whale Sanctuary (see Fig. 1.1. in Chapter 1) to reduce
exposure to shipping noise and collision risk.
Although sperm whales have large ranges and the study area is only a small part of their range, this research has shown that submarine canyons provide key foraging habitat for the species. Given that that the Kaikōura Canyon is a particularly productive submarine habitat
(De Leo et al. 2010), its value for the population is likely to be substantial in spite of its small size. Areas shown to be preferred habitat (Chapter 4) should be protected from impacts affecting prey, food web structure and the whales’ physical habitat (i.e., fishing, vessel noise, ship traffic). Specifically, the Kaikōura Canyon, Conway Trough and offshore gullies along the 1000 m contour line, should be areas of high priority for protection. The potential effects from activities affecting the whales’ foraging behaviour (i.e., anthropogenic noise, tourism) should be managed to minimise impact on the population. Overall, increasing the protection of sperm whales off Kaikōura would help to ensure their long-term survival in the area.
Achieving this objective would safeguard their ecological role as top predators, and their cultural and economic value for the local community and tourism industry.
Chapter 6 – Synthesis and conclusions 159
6.4. Moving beyond this thesis: future research
This research highlighted some important considerations for the study of foraging ecology of marine top predators in general. The use of long-term monitoring data was key to understand inter-annual variability in habitat use, which is essential in the study of long- lived species, a typical trait of top predators. The identification of individuals (through photography of natural marks) was extremely valuable in every analysis included in this thesis. It was particularly useful to link stable isotope information to population parameters, improve precision of estimates by pooling individual information (e.g., body size, residency patterns, isotope values), account for auto-correlation of repeated observations, and enable the estimation of abundance and densities of whales foraging in the area. In addition, it was advantageous to consider the studied system from a “wide-angle” perspective: sampling throughout the year rather than in one season, considering water-column characteristics rather than surface only, sampling across a wide range of trophic levels, and analysing foraging ecology at a range of temporal scales (i.e., from within-season isotopic measurements to inter-decadal patterns in habitat use). Lastly, this study emphasises the value of non-invasive research to examine foraging ecology of top predators, and the flexibility of small-boat surveys, which achieve substantial research time at lower costs.
This work has expanded our understanding of the ecology of sperm whales. As with all research, however, new questions have evolved and significant knowledge gaps remain.
Firstly, there is considerable scope to advance the study of the whales’ food web and foraging habits at Kaikōura. Information on particular prey species and their seasonal contribution to diet remains uncertain, and could be further investigated with more specific biochemical tools and more extensive sampling of prey. Sperm whales forage pelagically and demersally
(Miller et al. 2013a, Guerra et al. 2017), however it is still uncertain how these modes relate to different areas, habitats or prey. Diving depths are thought to correlate, at least partly, with prey aggregations within the deep scattering layer and benthic boundary layer (Guerra et al. 2017). Research on the spatial distribution and seasonal variability in demersal vs pelagic foraging would help to better understand the functional drivers of habitat selection.
A key knowledge gap is how little we know about sperm whale prey, particularly warty squid, arrow squid, ling and hāpuku. Further research on the diet, distribution and Chapter 6 – Synthesis and conclusions 160
movements of these species would help us understand how and why their abundance at
Kaikōura varies seasonally. This is not only relevant for better understanding habitat use by whales, but also the ecology and food web of the canyon. Studies of the whales’ foraging behaviour would be particularly useful if they were concurrent with prey sampling through hydro-acoustic surveys of their distribution and abundance (e.g., Benoit-Bird et al. 2004,
2008, 2013, Lawrence et al. 2016). Incorporating real-time prey fields into species distribution models could greatly expand our understanding of habitat use. In addition, relating differences in stable isotope ratios among individuals to differences in foraging distribution may help to link diet with habitat in a more specific way.
Due to the expense of boat-based surveys and the availability of boats, research carried out by the Marine Mammal Research Group has consistently focussed on two field seasons per year (typically November-January and June-July). Given the decline in whale numbers, and that spring (September-October) seems to have become a particularly scarce period, it may be instructive to extend the research. Autonomous acoustic recorders could be deployed to record the presence and relative abundance of whales throughout the year (e.g. Wong &
Whitehead 2014, Rayment et al. 2018). An important challenge in this would be developing a reliable mooring system in an area that is regularly fished.
I also recommend further study of the ecological significance of thermal boundary layers in spring and summer, including water column stratification, oceanic fronts, and the SAW wisp through the Mernoo Saddle. Strengthening of stratification throughout November appeared to be associated with an increase in the number of sperm whales in the study area (Chapter
4). This timing also coincides with an increase in the SAW wisp through the Mernoo Saddle
(Shaw & Vennell 2000), an oceanographic feature partly correlated with whale abundance
(Chapter 5). Investigating variability in these features in relation to prey aggregations will help reveal their potential importance for the canyon’s food web. In particular, changes in the timing of these processes could affect predator-prey relationships by decoupling important synchronies in food-web dynamics (Sims et al. 2001, Edwards & Richardson 2004).
A key question concerning the Kaikōura Canyon is what drives its extremely high productivity (De Leo et al. 2010, Leduc et al. 2014). Further study on the region’s oceanography and food web structure would advance our understanding of this dynamic Chapter 6 – Synthesis and conclusions 161
ecosystem. Of particular interest are potential trophic links for benthic-pelagic coupling and the role of the microbial loop in recycling carbon in the food web.
Kaikōura is one of the few places in the world where sperm whales can be found reliably so close to shore. Not surprisingly, most of what is known about this species in New Zealand comes from this location. Individuals move in and out of the study area, however, and some whales are not sighted for consecutive seasons (Childerhouse et al. 1995, Jaquet et al. 2000).
Importantly, nothing is known about recruitment: from where, and from which clans, do the subadult males that visit Kaikōura come from? Given the local decline in numbers, it is important to address two questions: (1) where are the whales when they are not at Kaikōura? and (2) what is the state of other populations around New Zealand? The movements of sperm whales can be tracked with satellite tags (Block et al. 2011, Mate et al. 2016), however there are two important drawbacks to this approach. Firstly, current tag technologies pose a risk to the animals, as their attachment system can cause significant damage through tissue loss and muscular trauma (Moore & Zerbini 2017). In addition, tags typically remain attached for periods of weeks to a few months (e.g., Mate et al. 2016) and have unknown but probably important effects on behaviour. This limits the value of the data, as the tracks are unlikely to provide robust information on what areas are important habitat, and much of the tracking may occur while individuals remain at Kaikōura. Until less invasive tags are developed costs to animal welfare probably outweigh the benefits of the data. An alternative approach is to carry out large-scale passive acoustic and photo-identification surveys (e.g.,
Whitehead et al. 2008, Pirotta et al. 2011), for example via transects across the 1000 m depth contour starting from Kaikōura. Such surveys would be expensive and time consuming.
Perhaps for now, the most practical way to learn about sperm whales beyond Kaikōura is to study other accessible populations. Sperm whales have been found consistently at the Otago
Canyons during surveys in 2016-2017 (Rayment, unpublished data). Assessing the status of this newly described population, its diet, habitat use and potential connectivity with the whales of Kaikōura would significantly expand our understanding of the species in New
Zealand. Lastly, it is important to continue to check for potential matches of ID photographs between the Kaikōura Catalogue and other whale sightings or strandings from around New
Zealand and surrounding regions. Chapter 6 – Synthesis and conclusions 162
6.5. Concluding remarks
This study has expanded our understanding of the ecology of sperm whales at Kaikōura, and draws attention to the rapid change that marine ecosystems are experiencing in modern times (Hoegh-Guldberg & Bruno 2010). The Kaikōura Canyon hosts a rich and productive ecosystem that is spatially and temporally diverse. The area is a foraging hotspot for sperm whales, apex predators which modify their habitat use and diet according to fluctuations in food supply. The long-term decline in abundance of whales signals that change is occurring within this unique system. Increasing our knowledge on how the whales use their environment, and how their habitat has changed over the last three decades will help to guide conservation of this population. This will help preserve its ecological role in the canyon and maintain its cultural and economic value to Kaikōura.
This study highlights the value of long-term monitoring studies in understanding change within populations of long-lived species. It also shows that much can be learnt about the ecology of marine mammals through the use of non-invasive methods. The ecology of deep- diving odontocetes and their relationship with submarine canyons poses many questions yet to be unfolded. The remoteness and difficult access to these intriguing species and fascinating habitats mean that our scientific knowledge is limited. An increasing body of evidence indicates that they are far from immune to anthropogenic pressures (Fernandez-
Arcaya et al. 2017). A robust understanding of their ecology and susceptibility to impacts is the first step to ensure their protection and future wellbeing. Appendix 1 163
Appendix 1. Supporting figures and tables for Chapters 2 and 3.
Figure A1.1. Diagnostic plots for the best GAMMs used to model the response of isotope ratios of sperm whales at Kaikōura. (a) δ¹³C and (b) δ¹⁵N.
(a)
(b)
Appendix 1 164
Table A1.1. Sources of error in the isotopic values of samples collected at Kaikōura. (1) Effect of lipid extraction (lipid free – raw) on the isotopic values and C:N ratios of animal samples; (2) Measured 2 duplicate error (√(subsamplea − subsampleb) ) from two aliquots of the same sample. All values are means ± 1 SD. Note that the change in δ¹³C after lipid extraction is expected, while the change in δ¹⁵N is a source of error as a secondary effect of the lipid extraction process.
1. Effect of lipid extraction Sample type n Δ δ¹⁵N (‰) Δ δ¹³C (‰) Δ C:N
Sperm whale skin 34 ±0.16 ± 0.14 +0.38 ± 0.20 -0.26 ± 0.12
Squid muscle 27 ±0.51 ± 0.36 +1.63 ± 0.66 -0.56 ± 0.22
Fish muscle 40 ±0.27 ± 0.13 +0.57 ± 0.61 -0.28 ± 0.41
2. Error measured between duplicate samples Sample type n Δ δ¹⁵N (‰) Δ δ¹³C (‰) Δ C:N
Sperm whale skin 28 0.10 ± 0.10 0.05 ± 0.05 0.03 ± 0.02
Squid muscle 8 0.16 ± 0.19 0.12 ± 0.12 0.07 ± 0.12
Fish muscle 8 0.10 ± 0.09 0.05 ± 0.05 0.02 ± 0.02
Macroalgae 7 0.35 ± 0.30 0.05 ± 0.03 0.28 ± 0.28
ALL 51 0.14 ± 0.17 0.06 ± 0.07 0.07 ± 0.14
Figure A1.2. Effect of lipid extraction via ASE (accelerated solvent extraction) on (a) δ¹⁵N and (b) δ¹³C of sloughed sperm whale skin. The black line represents a simple linear regression, and the dashed grey line represents a 1:1 relationship. Sample size = 34.
(a) (b) 17 -16
y = 0.96x + 0.66 y = 1.03x + 0.93 16 R² = 0.96 R² = 0.90
-17 15
14 ) ) after lipidextraction
) ) after extraction lipid -18
‰
‰ C C (
N ( 13
δ¹³ δ¹⁵
12 -19 12 13 14 15 16 17 -19 -18 -17 -16 untreated δ¹⁵N (‰) untreated δ¹³C (‰)
Appendix 1 165
Figure A1.3. Variation in δ¹³C of sperm whales at Kaikōura relative to year, month, sighting frequency, body length and sample skin type. Data points represent the raw data used in the GAMMs. Sample size = 90.
-15 -15
-16 -16
)
)
‰
‰ (
-17 ( -17
C
C
³
³
δ¹ δ¹
-18 -18
winter summer -19 -19 0 1 2 3 4 5 0 1 2 3 4 5 6 7 8 9 10 11 12 13 year month
-15 -15
-16 -16
) )
‰ ‰ (
-17 ( -17
C C
³ ³
δ¹ δ¹
-18 -18
-19 -19 0.00 0.05 0.10 0.15 0.20 0.25 0.30 12.5 13.5 14.5 15.5 16.5 sighting frequency body length (m)
-15
-16
) ‰
( -17
C
³ δ¹
-18
-19 1 type1 1 2 type2 2 3 skin type
Appendix 1 166
Figure A1.4. Variation in δ¹⁵N of sperm whales at Kaikōura relative to year, month, sighting frequency, body length and sample skin type. Data points represent the raw data used in the GAMMs. Sample size = 90.
18 18
17 17
16 16
) )
‰ 15 15
‰
N ( N (
δ¹⁵ 14 14 δ¹⁵
13 13
12 12 winter summer 11 11 0 1 2 3 4 5 0 1 2 3 4 5 6 7 8 9 10 11 12 13 year month
18 18
17 17
16 16
) )
15 15
‰ ‰
N ( N (
14 14
δ¹⁵ δ¹⁵
13 13
12 12
11 11 0.00 0.05 0.10 0.15 0.20 0.25 0.30 12.5 13.5 14.5 15.5 16.5 sighting frequency body length (m)
18
17
16 )
15
‰ N (
14 δ¹⁵
13
12
11 1 type1 1 2 type2 2 3 skin type
Appendix 1 167
Figure A1.5. Frequency distribution of the mean sighting frequencies of sperm whales at Kaikōura (2014-2017). The dashed line indicates a mean sighting frequency of 0.06, equivalent to an average of two sightings per field season after correcting for field effort (i.e., number of days of effort). This reference value was used as a cut-off to exclude whales with low sighting frequencies from the analysis in Chapter 3. Sample size = 37.
10
8
6
4
2 number of whales of number
0
0.06 0.08 0.00 0.01 0.02 0.03 0.04 0.05 0.07 0.09 0.10 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.20 0.21 0.22 0.23 0.24 0.25 0.26 0.27 0.28
sighting frequency
Appendix 1 168
Table A1.2. Contents of pelagic SPOM collected at Kaikōura from 5m depths. Values are expressed as percentages over the total number or area of organic components. Sample size = 6.
By number (%) By area (%)
Unicellular plankton 91 (85-94) 83 (77-91)
Organic matter particles 9 (5-14) 17 (8-22)
Diatoms 41 (17-57) 43 (10-54)
Dinoflagellates 27 (15-36) 13 (9-21)
Ciliates 13 (9-19) 20 (11-27)
Unidentified plankton cells 10 (4-21) 7 (3-17)
Figure A1.6. Plankton in SPOM from water samples collected at Kaikōura. Examples of oligotrichid ciliates (a,b), tintinnid ciliates (c,d), and diatoms (e,f).
Appendix 1 169
Figure A1.7. Isotopic signatures of surface SPOM (sample size = 25) and macroalgae samples (sample size = 51) collected in summer and winter.
12
10
8
6 N (‰)
δ¹⁵ 4
2 SPOM summer SPOM winter 0 Kelp summer Kelp winter -2 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8
δ¹³C (‰)
Appendix 1 170
Figure A1.8. Isotopic signatures of primary organic sources at Kaikōura. (a) Macroalgae signatures according to site of collection. (b) SPOM signatures according to site of collection. (c) Macroalgae signatures according to species. Error bars are 95% confidence intervals. ‘n’ = sample size.
(a) (b) 11 7
10 Macroalgae 6 SPOM sites sites Surface SPOM: 9 Site 1 5 Upper canyon
Conway trough
) )
Site 2 Offshore
‰ ‰
8 4 N ( N
Site 3 ( N 300m depth SPOM: δ¹⁵ δ¹⁵ Upper canyon 7 3
6 2 n = 51 n = 32 5 1 -27 -26 -25 -24 -23 -20 -18 -16 -14 -12 δ¹³C (‰) δ¹³C (‰)
(c) 11
Macroalgae species 10 Carpophyllum maschalocarpum 9
Durvillaea willana
) ‰
8 Ecklonia radiata N ( N
δ¹⁵ Landsburgia 7 quercifolia
Marginariella 6 boryana n = 51 Lessonia variegata 5 -22 -20 -18 -16 -14 -12 -10 δ¹³C (‰)
Appendix 2 171
Appendix 2. Supporting figures and tables for Chapter 4.
Figure A2.1. Correlation between values of temperature, salinity and fluorescence measured by the CTD SBE-19 and RBR-concerto. The grey dashed line represents a 1:1 slope with 0 origin. The red line in the fluorescence plot is a linear regression, used for the correction equation to transform SBE-19 fluorescence values. Sample size = 994 (from 10 different casts).
13.0 C) ° 12.5
12.0 concerto ( - y = 0.99x + 0.06 11.5 R² = 1.00
11.0
Temperature Temperature RBR 10.5 10.5 11.0 11.5 12.0 12.5 13.0 Temperature SBE-19 (°C)
35.0
34.8
34.6
34.4
concerto (psu) y = 0.99x + 0.21 - R² = 1.00 34.2
34.0 SalinityRBR
33.8 33.8 34.0 34.2 34.4 34.6 34.8 35.0 Salinity SBE-19 (psu)
- 4.0
3.5 g/L g/L Chl μ 3.0
2.5 )
concerto ( 2.0 a - y = 1.20x + 0.11 1.5 R² = 0.92 1.0 0.5
0.0 Fluorescence Fluorescence RBR 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Fluorescence SBE-19 (μg/L Chl-a)
Appendix 2 172
Figure A2.2. Comparison of monthly mean chlorophyll-a values obtained from in situ CTD data and remotely-sensed satellite data. The satellite data are derived from NOAA’s Aqua-Modis NPP LS3MI 8-day composites at 4 km resolution.
1.80
1.60
1.40
1.20
1.00
0.80
chla (mg/m^3) 0.60
0.40 CTD 0.20 Satellite 0.00
summer winter
Appendix 2 173
Figure A2.3. Spatial variability in oceanographic factors included in the species-distribution models. The y-axes show the absolute differences between CTD casts carried out on the same day, while the x-axes show the increasing distance between CTD casts. The smoothed splines (and 95% confidence bands) are represented by the black broken lines. The vertical red line marks the 2.5 nmi limit within which the variability of each factor is considered to be small enough to assume that oceanographic conditions are similar. Sample size = 486. ‘T’ = temperature, ‘S’ = salinity, ‘chl-a’ = relative chlorophyll a concentration, ‘T 10m’ = T at 10 m depth, ‘S 10m’ = S at 10 m depth, ‘T350’ = T at 350 m depth, ‘S350’ = S at 350 m depth, ‘max chl-a’ = maximum chl-a, ‘depth max chl-a’ = depth of the maximum chl-a, ‘TC depth’ = depth of the thermocline, ‘TC strength’ = variance in T within thermocline.
Appendix 2 174
(Figure A2.3, continued)
Appendix 2 175
Figure A2.4. Effect of (a) swell and (b) sea state on the detection distance of sperm whales at Kaikōura. Error bars are 95% confidence intervals. Sample size = 175 (search stations from which at least one sperm whale was detected).
(a) (b) 5.0 5.0 4.5 4.5 4.0 4.0 3.5 3.5 3.0 3.0 2.5 2.5 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 Meandistance toclosest whale (nmi) 0.0 0.0 0.5 1.0 1.5 2.0 0 1 2 3
Swell height (m) Sea state (beaufort scale)
Appendix 2 176
Table A2.1. Model selection of winter GAMMs based on data from 2016 and 2017 (including deep-water oceanographic variables), used to explain presence of sperm whales foraging off Kaikōura. Metrics of model performance and significance of smooth terms are shown for models with Akaike weights ≥5%. SW = sperm whale presence/absence; T = temperature; S = salinity, TC = thermocline; edf = estimated degrees of freedom; ∆AICC = difference in AICC score relative to best model in set; wi = Akaike weight (i.e., model probability); adjusted R2 = explained variance; (1|subsite) = random effect included in all GAMMs. Candidate variables included: seafloor depth, slope, aspect, surface salinity, temperature at 10 m depth, chl-a maximum, depth of chl-a maximum, temperature at 350m depth, depth of thermocline and strength of thermocline.
Note that although the temperature at 350 m depth is selected by the highest-ranking models, the effect has low statistical significance. Similarly, although there is some support for the model including thermocline strength, the effect has low statistical significance, and its inclusion does not increase the explained variance by much (0.397 vs 0.393).
Model selection table:
Rank Model edf ∆AICc wi Adjusted R2
1 SW ~ depth + max chl-a + depth max chl-a + T 350m + (1|subsite) 10 0 0.25 0.393 2 SW ~ depth + max chl-a + depth max chl-a + T 350m + aspect + (1|subsite) 11 1.5 0.12 0.423 3 SW ~ depth + max chl-a + depth max chl-a + T 350m + TC strength + (1|subsite) 12 2.15 0.09 0.397
Significance of smooth terms retained in each model:
depth max chl-a depth max chl-a T 350m aspect TC strength Model edf (p-value) edf (p-value) edf (p-value) edf (p-value) edf (p-value) edf (p-value)
1 3.58 (p=0.03) 3.72 (p<0.001) 3.49 (p<0.001) 3.46 (p=0.12) - - 2 3.58 (p=0.04) 3.69 (p<0.001) 3.50 (p<0.001) 3.46 (p=0.16) 1.28 (p=0.15) - 3 3.48 (p=0.04) 3.68 (p=0.002) 3.39 (p<0.001) 3.54 (p=0.13) - 3.10 (p=0.13) Appendix 2 177
Figure A2.5. Effect of habitat variables on the presence of sperm whales foraging at Kaikōura during winter (based on data from 2016 and 2017 and including deep-water oceanographic variables), estimated with the highest-ranking generalised additive model. (a) Seafloor depth, (b) temperature at 350 m depth, (c) sub-surface chl-a maximum, (d) depth of sub-surface chl-a maximum. The y-axes show the smooth function of each variable, with the estimated degrees of freedom in brackets. Shaded areas represent 95% confidence intervals of the response. The rug plots represent the x-value of each data point.
Note that the effect of temperature at 350 m on sperm whale presence probability is mostly evident at temperatures below 9°C (with a predicted increase in presence for colder temperatures), but the sample size for this range is very small, resulting in high uncertainty of the effect.
Appendix 2 178
Figure A2.6. Correlograms of the residuals of the best fitting models in (a) summer and (b) winter, to examine spatial autocorrelation. The y-axis indicates the degree of correlation, with distance (in metres) on the x-axis. The shaded area represents 95% pointwise bootstrap confidence intervals.
(a)
(b)
Appendix 2 179
Figure A2.7. Logistic regression quantile-quantile plots used to check model adequacy for (a) summer GAMs and (b) winter GAMMs used in species-distribution models for sperm whales. The green lines indicate 95% confidence bands.
(a)
(b)
Appendix 2 180
Table A2.2. Metrics of model performance of the (a) summer GAMs and (b) winter GAMMs used to predict sperm whale habitat suitability. The tables show the results of the validation tests for the best model based on model selection for each pair of years (top three rows), and for all years (bottom row, for comparative purposes). TSS = true skill statistic; AUC = area under the receiver operating curve; adjusted R2 = explained variance. Note that the explained deviance cannot be calculated for GAMMs.
(a) SUMMER
Model selection Deviance TSS TSS TSS AUC AUC AUC Training data Testing data Best model Adj. R2 based on explained (%) 2015 2016 2017 2015 2016 2017 2016, 2017 2016, 2017 2015 depth + slope + Tgrad350 + surS + surC 23.7 0.239 0.44 na na 0.80 na na
2015, 2017 2015, 2017 2016 depth + slope + Tgrad350 + surS + S350 27.4 0.285 na 0.32 na na 0.76 na
2015, 2016 2015, 2016 2017 depth + slope + Tgrad350 + surS + S350 + aspect 31.6 0.317 na na 0.36 na na 0.76 all years 3x two-year pairs 3x remaining year depth + slope + Tgrad350 + surS + S350 + surC 26.2 0.284 0.49 0.44 0.34 0.82 0.77 0.76
(b) WINTER
Model selection TSS TSS TSS AUC AUC AUC Training data Testing data Best model Adj. R2 based on 2015 2016 2017 2015 2016 2017 2016, 2017 2016, 2017 2015 depth + Cmax_D + Cmax + (1|subsite) 0.362 0.15 na na 0.63 na na
2015, 2017 2015, 2017 2016 depth + Cmax_D + aspect + (1|subsite) 0.226 na 0.52 na na 0.75 na
2015, 2016 2015, 2016 2017 depth + Cmax_D + Cmax + aspect + (1|subsite) 0.328 na na 0.37 na na 0.71 all years 3x two-year pairs 3x remaining year depth + Cmax_D + Cmax + aspect + (1|subsite) 0.300 0.19 0.45 0.37 0.65 0.76 0.71
Appendix 2 181
Table A2.3. Concurvity tests to examine correlation among candidate explanatory variables used in the species-distribution models in (a) summer and (b) winter. High correlation values (concurvity index > 0.3) are highlighted in grey. The variance explained by each univariate model (Adj. R2) was used to select the variable with most explanatory power among correlated terms. The variables retained in the full model are marked with a * symbol. ‘T’ = temperature, ‘S’ = salinity, ‘TC’ = thermocline, ‘chl-a’ = Chlorophyll a concentration. For example, in summer, TC strength and the vertical temperature gradient between the surface and 350 m (T gradient 350m) are highly correlated, as indicated by the concurvity index of 0.58. Since T gradient 350m explains more variance, it is retained in the full model.
(a) SUMMER
Adj. R2 T T T S S S TC TC T gradient chl-a chl-a depth chl-a Variable depth slope aspect (%) surface 10m 350m surface 10m 350m depth strength 350m surface max max depth * 13.10 - 0.06 0.04 0.02 0.02 0.05 0.10 0.08 0.05 0.02 0.01 0.02 0.01 0.06 0.02 slope * 0.97 - 0.03 0.01 0.00 0.02 0.02 0.02 0.01 0.02 0.01 0.01 0.04 0.01 0.01 aspect * 1.14 - 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
T surface 9.78 - 0.82 0.08 0.05 0.04 0.11 0.06 0.42 0.80 0.10 0.04 0.04
T 10m 9.74 - 0.07 0.07 0.06 0.07 0.03 0.57 0.91 0.05 0.04 0.04
T 350m 0.03 - 0.09 0.07 0.87 0.04 0.05 0.09 0.01 0.01 0.02
S surface * 3.81 - 0.59 0.09 0.00 0.02 0.05 0.02 0.01 0.01
S 10m 1.05 - 0.08 0.01 0.02 0.04 0.01 0.01 0.03
S 350m * 0.24 - 0.05 0.09 0.15 0.02 0.01 0.02
TC depth * 1.13 - 0.07 0.04 0.02 0.02 0.35
TC strength 7.83 - 0.58 0.04 0.02 0.02
T gradient 350m * 10.40 - 0.06 0.05 0.06 chl-a surface * 2.12 - 0.04 0.08 chl-a max. * 0.06 - 0.02 depth chl-a max. 0.45 - Appendix 2 182
(Table A2.3, continued)
(b) WINTER
Adj R2 chl-a chl-a depth chl- Variable depth slope aspect T surface T 10 m S surface S 10 m (%) surface max. a max. depth * 16.20 - 0.10 0.07 0.05 0.04 0.12 0.14 0.01 0.04 0.08 slope * 2.39 - 0.07 0.04 0.04 0.03 0.05 0.02 0.03 0.01 aspect * 11.90 - 0.02 0.02 0.02 0.02 0.00 0.02 0.02
T surface 0.62 - 0.90 0.24 0.22 0.02 0.05 0.06
T 10 m * 1.40 - 0.20 0.21 0.01 0.04 0.04
S surface * 0.12 - 0.79 0.14 0.24 0.15
S 10 m 0.09 - 0.11 0.26 0.13
Chl-a surface 5.62 - 0.51 0.17
Chl-a max. * 7.13 - 0.07 depth Chl-a max. * 6.19 -
Appendix 2 183
Figure A2.8. Locations of search stations where acoustic-visual surveys for sperm whales started from. Locations are shown for (a) summer and (b) winter. Sample size = 183 (summer), 104 (winter).
(a)
Appendix 2 184
(Figure A2.8 continued)
(b)
Appendix 2 185
Figure A2.9. Temporal variability in (a) daily sperm whale density, (b) vertical temperature gradient (surface – 350 m), and (c) surface salinity at Kaikōura over spring/summer (20 Oct – 20 Jan), using data from 2015 – 2017. The raw values are represented by circles and the trend is represented by a cubic smoothed spline (with a maximum of 4 degrees of freedom). Sperm whale density (SW/nmi) is the number of unique whales sighted in a day, standardised by survey effort (distance travelled searching for whales, in nautical miles). Temporal variability in vertical temperature gradient and surface salinity are shown because these were the two oceanographic variables that were important predictors of foraging habitat in summer (i.e., had high relative importance indices) and had the potential to be temporally dynamic (as opposed to seafloor depth or slope). ‘n’ = sample size.
n = 95
n = 278
n = 278
Appendix 2 186
Figure A2.10. Iso-surface maps of (a) surface salinity (n = 896) and (b) vertical temperature gradient (surface – 350 m; n = 486) at Kaikōura during summer, based on CTD data pooled over three years (2015 – 2017). The panels show the data for the full summer (20 Oct – 20 Jan), as well as at 30-day window intervals. Spatial interpolation was carried out in Ocean Data View (Schlitzer 2001) using weighted-average gridding. Note that in (a), additional surface salinity data are included, collected via CTD casts at the surface (1 m depth) and which were not used in the modelling analysis or shown in Fig. A2.9.
Appendix 3 187
Appendix 3. Supporting figures and tables for Chapter 5.
Figure A3.1. Model diagnostics for the analysis of sperm whale abundance in relation to environmental variables for (a) LMMs in the monthly analysis and (b) GLMs in the summer analysis. Plots include: deviance residuals as a function of their theoretical quantiles (Q-Q plot), histogram of residuals, and residuals as a function of the fitted values, to check for homogeneity of variance.
(a)
(b)
Appendix 3 188
Figure A3.2. Temporal auto-correlation function (ACF) plots for best fitting model of (a) monthly analysis and (b) seasonal analysis. The dashed lines indicate 95% confidence intervals, beyond which temporal auto-correlation is considered statistically significant.
(a)
(b)
Appendix 3 189
Table A3.1. Summary of effort and monthly estimates of average relative abundance index (AI) of sperm whales at Kaikōura. All AI estimates are number of whales/25 nmi/day.
Effort Effort Mean AI Range AI Month Summer (days) (nmi) (SD) (min - max)
Nov 1994 1994/95 8 213 3.49 (1.52) 1.09 - 5.55
Dec 1994 1994/95 15 412 5.53 (2.08) 2.50 - 10.41
Jan 1995 1994/95 6 217 3.38 (2.35) 0.51 - 6.80
Nov 1996 1996/97 9 164 3.32 (0.74) 1.86 - 4.05
Dec 1996 1996/97 13 344 3.63 (2.47) 1.55 - 11.08
Jan 1997 1996/97 5 58 6.15 (2.30) 3.52 - 9.72
Nov 1997 1997/98 12 296 3.34 (1.38) 1.97 - 6.48
Dec 1997 1997/98 10 243 7.73 (2.84) 2.46 - 12.92
Jan 1998 1997/98 8 269 5.06 (1.73) 2.11 - 7.52
Nov 1998 1998/99 13 382 4.73 (1.53) 1.50 - 6.93
Dec 1998 1998/99 8 272 7.11 (3.39) 1.77 - 20.44
Nov 1999 1999/00 6 79 3.98 (1.78) 1.13 - 7.07
Dec 1999 1999/00 7 189 5.00 (1.16) 3.34 - 6.36
Nov 2000 2000/01 9 325 2.19 (1.22) 0.82 - 4.74
Dec 2000 2000/01 9 267 2.47 (0.79) 1.58 - 3.74
Jan 2001 2000/01 5 134 3.52 (2.36) 1.70 - 6.95
Dec 2005 2005/06 11 185 5.59 (2.73) 1.73 - 8.83
Nov 2006 2006/07 7 129 3.59 (1.59) 1.71 - 6.24
Dec 2006 2006/07 13 318 3.51 (1.11) 2.19 - 5.24
Jan 2008 2007/08 5 125 2.86 (2.32) 0.70 - 6.27
Jan 2014 2013/14 9 238 3.29 (1.87) 1.32 - 6.16
Nov 2014 2014/15 5 100 1.16 (1.12) 0.00 - 2.43
Dec 2014 2014/15 6 137 3.01 (1.58) 1.45 - 5.92
Nov 2015 2015/16 10 229 1.32 (1.21) 0.00 - 3.18
Dec 2015 2015/16 14 241 3.74 (1.31) 1.87 - 6.76
Nov 2016 2016/17 7 139 1.23 (1.29) 0.00 - 3.13
Dec 2016 2016/17 9 187 2.09 (1.53) 0.00 - 5.31
Jan 2017 2016/17 5 110 1.91 (0.72) 0.99 - 2.91
Nov 2017 2017/18 17 588 1.31 (1.12) 0.00 - 3.59
Dec 2017 2017/18 13 367 2.31 (1.03) 0.77 - 4.32 Appendix 4 190
Appendix 4
Publication: Diverse foraging strategies by a marine top predator: sperm whales exploit pelagic and demersal habitats in the Kaikōura submarine canyon. (2017). M. Guerra, L.
Hickmott, J. van der Hoop, W. Rayment, E. Leunissen, E. Slooten, M. Moore. Deep-Sea
Research Part I, 128: 98-108.
Deep-Sea Research Part I 128 (2017) 98–108
Contents lists available at ScienceDirect
Deep-Sea Research Part I
journal homepage: www.elsevier.com/locate/dsri
Diverse foraging strategies by a marine top predator: Sperm whales exploit MARK pelagic and demersal habitats in the Kaikōura submarine canyon ⁎ M. Guerraa, , L. Hickmottb,c, J. van der Hoopd, W. Raymenta, E. Leunissena, E. Slootena, M. Mooree a University of Otago, Dunedin, New Zealand b Scottish Oceans Institute, University of St Andrews, St Andrews, UK c Open Ocean Consulting, Petersfield, Hants, UK d Aarhus University, Aarhus, Denmark e Woods Hole Oceanographic Institution, Woods Hole, USA
ARTICLE INFO ABSTRACT
Keywords: The submarine canyon off Kaikōura (New Zealand) is an extremely productive deep-sea habitat, and an im- Submarine canyon portant foraging ground for male sperm whales (Physeter macrocephalus). We used high-resolution archival tags Sperm whale to study the diving behaviour of sperm whales, and used the echoes from their echolocation sounds to estimate Foraging their distance from the seafloor. Diving depths and distance above the seafloor were obtained for 28 dives from Kaikoura six individuals. Whales foraged at depths between 284 and 1433 m, targeting mesopelagic and demersal prey Echolocation layers. The majority of foraging buzzes occurred within one of three vertical strata: within 50 m of the seafloor, Demersal mid-water at depths of 700–900 m, and mid-water at depths of 400–600 m. Sperm whales sampled during this study performed more demersal foraging than that reported in any previous studies – including at Kaikōura in further inshore waters. This suggests that the extreme benthic productivity of the Kaikōura Canyon is reflected in the trophic preferences of these massive top predators. We found some evidence for circadian patterns in the foraging behaviour of sperm whales, which might be related to vertical movements of their prey following the deep scattering layer. We explored the ecological implications of the whales’ foraging preferences on their habitat use, highlighting the need for further research on how submarine canyons facilitate top predator hot- spots.
1. Introduction et al., 2012). Furthermore, the presence of many top-predators tar- geting mesopelagic prey (e.g., Benoit‐Bird et al., 2004; Boren et al., Submarine canyons are complex topographic features that cross 2006) suggests that the area hosts a highly productive pelagic system. continental margins all over the globe, connecting the shallow con- The Kaikōura Canyon is also an important year-round foraging tinental shelves to deep ocean basins (Shepard and Dill, 1966). They are ground for male sperm whales (Physeter macrocephalus)(Childerhouse extremely productive, serving as hotspots of benthic and pelagic bio- et al., 1995; Jaquet et al., 2000), deep-diving predators (Papastavrou mass and diversity (De Leo et al., 2010; Vetter et al., 2010; van Oevelen et al., 1989; Watkins et al., 1993) that use echolocation to detect and et al., 2011), and are key habitats for top predators, including deep- locate prey (Møhl et al., 2000; Madsen et al., 2002). Although squid are diving cetaceans (Yen et al., 2004; Moors-Murphy, 2014). Despite being their primary food source (Okutani and Nemoto, 1964; Rice, 1989; globally numerous, submarine canyons are poorly studied, and the Santos et al., 1999), demersal fish appear to be an important component drivers behind their exceptional productivity are not well understood of their diet in some regions (Martin and Clarke, 1986), including the (De Leo et al., 2010; Moors-Murphy, 2014). Kaikōura/Cook Strait region of New Zealand (Gaskin and Cawthorn, The Kaikōura Canyon, off the east coast of New Zealand (Fig. 1), has 1967). been described as the most productive non-chemosynthetic habitat re- Deep-diving predators exploit specialist niches below the photic corded to date in the deep sea (De Leo et al., 2010). It harbours ex- zone, where most biomass is concentrated in two vertical bands: the ceptional biomass of infaunal and epifaunal invertebrates, including deep scattering layer (DSL, Johnson, 1948) and the benthic boundary nematodes, and also benthic-feeding fish (De Leo et al., 2010; Leduc layer (BBL, Marshall, 1965). The DSL is typically between 200 and
⁎ Corresponding author. E-mail address: [email protected] (M. Guerra). http://dx.doi.org/10.1016/j.dsr.2017.08.012 Received 31 March 2017; Received in revised form 19 July 2017; Accepted 28 August 2017 Available online 31 August 2017 0967-0637/ © 2017 Elsevier Ltd. All rights reserved. M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
Fig. 1. The Kaikōura submarine canyon, New Zealand. The known surface positions of tagged whales are shown (black circles) together with the tag ID. Note that whales could not always be fol- lowed and therefore some of their positions were unknown. Depth contours show 500, 1000 and 1500 m isobaths.
600 m deep, and is composed of mesopelagic organisms, dominated by main objective was to investigate which prey layers are targeted by micronektonic fish (Barham, 1966; Hays, 2003), while the BBL com- sperm whales in the Kaikōura Canyon system. We also aimed to identify prises benthopelagic organisms that move freely on and just above the circadian patterns in their foraging behaviour, and compare foraging sea bed (Angel and Boxshall, 1990). Studies of diving and acoustic preferences with other populations and previous observations at Kai- behaviour can reveal which types of prey are targeted by deep-diving kōura. We use the findings from this study to better understand how odontocetes (Teloni et al., 2008; Arranz et al., 2011; Miller et al., 2013). sperm whales use the different vertical strata in their habitat in order to The underwater behaviour of sperm whales has been investigated using meet their high energy requirements. a variety of methods including implanted transponders (Watkins et al., 1993), archival tags attached via suction cups (Johnson and Tyack, 2003; Watwood et al., 2006; Fais et al., 2015), tags attached with darts 2. Methods or barbs (Davis et al., 2007; Mate et al., 2016), and passive acoustics (Thode et al., 2002; Wahlberg et al., 2001; Wahlberg, 2002; Miller 2.1. Field site and data collection et al., 2013). Foraging ecology of sperm whales at Kaikōura was first studied in Field work took place off the Kaikōura coast, New Zealand (42.5°S, 2007 using a three-dimensional (3-D) passive acoustic array (Miller and 173.8°E; Fig. 1), from February 19 to March 3, 2013 (austral late Dawson, 2009; Miller et al., 2013). The study revealed that whales summer). The study area covered water depths of 800–1600 m ex- foraged throughout the water column, with the majority of foraging tending over the main canyon (ca. 20 km from the canyon head) and buzzes occurring in mid-water at depths of 400–550 m. The Miller et al. adjacent secondary canyons (Fig. 1). (2013) dataset included 78 recordings of full or partial dive cycles, with Adult male sperm whales were tagged with high-resolution digital whale identity known for 42 recordings (12 whales). These data were archival tags (DTag2) to collect acoustic, depth and movement data. gathered over one year in the upper portion of the Kaikōura Canyon Whales were tracked acoustically while underwater using a hand-held and around the Conway Rise (see Fig. 1). The foraging behaviour of directional hydrophone (Dawson, 1990), then after surfacing ap- sperm whales in the deeper parts of the canyon, and areas further off- proached slowly from behind using a 6 m rigid-hulled inflatable boat. shore, has remained unstudied. The tag was attached with four suction cups to the dorsal surface of the The abundance of sperm whales feeding at Kaikōura has suffered a whale using an 8 m handheld pole. The tags included a hydrophone, a recent decline, from nearly 100 individuals in 1991 to half that number depth sensor and three-axis accelerometers and magnetometers in 2007 (van der Linde, 2009). With no evidence for direct impacts on (Johnson and Tyack, 2003). Depth and orientation of the whales was survival, it is possible that the decline has been driven by a change in sampled at 50 Hz, decimated to 5 Hz for analysis, and sounds were distribution away from Kaikōura, potentially reflecting underlying sampled at 96 kHz. When possible, tagged whales were followed vi- ecological changes that affect the availability or distribution of their sually from the 56 m M/V (motor vessel) Alucia or from the 10 m RHIB prey. It is therefore essential to better understand what sustains the diet (rigid-hull inflatable boat) Northwind (tender to Alucia). Whales were of sperm whales and the environmental features which drive their identified prior to tagging via photos of the unique marks (nicks and distribution. Given the extreme benthic biomass of the Kaikōura notches) along the trailing edge of their flukes, and matched to a photo- Canyon (De Leo et al., 2010; Leduc et al., 2012), and that demersal fish identification catalogue of sperm whales sighted at Kaikōura since are known to constitute an important part of the diet of sperm whales in 1990. This was done to ensure that the same animal was not tagged the region (Gaskin and Cawthorn, 1967), we hypothesise that sperm more than once. The tags released after a pre-programmed duration and whales at Kaikōura are likely to forage extensively near the seafloor. floated to the surface where they were located by VHF radio tracking. To address this hypothesis, we attached acoustic multi-sensor tags All tag data were analysed with custom scripts (M. Johnson, 〈https:// (‘Dtags’, Johnson and Tyack, 2003) to sperm whales at Kaikōura. Our www.soundtags.org/dtags/dtag-toolbox/〉) in Matlab R2013a (Math- Works).
99 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
To assess the behavioural reactions of sperm whales to tagging at- tempts, we recorded whether whales exhibited a startle response (e.g., a swipe of the flukes), performed a shallow dive (i.e., not lifting flukes), displayed evasive behaviour by turning and swimming away from the tagging boat, or performed an early dive. A dive was considered to be ‘early’ if the whale’s time at the surface was < 4.5 min, which is half of the mean surface time recorded for sperm whales at Kaikōura (Douglas et al. 2005).
2.2. Dive cycle and foraging activity
While diving, sperm whales regularly produce loud broadband echolocation clicks (known as ‘usual clicks’) to search for prey (Weilgart and Whitehead, 1988), as well as buzzes (or ‘creaks’) which are as- sumed to indicate prey capture attempts (Miller et al., 2004; Fais et al., 2015). Acoustic recordings were examined acoustically and visually using spectrograms (512 point FFT, Hanning window, 50% overlap) to locate the start and end of the clicking phase of each dive, and to identify buzzes. Buzzes were defined as a series of clicks with inter-click Fig. 3. Echogram showing sperm whale clicks and echoes reflected from the seafloor. intervals shorter than 0.22 s (Teloni et al., 2008; Fais et al., 2015). An Each row represents the 2 s envelope of a single outgoing click, with colour representing fi automatic click finder in Matlab (“findallclicks” by M. Johnson, sound pressure level. Echolocation clicks appear as well-de ned dark red blocks, while the seafloor echoes appear as thinner yellow segments. The manual selection of the first 〈https://www.soundtags.org/dtags/dtag-toolbox/〉) was used to detect five seafloor echoes is indicated by small black circles. In this example the whale is ap- individual clicks for posterior echo analysis. proaching the seafloor, with its distance above it decreasing from about 200–180 m To facilitate the interpretation of the whales’ foraging behaviour, within this 30 s interval. The saw-tooth pattern of subsequent clicks reflects a dynamic we divided each dive cycle into distinct phases (Fig. 2), based on the inter-click interval as the whale searches for prey at varying ranges. (For interpretation of dive profiles and foraging vocalisations (Watwood et al., 2006; Miller the references to color in this figure legend, the reader is referred to the web version of et al., 2013; Fais et al., 2015). The ‘descent’ phase started when the this article.) whale fluked up, and ended with the first decrease in whale depth; this phase is therefore characterised by a continuous downwards orienta- acoustically at Kaikōura in near-shore waters, we also used an alter- tion. The ‘bottom’ phase of the dive followed immediately after the native definition of ‘bottom’ phase which continued until the whale descent phase and continued until the time when the whale was last became silent on ascent, following Miller et al. (2013). oriented downwards. The ‘ascent’ phase was characterised by a con- tinuous upwards orientation, starting immediately after the end of the fl fl bottom phase, and finishing at the time of surfacing. The ‘search’ phase 2.3. Sea oor echoes and distance above the sea oor comprised the part of the dive when the whale was producing usual echolocation clicks. The ‘foraging’ phase of the dive was defined as the Dtags record the sounds emitted by sperm whales, but also any fl period from the first to last buzz, representing the time when sperm echoes returning from the sea surface, sea oor or organisms in the whales were actively encountering prey. For the specific purpose of water (Thode et al., 2002; Arranz et al., 2011; Fais et al., 2015). The fl ‘ comparing foraging behaviour to that of sperm whales tracked distance of a sperm whale above the sea oor (also known as whale altitude’, Arranz et al., 2011; Fais et al., 2015) was thus obtained from the echoes generated by the whale’s clicks as they reflected from the sea bed (Thode et al., 2002; Arranz et al., 2011; Fais et al., 2015). In es- sence, this method taps into the whales’ sonic perception of their sur- roundings, eavesdropping on their bio-sonar to collect information that the whales themselves receive from their physical environment. Click echoes were automatically identified using a supervised click and echo detector in Matlab (“d3echotool” by M. Johnson, 〈https:// www.soundtags.org/dtags/dtag-toolbox/〉), and echoes from the sea- floor were then manually selected using echograms (Fig. 3). Echograms were constructed by stacking the envelopes of high-pass filtered 2 s duration sound segments synchronised to each produced click (Johnson et al., 2004; Arranz et al., 2011; Fais et al., 2015). The high-pass filter was used to remove flow noise, with a cut-off frequency set to 2 kHz. These segments enabled detection of seafloor echoes at ranges up to 1490 m from the whale, covering the entire water column within the research area. Reflections from the sea floor appeared on the echograms as sequences of echoes with well-defined onset times and slowly varying delay times from the outgoing click. The two way travel time (TWT) of each echo from the whale to the closest seafloor surface and back to the DTag was estimated using a supervised edge detector (re- solution < 500 μs). Whale distance above the seafloor was calculated as Fig. 2. Phases of an example foraging dive. The time-depth profile is shown by a blue line half the TWT multiplied by the speed of sound underwater. An average −1 while the whale is silent and a black line while it is producing echolocation clicks. Buzzes speed of sound of 1490 m s was assumed based on the sound speed (assumed prey capture attempts) are marked by red dots. Vertical dashed lines define the profile measured with a CTD to a depth of 550 m in the same area. The start and end of each phase. (For interpretation of the references to color in this figure seafloor depth was estimated by adding the whale’s distance above the legend, the reader is referred to the web version of this article.) seafloor to the whale’s depth.
100 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
Seafloor echoes were rarely detected during buzzes due to the low Table 2 sound pressure level and high repetition rate of clicks obscuring re- Summary of data derived from 28 foraging dives by six tagged sperm whales at Kaikōura. turning echoes. Whale distance above the seafloor during buzzes was Sample size (N) varies because the tag detached part way through four of the dives. All depths are in metres. thus estimated by using a value of seafloor depth obtained via linear interpolation between the previous and following estimates, as long as N Mean (SD) Median (range) these had been acquired within 120 s of the buzz. With this method we – could not account for the occurrence of irregularities in the seafloor Search-start depth 28 48 (38) 35 (18 162) Search-end depth 24 595 (215) 573 (303–1272) during the time of interpolation, inevitably adding some uncertainty Min depth of bottom phase 24 701 (260) 729 (222–1076) around the estimates of whales’ distance above the seafloor during Max depth of bottom phase 24 1074 (261) 1130 (535–1439) buzzes. Mean depth of bottom phase 24 924 (232) 978 (462–1226) To evaluate the accuracy of echo-derived distances we compared Start depth of bottom phase 28 806 (286) 888 (248–1387) End depth of bottom phase 24 885 (296) 1000 (474–1396) the whale’s depth (as recorded by the tag’s pressure sensor) with the Sensu Miller et al. (2013)*: distance from the whale to the surface derived from the TWT of surface Min depth of bottom phase 24 560 (194) 511 (222–1061) echoes. Whale depths were obtained by both these methods at 40 Mean depth of bottom phase 24 903 (220) 958 (452–1204) random times (including all tags), covering depths between 30 and End depth of bottom phase 24 595 (215) 573 (303–1272) 1130 m. The error in whale depth derived from surface echoes in- * Note: The definitions of start and maximum depths of the bottom phased used in this creased with distance to the surface, varying between 1 and 23 m study are the same as those in Miller et al. (2013). (mean = 10.8 m) over distances between 30 and 1135 m (mean = 547). In relative terms, the error varied by between 1.0% and 9.1% of the tag came off before the end of the dive) (Table 1). The tag-on-an- ’ the whale s depth (mean = 2.3%). For the purpose of this study, we imal times varied between 2.3 and 8.1 h (mean = 5.6). Tagged whales consider this error to be acceptable. performed foraging dives with a mean duration of 50 min (range 33–66 min) and a mean bottom-phase depth of 924 m (462–1226 m) 2.4. Data analysis (2Table). The maximum depth recorded was 1439 m. Whales started echolocating at a mean depth of 48 m (18–162 m) after a silent descent Data from all tagged whales were pooled for analysis. Our analysis lasting on average 34 s (15–109 s). At the end of the dive, whales focused on foraging dives only, thus excluding shallow dives in which stopped clicking at a mean depth of 595 m (303–1272 m), followed by a no usual clicks or buzzes were detected. The depth and duration of each silent ascent of 7 min on average (5–14 min). The search phase of the dive phase was calculated. We also recorded the duration of each buzz dive lasted for 42 min on average (27–55 min) while the foraging phase and the inter-buzz interval (IBI, measured as the time between the last lasted 33 min (13–48 min). click of a buzz and the first click of the following one) to compare Between consecutive foraging dives, whales spent on average foraging behaviour at different depths and distances from the seafloor. 15 min at the surface (median 12 min, range 6–41 min). On two occa- To explore circadian differences in foraging behaviour we compared sions, two different whales spent 4.2 and 1.5 h travelling silently, al- the characteristics of the buzzes that took place before and after sunset. ternating between swimming at the surface and at depths of 10–20 m. Due to the small sample size of recorded night dives, we used this Echoes from the seafloor were detected for a total of 20,911 echo- analysis to identify knowledge gaps and guide research questions rather location clicks, with the estimated whale distance from the seafloor than test diel differences empirically. varying from 6 to 1418 m. Two general foraging modes were observed, Due to the low number of dives obtained in this study and its limited which we defined as pelagic and demersal (Figs. 4 and 5). Pelagic dives spatial and temporal scope, we limited statistical testing to only a few (n = 8) had a relatively narrow bottom phase at average depths of relevant comparisons to assess the rates and durations of buzzes. For either ca. 500 m or 850 m, and at least 50 m from the seafloor (but these cases we chose a simple confidence-interval approach (Di Stefano, typically > 200 m). Demersal dives (n = 12) had a bottom phase at or 2004; Foody, 2009), whereby differences in means and 95% confidence near (< 50 m) the seafloor (mean distance from the seafloor = 23 m, intervals (CI) were used as the measure of magnitude and uncertainty of 9–44), with variable depth as the whales roughly followed the contours each comparison (Di Stefano, 2004). Diagnostic plots (histograms of of the sea bed. In a few dives (n = 4) we observed a combination of residual distributions and normal Q-Q plots) were used to verify as- both modes, with part of the bottom phase in mid-water before and/or sumptions of normality in buzz patterns. after foraging on the seafloor. Both types of foraging were recorded for all the tagged whales. Regardless of the foraging mode, dives were 3. Results bracketed by steep descents and ascents typical of sperm whale dives (Watkins et al., 1993; Watwood et al., 2006). Whales often foraged 3.1. General foraging behaviour opportunistically in the descents and ascents to/from the bottom phases of demersal dives, as shown by occasional buzzes during these periods A total of 33.6 h of combined acoustic and movement data were (e.g., Fig. 4b). Across the 24 complete dives, 57% of the total time of collected from six successful tag deployments on six different mature active foraging was spent within 50 m of the seafloor. male sperm whales, including 24 complete foraging dive cycles (des- There was a positive and significant correlation between the depth cent-bottom-ascent) and another four incomplete dives (i.e., in which
Table 1 Summary of tag deployments. Only foraging dives are included here, defined by the presence of usual clicks.
Tag ID Date of deployment Start time Programmed tag-on time (h) Tag-on time (h) N dives N complete dives
050a 19-Feb-13 12:09 p.m. 3.0 2.3 3 2 051a 20-Feb-13 11:43 a.m. 3.0 3.7 4 3 056a 25-Feb-13 12:36 p.m. 6.0 8.1 7 7 057a 26-Feb-13 1:00 p.m. 6.0 6.3 2 2 061a 2-Mar-13 4:24 p.m. 6.0 6.8 7 6 061b 2-Mar-13 5:09 p.m. 6.0 6.4 5 4 TOTAL 33.6 28 24
101 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
Fig. 4. Foraging activity of sperm whales in relation to the seafloor. Time-depth profiles of whales during example pelagic (a), demersal (b), and combined (c) dives. Black lines represent the portion of the dive when the whale was echolocating, red dots represent buzzes and green dots represent seafloor depths estimated from echo reflections. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
known. The correlation between this time interval and the mean buzz depth during a dive was also positive and significant, but with a much lower predictive power (r2 = 0.33, p = 0.003, n = 24), and therefore has less potential as a biologically meaningful proxy.
3.2. Vertical distribution of buzzes
A total of 438 buzzes were detected in 28 foraging phases. The number of buzzes per dive ranged from 9 to 38, with a mean of 17 (Table 3). Buzz duration had a normal distribution with a mode at 6–8 s, varying from 2 to 50 s. The mean buzz depth per dive was 869 m (475–1205 m), and whales emitted buzzes at depths between 284 and
Table 3 Summary of buzz depths and distance from the seafloor produced by six tagged sperm whales at Kaikōura. All depths are in metres.
Fig. 5. Types of foraging dives performed by tagged sperm whales. A histogram of the N Mean (SD) Median (range) mean bottom phase depths for all complete foraging dives (n = 24), distinguishing be- tween dive types. No. of buzzes per dive 24 17 (7) 15 (9–38) Depth of first buzz 27 611 (208) 511 (284–1033) Depth of last buzz 24 663 (193) 633 (421–1261) at the end of the search phase (marked by the last echolocation click Min buzz depth 24 549 (156) 496 (284–949) during a dive) and the time taken by the whale to resurface from this Max buzz depth 24 1056 (261) 1099 (508–1433) depth (r2 = 0.86, p < 0.001, n = 24). Although this relationship is Mean buzz depth 24 869 (208) 891 (475–1205) fl – based on a small sample size, it has the potential to be used as a useful Min buzz distance from the sea oor 24 182 (315) 17 (6 953) Max buzz distance from the seafloor 24 623 (273) 601 (116–1052) proxy of foraging depth when the time from the last click to surfacing is Mean buzz distance from the seafloor 24 319 (299) 164 (23–997)
102 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
Fig. 6. Vertical distribution of sperm whale buzzes. Histograms of depth (a) and distance from the seafloor (b) of buzzes recorded during foraging phases. Depths are the lower bounds of 50 m bins.
1433 m (Table 3). On average, the first buzz in a dive was made 8.4 min patterns in buzz duration and inter-buzz intervals separately for each after fluking up (range 4.7–14.4 min). whale. There was high inter-individual variability, especially in inter- Buzz depth had a clear bimodal distribution (Fig. 6a), with one buzz intervals. Consistent with the overall trend, five out of the six mode at 850–950 m (accounting for 25% of all buzzes) and a second tagged whales produced longer buzzes near the seafloor than at dis- mode at 450–550 m (17% of all buzzes). The majority of the buzzes tances > 50 m (Fig. 7a), and four whales produced longer inter-buzz (71%) occurred at depths greater than 800 m. intervals at the seafloor than in the water column (Fig. 7b). Estimates of whale distance from the seafloor were possible via No surface feeding was observed or recorded during this study. The linear interpolation for 354 out of a total of 438 recorded buzzes. The minimum buzz depth was 284 m, indicating that no foraging occurred remaining 84 buzzes were recorded while the whale was foraging near at or near the surface. the seafloor during times in which seafloor echoes were not detected within 120 s either side of the buzz. Because whale departures from the 3.3. Day/night patterns in foraging behaviour seafloor were always obvious in the echograms and there was no in- dication of such departures before any of these 84 buzzes, we assumed Data on foraging behaviour after sunset were available for three of that they were performed within 50 m of the seafloor. the tagged whales, comprising seven hours of nocturnal data from nine Sperm whale buzzes took place at distances from the seafloor dives, with 142 recorded buzzes. Due to the small number of dives varying from 6 to 1052 m. The distribution of buzz distance from the recorded at night, diel patterns in foraging behaviour were only in- seafloor had one clear mode within 50 m of the seafloor (Fig. 6b). The vestigated by comparing depth and distance from the seafloor of rest of the buzzes (n = 274) were fairly evenly distributed across dis- buzzes. tances between 50 and 1000 m from the seafloor, with perhaps a second There was some evidence for diel differences in the vertical dis- mode at 400–450 m from the seafloor. Following the definition of the tribution of buzzes (Fig. 8a–c). The mode in buzz depth at 450–550 m benthic boundary layer (BBL) as extending from the seafloor to 200 m was almost exclusively (97%) composed of day-time buzzes, while the above it (Angel and Boxshall, 1990), 48% (n = 211) of buzzes occurred mode at 850–950 m was most evident for night-time buzzes. This diel within this layer, indicating a substantial amount of benthopelagic variation was most evident in the depths of pelagic buzzes (Fig. 8b,c). foraging. Of the 175 pelagic buzzes recorded during day-time, a large proportion To identify vertical patterns in foraging behaviour more clearly, we (41%) were made at depths of 450–550 m, with a smaller proportion assessed the distribution of buzzes simultaneously across whale depth (22%) at 850–950 m (Fig. 8b). In contrast, of the 99 pelagic buzzes and distance from the seafloor. The majority of buzzes fell within one of recorded at night, only 2% were made at 450–550 m deep, while the three clusters: 37% of all buzzes were produced within 50 m of the vast majority (63%) were produced at 850–950 m (Fig. 8c). The mode seafloor (depths of 800–1400 m), 35% were in mid-water at depths of in buzz distance from the seafloor at less than 50 m was apparent for 700–1000 m, and 21% were in mid-water at depths of 400–600 m (see both day-time and night-time buzzes. Fig. 7a). Distance from the seafloor appeared to influence buzz patterns. Buzzes emitted at or near the seafloor (distance < 50 m) were longer 3.4. Reaction to tagging attempts and tag deployment (mean buzz duration = 15.9 s, range = 2.0–49.7 s, n = 164) than buzzes emitted at distances > 50 m (mean = 9.0 s, range = 3.2–27.1 s, We made 27 close approaches (< 200 m) to sperm whales to at- n = 274). The rate of buzzes was also di fferent between pelagic and tempt a tag deployment, with 8 successful tag attachments. One tag was demersal prey, with longer inter-buzz intervals at the seafloor (mean IBI never recovered, and another malfunctioned. Behavioural reactions to = 149.0 s, range = 2–976, n = 164) than in the water column (mean the tagging attempt included startles (n = 3), evasive behaviours such IBI = 90.5 s, range = 4–471, n = 246). This effect was statistically as swimming away from the boat (n = 9), shallow diving (n = 15), and significant for both buzz duration and IBI (95% CI of the difference early diving (n = 8). After shallow diving, whales often resurfaced between means = 5.2–8.6 s and 26.0–90.9 s, respectively). To reduce within a few minutes. Note that whales could exhibit multiple reactions the influence of within-individual autocorrelation, we analysed the during the same tagging attempt. No reaction to a tagging attempt was recorded on five occasions.
103 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
Fig. 7. Individual patterns of sperm whale buzzes. Mean buzz duration (a) and inter-buzz interval (b) of buzzes recorded during foraging phases. Demersal buzzes (black) include those produced within 50 m of the seafloor, while pelagic buzzes (white) include those produced above 50 m of the seafloor. Error bars represent 95% confidence intervals (CI), with the * symbol indicating lack of overlap in the CI of demersal and pelagic buzzes.
Fig. 8. Diel patterns in foraging behaviour of sperm whales at Kaikōura. (a) Buzz depth vs distance from the seafloor, with day buzzes represented by grey triangles and night buzzes by black circles. (b, c) Histogram of buzz depths recorded during the day (b) and at night (c) for pelagic (light grey, > 50 m from the seafloor) and demersal buzzes (dark grey, < 50 m from the seafloor). Depths in (b) and (c) are the lower bounds of 50 m bins.
104 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
4. Discussion observed inshore-offshore differences in habitat use. On the other hand, the dominance of pelagic foraging by sperm whales in the upper canyon This study investigated the foraging behaviour of male sperm might simply reflect a particularly high pelagic productivity there. whales over the Kaikōura submarine canyon. While the data are from a Steep slopes – which are more prevalent in the upper Kaikōura canyon – limited period and from only six whales, they provide novel insights have been shown to increase the biomass of faunal assemblages by into the use of the water column by sperm whales at Kaikōura. Assumed truncating the vertical distribution of pelagic organisms (Mauchline prey capture attempts, indicated by buzzes, occurred at all depths and Gordon, 1991; Sutton et al., 2008). Canyon heads and upper slopes below the epipelagic zone, from ca. 300–1430 m. Depth sensor data and are also known to facilitate aggregations of pelagic zooplankton echoes from the seafloor indicated that sperm whales searched for prey (Macquart-Moulin and Patriti, 1996). If pelagic prey are more abun- within three main vertical strata. Demersal prey were targeted when dant, or if prey energy density is higher, shallower pelagic dives might whales foraged within 50 m of the seafloor (and as close as 6 m) at be more economical than deep dives to the seafloor. Further research on depths of 900–1300 m, while mesopelagic prey were hunted in mid- the drivers of habitat use by sperm whales at Kaikōura, with simulta- water at depths of around 500 m and around 850 m. The wide and neous sampling of prey, is necessary to clarify the observed differences. multimodal distribution of diving depths by sperm whales is consistent with previous studies (Teloni et al., 2008; Miller et al., 2013; Fais et al., 4.2. Sperm whale prey 2015). It reflects the flexibility of their long-range sampling strategies to detect patchy food resources (Fais et al., 2015), and adaptability to The average inter-buzz interval (IBI), and duration of sperm whale track them down over vast volumes of water. buzzes were generally different when whales were foraging in mid- Average diving and foraging depths in our study were much deeper water vs near the seafloor, suggesting that whales might be targeting (ca. 200–500 m more) than those previously reported for male sperm different prey types in these two habitats. Although there was some whales in Kaikōura (Miller et al., 2013) and northern Norway (Teloni individual variation, typically longer intervals between buzzes while et al., 2008), and slightly deeper (ca. 100–200 m more) than those in foraging within 50 m of the seafloor suggests that prey are more widely the Atlantic Ocean, Gulf of Mexico and Ligurian Sea (Watwood et al., spread in this habitat than in the water column. Longer buzzes for prey 2006). The drivers of these differences remain unknown, but the deeper found close to the seafloor might reflect longer chases, perhaps of faster foraging tendencies observed in this study appear to be mostly driven or more motile prey. Interestingly, previous findings at Kaikōura from by the high incidence of foraging at or near the seafloor. Moreover, the Miller et al. (2013) showed that buzzes near the seafloor had fewer water depths in the areas studied by Teloni et al. (2008) and Watwood sharp turns than pelagic ones. It was suggested that whales might line et al. (2006) were generally deeper than those in our research area, up slow-moving or unsuspecting prey from a distance. The combined suggesting that the shallower foraging depths recorded in their studies findings highlight the fact that much of the foraging ecology of sperm were not driven by demersal foraging at shallower bathymetries. The whales remains unknown, but indicate that they have the flexibility to average number of buzzes per dive was almost double those previously adjust their hunting tactics to efficiently target a variety of prey types. recorded in Norway (Teloni et al., 2008), and similar to those reported The fact that they forage on different prey types is also likely to be by Watwood et al. (2006), suggesting relatively high prey densities at reflected in the multimodal nature of their preferred diving depths. Kaikōura. Further research would be necessary to clarify the energetic balance of the different prey layers, but it appears that prey at the seafloor are 4.1. Demersal foraging by pelagic predators sparser and require longer (i.e., more energy) to catch, in addition to being found at greater depths. Perhaps they have a higher caloric value The sperm whales tagged in our study spent more than half of their compared to pelagic prey, outweighing the increased energetic costs to active foraging time within 50 m of the seabed. Compared to pelagic obtain them. foraging, the prevalence of benthic or demersal foraging is generally Although the trophic pathway from the productive benthos to the reported to be infrequent (Watwood et al., 2006; Miller et al., 2013). prey targeted by sperm whales at the seafloor is unknown, benthope- Male sperm whales in the Andøya Canyon (Norway) performed a few lagic fish are a likely candidate to mediate this link. Ling (Genypterus dives to the seafloor, but pelagic foraging was dominant (Teloni et al., blacodes) and hāpuku (Polyprion oxygeneios, also known as grouper) are 2008; Fais et al., 2015). The high proportion of demersal foraging found two fish species common at Kaikōura (Francis, 1979) which are fre- in the present study suggests that the extreme benthic productivity of quent prey for New Zealand sperm whales (Gaskin and Cawthorn, the Kaikōura Canyon (De Leo et al., 2010) plays an important role in 1967). Ling is a demersal species which feeds primarily on benthic prey, sustaining the diet of these deep-diving predators. including macrourid fishes (Clark, 1985), which are extremely abun- Intriguingly, the high incidence of demersal foraging found in our dant in the Kaikōura Canyon (De Leo et al., 2010). Hāpuku have a more study contrasts with the mostly pelagic hunting by sperm whales re- flexible vertical distribution, foraging on demersal and mesopelagic corded at Kaikōura in a previous study (Miller et al., 2013), in which squid and fish (Nguyen et al., 2015; Tromp et al., 2016). Further studies only one out of 28 full dives had a bottom phase foraging on the sea- are necessary to better understand prey preferences and the relative floor. The whales sampled by Miller et al. (2013) were tracked further contribution of benthic vs pelagic energy sources to the diet of sperm inshore (but only by a few km) and in generally shallower waters than whales. As sperm whales rarely strand at Kaikōura, precluding stomach the whales tagged in the present study. Given the differences in contents analysis, stable isotope analyses (Ruiz-Cooley et al., 2004; bathymetry, seasonality, methodology and individual whales sampled, Newsome et al., 2010) might provide useful insights into their trophic the study by Miller et al. (2013) and the present one are not directly ecology. comparable. The observed difference, however, raises the question of The great majority of the available pelagic biomass in deep-sea why the exploitation of benthic biomass was rare in Miller et al., 2013 systems below the photic zone is found within the DSL, composed of (which covered a period of one year), yet so commonplace in the pre- mesopelagic organisms (Johnson, 1948; Hays, 2003). Many of these sent one. mesopelagic animals undergo a diel vertical migration (Gjøsaeter and Submarine canyons have complex patterns of topography and hy- Kawaguchi, 1980) from their day-time depths below the twilight zone drography, creating heterogenous habitats with strong gradients in (typically 400–800 m) up to the productive surface waters where they biomass distribution and community structure (McClain and Barry, feed at night in the shelter of darkness. However, some mesopelagic 2010; Ramirez-Llodra et al., 2010). It is possible that the organisms that organisms do not undergo a diel migration and remain in the deeper support the demersal component of the sperm whales’ diet are found parts of the DSL during the night (Childress, 1995; Kaartvedt et al., mostly in the deeper and lower parts of the canyon, resulting in the 2009). The foraging by tagged sperm whales in Kaikōura at depths of
105 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
400–600 m fall within the typical boundaries of the lower parts of DSLs resurfaced within a few minutes after shallow diving suggested that this reported globally (O'Driscoll et al., 2009; Netburn and Koslow 2015; was used as an evasive manoeuvre. Although the physiological costs to Klevjer et al., 2012, 2016). Although simultaneous sampling of sperm the animals are unknown, they are unlikely to be long-lasting. How- whale behaviour and prey distribution has not yet been attempted at ever, this evidence suggests that sperm whales may experience some Kaikōura, at least a fraction of the mesopelagic prey targeted by sperm stress from tagging attempts, and that foraging behaviour measured via whales is likely to be associated with the lower boundaries of the DSL. attached tags may be modified (Miller et al., 2005, 2009). Passive Based on stomach contents analysis, the primary prey item of New acoustic studies (Wahlberg, 2002; Miller et al., 2013) can be used to Zealand sperm whales are Onychoteuthid squid, and predominantly the study foraging behaviour in detail, with minimal influence on the target warty squid Onykia ingens (formerly known as Moroteuthis ingens) animals. This should be considered when choosing the most appropriate (Gaskin and Cawthorn, 1967). Warty squid undergo a gradual onto- method to study diving and foraging behaviour. genic migration from epipelagic waters into deeper demersal habitats as they mature (Jackson, 1997), with depth ranges of 300–1400 m 4.4. Ecological implications of the observed foraging behaviour (Jackson et al., 2000). Adults are demersal at night but appear to mi- grate vertically into pelagic waters during the day, where they feed on The present study has provided useful data for understanding which mesopelagic myctophids associated with the DSL (Jackson et al., 1998, strata might be important for foraging sperm whales at Kaikōura. 2000). We found some evidence in this study for circadian shifts in the Oceanographic characteristics of mid-water depths at ca. 500 and foraging depths of sperm whales, and a likely association of pelagic 900 m (e.g., temperature, salinity and dissolved oxygen), in addition to prey captures with depths typical of lower-DSL boundaries (i.e., ca. physical aspects of the seafloor (e.g., slope, aspect and sediment type) 500 m). Warty squid are found at Kaikōura, and mature specimens have would be useful additions to the candidate set of environmental vari- been found floating at the surface with marks consistent with sperm ables used to investigate the habitat preferences of sperm whales. The whale teeth (Guerra, pers. obs.) It seems very plausible that sperm apparent differences in foraging behaviour between inshore and off- whales are targeting warty squid during their dives to 400–600 m in shore habitats at Kaikōura (Miller et al., 2013, present study) provide a mid-water during the day, and possibly during demersal foraging at useful framework for future research aimed at understanding intra- night. Strong associations with the DSL have been shown for many population variability in the spatial distribution of sperm whales. deep-diving top predators including pilot whales (Abecassis et al., By feeding at depth and releasing fecal plumes close to the surface 2015), elephant seals (Saijo et al., 2016) and Blainville’s beaked whales (Kooyman et al., 1981; Whitehead, 1996), whales play an important (Arranz et al., 2011), for which circadian changes in foraging behaviour role in ecosystem connectivity and enhancing productivity (Lavery have been associated with changes in the vertical distribution of their et al., 2010; Roman and McCarthy, 2010; Sutton, 2013). They act as an prey. ‘upward whale pump’, recycling organic carbon and nutrients from the deep by bringing them to the photic zone, where they become available 4.3. Use of Dtags to study sperm whale behaviour to phytoplankton (Roman and McCarthy, 2010; Roman et al., 2014). The vertical flux of nutrients is particularly great in the case of sperm Dtags can provide useful insights into the foraging behaviour of whales (Roman et al., 2014), due to their extreme foraging depths and deep-diving mammals in relation to the seafloor, in addition to pro- their uptake of food sources across the entire water column. In Kai- viding night-time behaviour information. They can also offer useful kōura, where the density of sperm whales has been historically high information on the density of organisms in the water column, by (Childerhouse et al., 1995), their contribution to the biological pump quantifying the backscatter from animals that are ensonified by the could be an important driver of the canyon’s high productivity. The fact whales (Arranz et al., 2011). However, they have some limitations – that abundance of sperm whales at Kaikōura appears to be declining such as not being able to provide accurate georeferenced locations – (van der Linde, 2009), raises questions about whether this productivity which prevent relating the whales’ foraging behaviour to other relevant will be sustained. Unexpected responses of coastal ecosystems to an- physical characteristics of their habitat. For example, we were not able thropogenic stressors arise from insufficient knowledge about complex to address if demersal foraging was more or less prevalent over parti- food web interactions and feedback loops (Doak et al., 2008). It is cular topographic features such as canyon slopes. Passive acoustic therefore essential to better understand food web structures, mechan- studies using three-dimensional arrays (e.g., Wahlberg, 2002; Miller isms and functions. Elucidating the processes that result in the extra- and Dawson, 2009; Miller et al., 2013) are an alternative research tool, ordinary biological assemblages associated with submarine canyons is with the advantage of providing high-resolution georeferenced 3D lo- key to their long-term protection. cations for relating whale movement and behaviour to their physical environment. Acknowledgements We found a strong correlation between a whale’s depth at the end of the search phase and the time taken to resurface, revealing a potential Support for this work was provided from contributions by Ray Dalio proxy for foraging depth when the underwater track of a whale is un- to the WHOI Access to the Sea Fund. We thank the owners, captain and known. The duration of this silent ascent is an easy metric to record crew of the M/V Alucia for their dedicated assistance. For assistance in while acoustically tracking a whale. The application of this variable is the field we thank Moira Brown, Andreas Fahlman, Don LeRoi, Wayne however limited to sperm whales foraging individually and it only Perryman, Luis Lamar and Maryann Morin. The New Zealand Whale provides an estimate of depth for the final phase of a dive. In spite of and Dolphin Trust and the University of Otago Marine Mammal these caveats, it could be useful for investigating temporal and spatial Research Group provided equipment and access to the photo-identifi- differences in foraging depth between seasons, or for assessing an- cation catalogue from Kaikōura. Thanks to Mark Johnson for making thropogenic impacts on behaviour. the Matlab scripts for Dtag analysis freely available, and for helpful Our observations of sperm whale behaviour during tagging attempts suggestions, and to Brian Miller for helpful comments and advice. We revealed behavioural reactions to close vessel approaches and tagging. thank Mike Morrissey and Whale Watch Kaikōura for logistical support. Tendencies to reduce time at the surface, for example, correspond to Sperm whales were tagged under permit 35604-MAR issued by the New changes expected under a stress response (Richter et al., 2006; Lusseau, Zealand Department of Conservation to Michael Moore, and with the 2003). The rate of shallow diving (55% of all approaches) was sub- consent of the University of Otago Animal Ethics Committee (AEC stantially higher than the average rate recorded for Kaikōura sperm number 73/2012) issued to William Rayment. The manuscript was whales (6% of 677 approaches, Otago University Marine Mammal Re- greatly improved by comments and analysis suggestions from an search Group, unpublished data). The fact that sperm whales often anonymous reviewer.
106 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
References 2010.0863. Leduc, D., Rowden, A.A., Bowden, D.A., Probert, P.K., Pilditch, C.A., Nodder, S.D., 2012. Unimodal relationship between biomass and species richness of deep-sea nematodes: Abecassis, M., Polovina, J., Baird, R.W., Copeland, A., Drazen, J.C., Domokos, R., Oleson, implications for the link between productivity and diversity. Mar. Ecol. Prog. Ser. E., Jia, Y., Schorr, G.S., Webster, D.L., Andrews, R.D., 2015. Characterizing a foraging 454, 53–64. hotspot for short-finned pilot whales and Blainville’s beaked whales located off the Lusseau, 2003. Male and female bottlenose dolphins Tursiops spp. have different strategies west side of Hawai‘i Island by using tagging and oceanographic data. PLoS One 10, to avoid interactions with tour boats in doubtful sound, New Zealand. Mar. Ecol. e0142628. Prog. Ser. 257, 267–274. Angel, M.V., Boxshall, G.A., 1990. Life in the benthic boundary layer: connections to the Macquart-Moulin, C., Patriti, G., 1996. Accumulation of migratory micronekton crusta- mid-water and sea floor. Philos. Trans. R. Soc. A 33, 15–28. ceans over the upper slope and submarine canyons of the northwestern mediterra- Arranz, P., Aguilar Soto, N., Madsen, P.T., Brito, A., Bordes, F., Johnson, M.P., 2011. nean. Deep-Sea Res. Part I Oceanogr. Res. Pap. 43, 579–601. Following a foraging fish-finder: diel habitat use of Blainville’s beaked whales re- Madsen, P., Wahlberg, M., Møhl, B., 2002. Male sperm whale (Physeter macrocephalus) vealed by echolocation. PLoS One 6, e28353. acoustics in a high-latitude habitat: implications for echolocation and communica- Barham, E., 1966. Deep scattering layer migration and composition: observations from a tion. Behav. Ecol. Sociobiol. 53, 31–41. diving saucer. Science 151, 1399–1403. Marshall, N.B., 1965. Systematical and biological studies of the macrourid fishes Benoit‐Bird, K.J., Würsig, B., Mfadden, C.J., 2004. Dusky dolphin (Lagenorhynchus ob- (Anacanthini-Teleostii). Deep-Sea Res. I 12, 299–322. scurus) foraging in two different habitats: active acoustic detection of dolphins and Martin, A.R., Clarke, M.R., 1986. The diet of sperm whales (Physeter macrocephalus) their prey. Mar. Mamm. Sci. 20, 215–231. captured between Iceland and Greenland. J. Mar. Biol. Assoc. U.K. 1, 779–790. Boren, L.J., Muller, C.G., Gemmell, N.J., 2006. Colony growth and pup condition of the Mate, B.R., Irvine, L.M., Palacios, D.M., 2016. The development of an inter- New Zealand fur seal (Arctocephalus forsteri) on the Kaikoura coastline compared with mediate‐duration tag to characterize the diving behavior of large whales. Ecol. Evol. other east coast colonies. Wild Res. 33, 497–505. http://dx.doi.org/10.1002/ece3.2649. Childerhouse, S.J., Dawson, S.M., Slooten, E., 1995. Abundance and seasonal residence of Mauchline, J., Gordon, J.D.M., 1991. Oceanic pelagic prey of benthopelagic fish in the sperm whales at Kaikōura, New Zealand. Can. J. Zool. 73, 723–731. benthic boundary layer of a marginal oceanic region. Mar. Ecol. Prog. Ser. 74, Childress, J.J., 1995. Are there physiological and biochemical adaptations of metabolism 109–115. in deep-sea animals? Trends Ecol. Evol. 10, 30–36. McClain, C.R., Barry, J.P., 2010. Habitat heterogeneity, disturbance, and productivity Clark, M.R., 1985. The food and feeding of seven fish species from the Campbell Plateau, work in concert to regulate biodiversity in deep submarine canyons. Ecology 91, New Zealand. NZ J. Mar. Freshw. Res. 19, 339–363. 964–976. Davis, R.W., Jaquet, N., Gendron, D., Markaida, U., Bazzino, G., Gilly, W., 2007. Diving Miller, B., Dawson, S., 2009. A large-aperture low-cost hydrophone array for tracking behavior of sperm whales in relation to behavior of a major prey species, the jumbo whales from small boats. J. Acoust. Soc. Am. 126, 2248–2256. squid, in the Gulf of California, Mexico. Mar. Ecol. Prog. Ser. 333, 291–302. Miller, B., Dawson, S., Vennell, R., 2013. Underwater behavior of sperm whales off Dawson S., 1990. Building the CETOS directional hydrophone. 〈http:// Kaikoura, NZ, as revealed by a three-dimensional hydrophone array. J. Acoust Soc. whaledolphintrust.org.nz/wp-content/uploads/Building-Directional-HPs.pdf〉. Am. 134, 2690–2700. De Leo, F.C., Smith, C.R., Rowden, A.R., Bowden, D.A., Clark, M.R., 2010. Submarine Miller, P.J.O., Johnson, M., Madsen, P.T., Biassoni, N., Quero, M., Tyack, P.L., 2009. canyons: hotspots of benthic biomass and productivity in the deep sea. Proc. R. Soc. B Using at-sea experiments to study the effects of airguns on the foraging behavior of 277, 2783–2792. sperm whales in the Gulf of Mexico. Deep-Sea Res. I 56, 1168–1181. Di Stefano, J., 2004. A confidence interval approach to data analysis. For. Ecol. Manag. Miller, P.J.O., Johnson, M.P., Tyack, P.L., 2004. Sperm whale behaviour indicates the use 187, 173–183. of echolocation click buzzes ‘creaks' in prey capture. Proc. R. Soc. Lond. B 271, Doak, D.F., Estes, J.A., Halpern, B.S., Jacob, U., Lindberg, D.R., Lovvorn, J., Monson, 2239–2247. D.H., Tinker, M.T., Williams, T.M., Wootton, J.T., Carroll, I., 2008. Understanding Miller P.J.O., Tyack P.L., Biassoni N., Johnson M.P., 2005. Short-term effects of suction and predicting ecological dynamics: are major surprises inevitable. Ecology 89, cup tagging on the deep-diving behaviour of sperm whales. In: Proceedings of the 952–961. Abstracts in Second International Symposium of Biologging Science, St. Andrews, Fais, A., Soto, N.A., Johnson, M., Pérez-González, C., Miller, P.J.O., Madsen, P.T., 2015. Scotland. Sperm whale echolocation behaviour reveals a directed, prior-based search strategy Møhl, B., Wahlberg, M., Madsen, P.T., Miller, L.A., Surlykke, A., 2000. Sperm whale informed by prey distribution. Behav. Ecol. Sociobiol. 69, 663–674. clicks: directionality and source level revisited. J. Acoust. Soc. Am. 107, 638–648. Foody, G.M., 2009. Classification accuracy comparison: hypothesis tests and the use of Moors-Murphy, H.B., 2014. Submarine canyons as important habitat for cetaceans, with confidence intervals in evaluations of difference, equivalence and noninferiority. special reference to the Gully: a review. Deep Sea Res. II: Top. Stud. Oceanogr. 104, Remote Sens. Environ. 113, 1658–1663. 6–19. Francis MP (1979) Checklist of the marine fishes of Kaikoura, New Zealand. Mauri ora 7: Netburn, A.N., Koslow, J.A., 2015. Dissolved oxygen as a constraint on daytime deep pp.63–71. scattering layer depth in the southern California current ecosystem. Deep-Sea Res. Gaskin, D.E., Cawthorn, M.W., 1967. Diet and feeding habits of the sperm whale (Physeter Part I Oceanogr. Res. Pap. 104, 149–158. catadon L.) in the Cook Strait region of New Zealand. N.Z. J. Mar. Freshw. Res. 2, Newsome, S.D., Clementz, M.T., Koch, P.L., 2010. Using stable isotope biogeochemistry to 156–179. study marine mammal ecology. Mar. Mamm. Sci. 26, 509–572. Gjøsaeter, J.K., Kawaguchi, K., 1980. A review of the world resources of mesopelagic fish. Nguyen, H.M., Rountrey, A.N., Meeuwig, J.J., Coulson, P.G., Feng, M., Newman, S.J., FAO Fish. Tech. Pap. 193, 151. Waite, A.M., Wakefield, C.B., Meekan, M.G., 2015. Growth of a deep-water, pre- Hays, G.C., 2003. A review of the adaptive significance and ecosystem consequences of datory fish is influenced by the productivity of a boundary current system. Sci. zooplankton diel vertical migrations. Hydrobiologia 503, 163–170. Rep. 5. Jackson, G.D., 1997. Age, growth and maturation of the deepwater squid Moroteuthis O'Driscoll, R.L., Gauthier, S., Devine, J.A., 2009. Acoustic estimates of mesopelagic fish: ingens (Cephalopoda: onychoteuthidae) in New Zealand waters. Polar Biol. 17, as clear as day and night? ICES J. Mar. Sci. 66. 268–274. Okutani, T., Nemoto, T., 1964. Squids as the food of sperm whales in the Bering Sea and Jackson, G.D., McKinnon, J.F., Lalas, C., Ardern, R., Buxton, N.G., 1998. Food spectrum of Alaska Gulf. Sci. Rep. Whales Res Inst. 18. the deepwater squid Moroteuthis ingens (Cephalopoda: onychoteuthidae) in New Papastavrou, V., Smith, S.C., Whitehead, H., 1989. Diving behavior of the sperm whale, Zealand waters. Polar Biol. 20, 56–65. Physeter macrocephalus,off the Galapagos Islands. Can. J. Zool. 67, 839–846. Jackson, G.D., Shaw, A.G., Lalas, C., 2000. Distribution and biomass of two squid species Ramirez-Llodra, E., Brandt, A., Danovaro, R., De Mol, B., Escobar, E., German, C., Levin, off southern New Zealand: Nototodarus sloanii and Moroteuthis ingens. Polar Biol. 23, L., Arbizu, P., Menot, L., Buhl-Mortensen, P., Narayanaswamy, B.E., 2010. Deep, 699–705. diverse and definitely different: unique attributes of the world's largest ecosystem. Jaquet, N., Dawson, S., Slooten, E., 2000. Seasonal distribution and diving behaviour of Biogeosciences 7, 2851–2899. male sperm whales off Kaikōura: foraging implications. Can. J. Zool. 78, 407–419. Rice, D.W., 1989. Sperm whale. Physeter macrocephalus Linnaeus, 1758. In: Ridgway, H. Johnson, M.W., 1948. Sound as a tool in marine ecology, from data on biological noises (Ed.), Handbook of Marine Mammals 4 Academic Press, London. and the deep scattering layer. J. Mar. Res. 7, 443–458. Richter, C., Dawson, S., Slooten, E., 2006. Impacts of commercial whale watching on male Johnson, M.P., Madsen, P.T., Zimmer, W.M.X., Aguilar de Soto, N., Tyack, P.L., 2004. sperm whales at Kaikoura, New Zealand. Mar. Mamm. Sci. 22, 46–63. Beaked whales echolocate on prey. Biol. Lett. 271, 383–386. Roman, J., Estes, J.A., Morissette, L., Smith, C., Costa, D., McCarthy, J., Nation, J.B., Johnson, M.P., Tyack, P.L., 2003. A digital acoustic recording tag for measuring the re- Nicol, S., Pershing, A., Smetacek, V., 2014. Whales as marine ecosystem engineers. sponse of wild marine mammals to sound. IEEE J. Ocean. Eng. 28, 3–12. Front. Ecol. Environ. 12, 377–385. Kaartvedt, S., Røstad, A., Klevjer, T.A., Staby, A., 2009. Use of bottommounted echo Roman, J., McCarthy, J.J., 2010. The whale pump: marine mammals enhance primary sounders in exploring behavior of mesopelagic fishes. Mar. Ecol. Prog. Ser. 395, productivity in a coastal basin. PLoS One 5, e13255. 109–118. Ruiz-Cooley, R.I., Gendron, D., Aguíñiga, S., Mesnick, S., Carriquiry, J.D., 2004. Trophic Klevjer T.A., Irigoien X., Røstad A., Fraile-Nuez E., Benítez-Barrios V.M., Kaartvedt S., relationships between sperm whales and jumbo squid using stable isotopes of C and (2016) Large scale patterns in vertical distribution and behaviour of mesopelagic N. Mar. Ecol. Prog. Ser. 277, 275–283. scattering layers. Scientific Reports 6:19873. Saijo, D., Mitani, Y., Abe, T., Sasaki, H., Goetsch, C., Costa, D.P., Miyashita, K., 2016. Klevjer, T.A., Torres, D.J., Kaartvedt, S., 2012. Distribution and diel vertical movements Linking mesopelagic prey abundance and distribution to the foraging behavior of a of mesopelagic scattering layers in the Red Sea. Mar. Biol. 159, 1833–1841. deep-diving predator, the northern elephant seal. Deep Sea Res. II: Top. Stud. Kooyman, G.L., Castellini, M.A., Davis, R.W., 1981. Physiology of diving marine mam- Oceanogr. http://dx.doi.org/10.1016/j.dsr2.2016.11.007. mals. Annu. Rev. Physiol. 43, 343–356. Santos, M.B., Pierce, G.J., Boyle, P.R., Reid, R.J., Ross, H.M., Patterson, I.A.P., Kinze, C.C., Lavery, T.J., Roudnew, B., Gill, P., Seymour, J., Seuront, L., Johnson, G., Mitchell, J.G., Tougaard, S., Lick, R., Piatkowski, U., Hernandez-Garcia, V., 1999. Stomach contents Smetacek, V., 2010. Iron defecation by sperm whales stimulates carbon export in the of sperm whales Physeter macrocephalus stranded in the North Sea 1990–1996. Mar. Southern Ocean. Proc. R. Soc. Lond. B Biol. Sci. http://dx.doi.org/10.1098/rspb. Ecol. Prog. Ser. 183, 281–294.
107 M. Guerra et al. Deep-Sea Research Part I 128 (2017) 98–108
Shepard, F.P., Dill, R.F., 1966. Submarine Canyons and other Sea Valleys. Rand Danovaro, R., 2011. Canyon conditions impact carbon flows in food webs of three McNally & Co, Chicago, IL. sections of the Nazaré canyon. Deep Sea Res. II: Top. Stud. Oceanogr. 58, 2461–2476. Sutton, T.T., 2013. Vertical ecology of the pelagic ocean: classical patterns and new Vetter, E.W., Smith, C.R., De Leo, F.C., 2010. Hawaiian hotspots: enhanced megafaunal perspectives. J. Fish. Biol. 83, 1508–1527. abundance and diversity in submarine canyons on the oceanic islands of Hawaii. Mar. Sutton, T.T., Porteiro, F.M., Heino, M., Byrkjedal, I., Langhelle, G., Anderson, C.I., Horne, Ecol. 31, 183–199. J., Søiland, H., Falkenhaug, T., Godø, O.R., Bergstad, O.A., 2008. Vertical structure, Wahlberg, M., 2002. The acoustic behaviour of diving sperm whales observed with a biomass and topographic association of deep-pelagic fishes in relation to a mid-ocean hydrophone array. J. Exp. Mar. Biol. Ecol. 281, 53–62. ridge system. Deep Sea Res. II: Top. Stud. Oceanogr. 55, 161–184. Wahlberg, M., Møhl, B., Madsen, P.T., 2001. Estimating source position accuracy of a Teloni, V., Johnson, M.P., Miller, J.O.P., Madsen, P.M., 2008. Shallow food for deep di- large-aperture hydrophone array for bioacoustics. J. Acoust. Soc. Am. 109, 397–406. vers: dynamic foraging behavior of male sperm whales in a high latitude habitat. J. Watkins, W.A., Daher, M.A., Fristrup, K.M., Howald, T.J., Notarbartolo di Sciara, G., Exp. Mar. Biol. Ecol. 354, 119–131. 1993. Sperm whales tagged with transponders and tracked underwater by sonar. Mar. Thode, A., Mellinger, D.K., Stienessen, S., Martinez, A., Mullin, K., 2002. Depth-depen- Mamm. Sci. 9, 55–67. dent acoustic features of diving sperm whales (Physeter macrocephalus) in the Gulf of Watwood, S., Miller, P., Johnson, M., Madsen, P., Tyack, P., 2006. Deep-diving foraging Mexico. J. Acoust. Soc. Am. 112, 308–321. behaviour of sperm whales (Physeter macrocephalus). J. Anim. Ecol. 75, 814–825. Tromp, J.J., Jones, P.L., Symonds, J.E., Walker, S.P., Pope, A., Pether, S.M., Afonso, L.O., Weilgart, L.S., Whitehead, H., 1988. Distinctive vocalizations from mature male sperm 2016. Effects of commercial diets and temperature on the growth performance and whales (Physeter macrocephalus). Can. J. Zool. 66, 1931–1937. stress response of hapuku (Polyprion oxygeneios). Aquaculture 452, 128–133. Whitehead, H., 1996. Variation in the feeding success of sperm whales: temporal scale, van der Linde, M., 2009. Long-term Monitoring of Sperm Whales in Kaikoura, New spatial scale and relationship to migrations. J. Anim. Ecol. 65, 429–438. Zealand: Data-management, Abundance, and Population Dynamics. University of Yen, P.P., Sydeman, W.J., Hyrenbach, K.D., 2004. Marine bird and cetacean associations Otago, New Zealand (MSc dissertation). with bathymetric habitats and shallow-water topographies: implications for trophic Van Oevelen, D., Soetaert, K., García, R., De Stigter, H.C., Cunha, M.R., Pusceddu, A., transfer and conservation. J. Mar. Syst. 50, 79–99.
108
References
Aarts GA, MacKenzie M, McConnell B, Fedak M, Matthiopoulos J (2008) Estimating space-use and habitat preference from wildlife telemetry data. Ecography 31:140–160. Abend AG, Smith TD (1995) Differences in ratios of stable isotopes of nitrogen in long-finned pilot whales (Globicephala melas) in the western and eastern North Atlantic. ICES Journal of Marine Science 52:837-841. Aguilar Soto N, Johnson M, Madsen PT, Tyack PL, Bocconcelli A, Fabrizio Borsani J (2006) Does intense ship noise disrupt foraging in deep-diving Cuvier's beaked whale (Ziphius cavirostris)? Marine Mammal Science, 22: 690-699. Ainley DG, Sydeman WJ, Norton J (1995) Upper trophic level predators indicate interannual negative and positive anomalies in the California Current food web. Marine Ecology Progress Series 118:69-79. Airoldi L, Beck MW (2007) Loss, status and trends for coastal marine habitats of Europe. Oceanography and Marine Biology 357-417. CRC Press. Akaike H (1973) Information theory and an extension of the maximum likelihood principle. ‘2nd International Symposium on Information Theory’ (F. Csáki, ed.) 267-281. Akadémiai Kiadó, Budapest. Allen SE, Vindeirinho C, Thomson RE, Foreman MGG, Mackas DL (2001) Physical and biological processes over a submarine canyon during an upwelling event. Canadian Journal of Fisheries and Aquatic Sciences 58:671–684. Allouche O, Tsoar A, Kadmon, R (2006) Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). Journal of Applied Ecology 43:1223–1232. Altabet MA, François R (1994) Sedimentary nitrogen isotopic ratio as a recorder for surface ocean nitrate utilization. Global Biogeochemical Cycles 8:103-116. Anderson CI, Rodhouse PG (2001) Life cycles, oceanography and variability: ommastrephid squid in variable oceanographic environments. Fisheries Research 54:133-143. Anderson DR (2008) Model Based Inference in the Life Sciences: A Primer on Evidence. Springer, New York. Anderson OF, Bagley NW, Hurst RJ, Francis MP, Clark MR, McMillan PJ (1998) Atlas of New Zealand fish and squid distributions from research bottom trawls. NIWA Technical Report 42:303. Andrews KS, Harvey CJ (2013) Ecosystem-level consequences of movement: seasonal variation in the trophic impact of a top predator. Marine Ecology Progress Series 473:247-60. Angel MV, Boxshall GA (1990) Life in the benthic boundary layer: connections to the mid-water and sea floor. Philosophical Transactions of the Royal Society A 33:15–28. Arnould JPY, Cherel Y, Gibbens J, White JG, Littnan CL (2011) Stable isotopes reveal inter-annual and interindividual variation in the diet of female Australian fur seals. Marine Ecology Progress Series 422:291−302. Arregui M, Josa M, Aguilar A, Borrell A (2017) Isotopic homogeneity throughout the skin in small cetaceans. Rapid Communications in Mass Spectrometry 31:1551-1557. Au WWL (1993) The Sonar of Dolphins. Springer, New York. Austin MP (2002) Spatial prediction of species distribution: an interface between ecological theory and statistical modelling. Ecological Modelling 157:101-18. References 204
Azam F, Fenchel T, Field J, Gray J, Meyer-Reil L, Thingstad F (1983) The ecological role of water- column microbes in the sea. Marine Ecology Progress Series 10:257−263. Azam F, Smith DC, Steward GF, Hagström Å (1993) Bacteria-organic matter coupling and its significance for oceanic carbon cycling. Microbial Ecology 28:167–179. Azzellino A, Gaspari S, Airoldi S, Nani B (2008) Habitat use and preferences of cetaceans along the continental slope and the adjacent pelagic waters in the western Ligurian Sea. Deep-Sea Research 55:296–323. Baguley T (2009) Standardized or simple effect size: What should be reported? British Journal of Psychology 100:603-617. Baier CT, Napp JM (2003) Climate-induced variability in Calanus marshallae populations. Journal of Plankton Research 25:771-82. Bailey H, Thompson PM (2009) Using marine mammal habitat modelling to identify priority conservation zones within a marine protected area. Marine Ecology Progress Series 378:279–287. Baker EJ, Clauss M, Clements KD (2016) Selection and intake of algal species in butterfish (Odax pullus; Labridae). Marine Biology 163: 136. Baker JD, Polovina JJ, Howell EA (2007) Effect of variable oceanic productivity on the survival of an upper trophic predator, the Hawaiian monk seal Monachus schauinslandi. Marine Ecology Progress Series 346:277–283. Barber RT, Chavez FP (1983) Biological consequences of El Niño. Science 222:1203–1210. Barbet-Massin M, Jiguet F, Albert CH, Thuller W (2012) Selecting pseudo-absences for species distribution models: how, where and how many? Methods in Ecology and Evolution 3:327-338 Barlow J, Cameron GA (2003) Field experiments show that acoustic pingers reduce marine mammal by-catch in the California drift gill net fishery. Marine Mammal Science 19:265-283. Barlow J, Kahru M, Mitchell BG (2008b) Cetacean biomass, prey consumption and primary production requirements in the California Current ecosystem. Marine Ecology Progress Series 371:285-95. Barlow J, Rankin S, Dawson S (2008a) A guide to constructing hydrophones and hydrophone arrays for monitoring marine mammal vocalizations. NOAA Technical Memorandum NMFS, Technical Report No. 417. Barlow J, Taylor B (2005) Estimates of sperm whale abundance in the Northeastern temperate Pacific from a combined acoustic and visual survey. Marine Mammal Science 21:429–445. Bartoń K (2016) MuMIn: multi-model inference. R package, version 1.15.6. URL: http://r-forge.r- project.org/projects/mumin/. Bates D, Maechler M, Bolker B, Walker S, Christensen RHB, Singmann H, Dai B, Grothendieck G, Green P (2016) Package ‘Lme4’, linear mixed-effects models using Eigen and S4. URL: http: //lme4.r-forge.r-project.org/. Baum JK, Myers RA, Kehler DG, Worm B, Harley SJ, Doherty PA (2003). Collapse and Conservation of Shark Populations in the Northwest Atlantic. Science 299:389–393. Bearzi G, Agazzi S, Gonzalvo J, Costa M, Bonizzoni S, Politi E, Piroddi C, Reeves PR (2008) Overfishing and the disappearance of short-beaked common dolphins from western Greece. Endangered Species Research 5:1-12. Becker EA, Foley DG, Forney KA, Barlow J, Redfern JV, Gentemann CL (2012) Forecasting cetacean abundance patterns to enhance management decisions. Endangered Species Research 16: 97-112. Becker EA, Forney KA, Foley DG, Smith RC, Moore TJ, Barlow J (2014) Predicting seasonal density patterns of California cetaceans based on habitat models. Endangered Species Research 23:1-22. References 205
Beentjes MP, Francis MP (1999) Movement of hapuku (Polyprion oxygeneios) determined from tagging studies. New Zealand Journal of Marine and Freshwater Research 33: 1-12. Behrenfeld MJ, Falkowski PG (1997) Photosynthetic rates derived from satellite-based chlorophyll concentration. Limnology and Oceanography 42:1–20. Bejder L, Samuels AM, Whitehead HA, Gales N, Mann J, Connor R, Heithaus M, Watson‐Capps JA, Flaherty C, Kruetzen M (2006) Decline in relative abundance of bottlenose dolphins exposed to long‐term disturbance. Conservation Biology 20:1791-8. Belkin IM, O'Reilly JE (2009) An algorithm for oceanic front detection in chlorophyll and SST satellite imagery. Journal of Marine Systems 78:319-26. Benoit-Bird KJ, Au WW (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 KJ, Battaile BC, Heppell SA, Hoover B, Irons D, Jones N, Kuletz KJ, Nordstrom CA, Paredes R, Suryan RM, Waluk CM, Trites AW (2013) Prey Patch Patterns Predict Habitat Use by Top Marine Predators with Diverse Foraging Strategies. PLoS One 8(1):e53348. Benoit-Bird KJ, Gilly WF, Au WW, Mate B (2008) Controlled and in situ target strengths of the jumbo squid Dosidicus gigas and identification of potential acoustic scattering sources. The Journal of the Acoustical Society of America 123:1318-1328. Benoit-Bird KJ, Würsig B, McFadden CJ (2004) Dusky Dolphin (Lagenorhynchus obscurus) Foraging in Two Different Habitats: Active Acoustic Detection of Dolphins and Their Prey. Marine Mammal Science 20:215–231. Best PB (1979) Social organization in sperm whales, Physeter microcephalus. Behaviour of marine animals (ed. H.E. Winn and B.L. Olla) New York: Plenum Press: 227–289. Best PB (1983) Sperm whale stock assessments and the relevance of historical whaling records. Reports of the International Whaling Commission Special Issue 5: 41-56. Best PB (1999) Food and feeding of sperm whales Physeter macrocephalus off the west coast of South Africa. South African Journal of Marine Science 21:393-413. Best PB, Canham PAS, Macleod N (1984) Patterns of reproduction in sperm whales, Physeter macrocephalus. Reports of the International Whaling Commission 6:51-79. Beyer HL (2012) Geospatial Modelling Environment (Version 0.7.3.0). URL: http://www.spatialecology.com/gme Bhaskaran B, Mullan A (2003). El Niño-related variations in the southern Pacific atmospheric circulation: model versus observations. Climate Dynamics 20:229-39. Bjørge A (2002) How persistent are marine mammal habitats in an ocean of variability? In: Marine Mammals: Biology and Conservation, 63-91. Springer, Boston, MA. Bjørnstad ON (2008) Ncf: spatial nonparametric covariance functions. R package version 1.1-1. URL: http://onb.ent.psu.edu/onb1/R. Bjørnstad ON, Falck W (2001) Nonparametric spatial covariance functions: estimation and testing. Environmental and Ecological Statistics 8:53–70. Block BA, Jonsen ID, Jorgensen SJ, Winship AJ, Shaffer SA, Bograd SJ, Hazen EL, Foley DG, Breed GA, Harrison AL, Ganong JE (2011) Tracking apex marine predator movements in a dynamic ocean. Nature 475(7354):86-90. Bode A, Alvarez-Ossorio MT, Varela M (2006) Phytoplankton and macrophyte contributions to littoral food webs in the Galician upwelling estimated from stable isotopes. Marine Ecology Progress Series 318:89-102. References 206
Bodin N, Budzinski H, Le Ménach K, Tapie N. (2009) ASE extraction method for simultaneous carbon and nitrogen stable isotope analysis in soft tissues of aquatic organisms. Analytica Chimica Acta 643:54-60. Boecklen WJ, Yarnes CT, Cook BA, James AC (2011) On the use of stable isotopes in trophic ecology. Annual Review of Ecology, Evolution, and Systematics 42:411-440. Bogert BP, Healy MJR, Tukey JW (1963) The quefrency analysis time series for echoes: cepstrum, pseudoautocovariance, cross-cepstrum, and saphe cracking. Rosenblatt, M. (eds). Symposium on Time Series Analysis. Wiley, New York 209-43. Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MHH, White JSS (2009) Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology and Evolution 24:127–135. Bolnick DI, Amarasekare P, Araújo MS, Bürger R and others (2011) Why intraspecific trait variation matters in community ecology. Trends in Ecology and Evolution 26:183−192. Bolnick DI, Svanbäck R, Araújo MS, Persson L (2007) Comparative support for the niche variation hypothesis that more generalized populations also are more heterogeneous. Proceedings of the National Academy of Science of USA 104:10075−10079. Bolnick DI, Svanbäck R, Fordyce JA, Yang LH, David JM, Hulsey CD, Forister ML (2003) The ecology of individuals: incidence and implications of individual specialization. The American Naturalist 161:1−28. Bonjean F, Lagerloef GS (2002) Diagnostic model and analysis of the surface currents in the tropical Pacific Ocean. Journal of Physical Oceanography 32:2938-54. Booth GD, Niccolucci MJ, Schuster EG (1994) Identifying proxy sets in multiple linear regression: an aid to better coefficient interpretation. Research paper INT-470. United States Department of Agriculture, Forest Service, Ogden, USA. Boren LJ, Muller CG, Gemmell NJ (2006) Colony growth and pup condition of the New Zealand fur seal (Arctocephalus forsteri) on the Kaikoura coastline compared with other east coast colonies. Wildlife Research 33:497–505. Borrell A, Abad‐Oliva N, Gómez‐Campos E, Giménez J, Aguilar A (2012) Discrimination of stable isotopes in fin whale tissues and application to diet assessment in cetaceans. Rapid Communications in Mass Spectrometry 26:1596-1602. Borrell A, Velásquez Vacca A, Pinela AM, Kinze C, Lockyer CH, et al. (2013) Stable isotopes provide insight into population structure and segregation in Eastern North Atlantic sperm whales. PLoS One 8(12): e82398. Boschker HTS, Middelburg JJ (2002) Stable isotopes and biomarkers in microbial ecology. FEMS Microbiology Ecology 40:85-95. Bosley KL, Lavelle JW, Brodeur RD, Wakefield WW, Emmett RL, Baker ET, Rehmke KM (2004) Biological and physical processes in and around Astoria submarine Canyon, Oregon, USA. Journal of Marine Systems 50:21-37. Bost CA, Cotté C, Bailleul F, Cherel Y, Charrassin JB, Guinet C, Ainley DG, Weimerskirch H (2009) The importance of oceanographic fronts to marine birds and mammals of the southern oceans. Journal of Marine Systems 78: 363-76. Bost CA, Cotté C, Terray P, Barbraud C, Bon C, Delord K, Gimenez O, Handrich Y, Naito Y, Guinet C, Weimerskirch H (2015) Large-scale climatic anomalies affect marine predator foraging behaviour and demography. Nature communications, 6. Article 8220 (2015). References 207
Boutton TW (1991) Stable carbon isotope ratios of natural materials: II. Atmospheric, terrestrial, marine, and freshwater environments. Carbon isotope techniques 1:173. Bowes RE, Thorp JH (2015) Consequences of employing amino acid vs bulk-tissue stable isotope analysis: a laboratory trophic position experiment. Ecological Society of America 6:1-12. Bowles AE, Smultea M, Würsig B, DeMaster DP, Palka D (1994) Relative abundance and behavior of marine mammals exposed to transmissions from the Heard Island Feasibility Test. Journal of the Acoustical Society of America 96:2469-84. Boyd P, LaRoche J, Gall M, Frew R, McKay RML (1999) The role of iron, light and silicate in controlling algal biomass in sub-Antarctic water SE of New Zealand. Journal of Geophysical Research 104: 13395-13408. Bradford JM (1972) Systematics and Ecology of New Zealand Central East Coast Plankton Sampled at Kaikoura. Bulletin 207. Department of Scientific and Industrial Research, New Zealand. Bradford-Grieve JM, Boyd PW, Chang FH, Chiswel S, Hadfield M, Hall JA, Shushkina, EA (1999) Pelagic ecosystem structure and functioning in the Subtropical Front region east of New Zealand in austral winter and spring 1993. Journal of Plankton Research 21:405-428. Bräger S (1998) Feeding associations between white-fronted terns and Hector's dolphins in New Zealand. The Condor 100:560–562. Bremner G, Johnstone P, Bateson T, Clarke P (2009) Unreported bycatch in the New Zealand west coast South Island hoki fishery. Marine Policy 33:504-512. Brock PM, Hall AJ, Goodman SJ, Cruz M, Acevedo-Whitehouse K (2013) Immune activity, body condition and human-associated environmental impacts in a wild marine mammal. PLoS One 8(6):e67132. Brotons L, Thuiller W, Araújo MB, Hirzel AH (2004) Presence‐absence versus presence‐only modelling methods for predicting bird habitat suitability TL -27. Ecography (Cop) 27 VN- r:437–448. Brough T, Rayment W, Slooten E, Dawson S (2018) Fine scale distribution for a population of New Zealand's only endemic dolphin (Cephalorhynchus hectori) shows long‐term stability of coastal hotspots. Marine Mammal Science. Browning NE, Dold C, Jack IF, Worthy GA (2014) Isotope turnover rates and diet–tissue discrimination in skin of ex situ bottlenose dolphins (Tursiops truncatus). Journal of Experimental Biology 217:214-221. Budge SM, Iverson SJ, Koopman HN (2006) Studying trophic ecology in marine ecosystems using fatty acids: A primer on analysis and interpretation. Marine Mammal Science 22:759–801. Burkholder DA, Heithaus MR, Fourqurean JW, Wirsing A, Dill LM (2013) Patterns of top‐down control in a seagrass ecosystem: could a roving apex predator induce a behaviour‐mediated trophic cascade? Journal of Animal Ecology 82:1192-202. Burnham KP, Anderson DR (2002) Model selection and multimodel inference: a practical information- theoretic approach. Springer, New York, NY Burnham KP, Anderson DR, Huyvaert KP (2011) AIC model selection and multimodel inference in behavioural ecology: some background, observations and comparisons. Behavioral Ecology and Sociobiology 65:23-35. Burton RK, Koch PL (1999) Isotopic tracking of foraging and long-distance migration in northeastern Pacific pinnipeds. Oecologia 119:578–585. Busquets-Vass G, Newsome SD, Calambokidis J, Serra-Valente G, Jacobsen JK, Aguíñiga-García S, et al. (2017) Estimating blue whale skin isotopic incorporation rates and baleen growth rates: References 208
implications for assessing diet and movement patterns in mysticetes. PLoS ONE 12(5): e0177880 Butchart SHM, Walpole M, Collen B, van Strien A, Scharleman JPW, Almond RE, et al (2010) Global biodiversity: indicators of recent declines. Science 328:1164–1168. Calbet A, Saiz E (2005) The ciliate-copepod link in marine ecosystems. Aquatic Microbial Ecology 38:157-167. Canals M, Puig P, de Madron XD, Heussner S, Palanques A, Fabres J (2006) Flushing submarine canyons. Nature 444:354–357. Cantor M, Eguiguren A, Merlen G, Whitehead H (2017). Galápagos sperm whales (Physeter macrocephalus): waxing and waning over three decades. Canadian Journal of Zoology 2017, 95:645-652. Carlier A, Le Guilloux E, Olu K, Sarrazin J, Mastrototaro F, Taviani M, Clavier J (2009) Trophic relationships in a deep Mediterranean cold-water coral bank (Santa Maria di Leuca, Ionian Sea). Marine Ecology Progress Series 397:125-137. Carrillo M, Ritter F (2010) Increasing numbers of ship strikes in the Canary Islands: proposals for immediate action to reduce risk of vessel-whale collisions. Journal of Cetacean Research and Management 11: 131–138. Carroll MJ, Butler A, Owen E, Ewing SR, Cole T, Green JA, Soanes LM, Arnould JP, Newton SF, Baer J, Daunt F (2015) Effects of sea temperature and stratification changes on seabird breeding success. Climate Research 66:75-89. Carroll G, Hedley S, Bannister J, Ensor P, Harcourt R (2014) No evidence for recovery in the population of sperm whale bulls off Western Australia , 30 years post-whaling. Endangered Species Research 24:33–43. Caut S, Angulo E, Courchamp F (2009) Variation in discrimination factors (Δ15N and Δ13C): the effect of diet isotopic values and applications for diet reconstruction. Journal of Applied Ecology 46:443-453. Chen CS, Chiu TS, Huang WB (2007) The spatial and temporal distribution patterns of the Argentine short-finned squid, Illex argentinus, abundance in the Southwest Atlantic and the effects of environmental influences. Zoological Studies 46:111-122. Cherel Y, Hobson KA (2007) Geographical variation in carbon stable isotope signatures of marine predators: tool to investigate their foraging areas in the Southern Ocean. Marine Ecology Progress Series 329:281−287. Cherel Y, Hobson KA, Weimerskirch H (2000) Using stable-isotope analysis of feathers to distinguish moulting and breeding origins of seabirds. Oecologia 122:155–162. Cherel Y, Weimerskirch H (1999) Spawning cycle of onychoteuthid squids in the southern Indian Ocean: new information from seabird predators. Marine Ecology Progress Series 188:93-104. Childerhouse SJ, Dawson SM, Slooten E (1995) Abundance and seasonal residence of sperm whales at Kaikōura, New Zealand. Canadian Journal of Zoology 73:723-731. Childerhouse SJ, Dawson SM, Slooten E (1996) Stability of fluke marks used in individual photo- identification of male sperm whales at Kaikoura, New Zealand. Marine Mammal Science 12:447-51. Chilvers BL, Corkeron PJ, Puotinen ML (2003) Influence of trawling on the behaviour and spatial distribution of Indo-Pacific bottlenose dolphins (Tursiops aduncus) in Moreton Bay, Australia. Canadian Journal of Zoology 81:1947-1955. References 209
Chiswell SM (1996) Variability in the southland current, New Zealand. New Zealand Journal of Marine and Freshwater Research 30:1-17. Chiswell SM, Bostock HC, Sutton PJH, Williams MJM (2015) Physical oceanography of the deep seas around New Zealand: a review. New Zealand Journal of Marine and Freshwater Research 49:286– 317. Chiswell SM, Schiel DR (2001) Influence of along-shore advection and upwelling on coastal temperature at Kaikoura Peninsula, New Zealand. New Zealand Journal of Marine and Freshwater Research Res 35:307–317. Chiswell SM, Sutton PJ (1998) A deep eddy in the Antarctic Intermediate Water north of the Chatham Rise. Journal of Physical Oceanography 28:535-40. Chiswell SM, Zeldis JR, Hadfield MG, Pinkerton MH (2017) Wind-driven upwelling and surface chlorophyll blooms in Greater Cook Strait. New Zealand Journal of Marine and Freshwater Research 51:465-89. Christal J, Whitehead H (1997) Aggregations of mature male sperm whales on the Galapagos Islands breeding ground. Marine Mammal Science 13:59-69. Christensen JT, Richardson K (2008) Stable isotope evidence of long-term changes in the North Sea food web structure. Marine Ecology Progress Series 368:1-8. Clark MR (1985) The food and feeding of seven fish species from the Campbell Plateau, New Zealand. New Zealand Journal of Marine and Freshwater Research 19:339-363. Clark MR, Anderson OF, Dunn MR (2003) New Zealand Fishery Assessment Report No. 2003/60. Ministry of Fisheries, Wellington, New Zealand. Clarke MR (1956) Sperm whales of the Azores. Discovery Report 28:237-298. Clarke MR (1977) Beaks, nets and numbers. Symposia of the Zoological Society of London 38:89-126. Clarke MR, Martins HR, Pascoe, P (1993) The diet of sperm whales (Physeter macrocephalus Linnaeus 1758) off the Azores. Philosophical Transactions of the Royal Society of London B: Biological Sciences 339:67-82. Coakes AK, Whitehead H (2004) Social structure and mating system of sperm whales off Northern Chile. Canadian Journal of Zoology 82:1360-1369. Cohen JE (1994) Marine and continental food webs: three paradoxes? Philosophical Transactions of the Royal Society of London B, 343:57–69. Collins MA, Pierce GJ, Boyle PR (1997) Population indices of reproduction and recruitment in Loligo forbesi (Cephalopoda: Loliginidae) in Scottish and Irish waters. Journal of Applied Ecology 34:778–786. Connolly RM, Guest MA, Melville AJ, Oakes JM (2004) Sulfur stable isotopes separate producers in marine food-web analysis. Oecologia 138:161-7. Cornwall CE, Phillips NE, McNaught DC (2009) Feeding preferences of the abalone Haliotis iris in relation to macroalgal species, attachment, accessibility and water movement. Journal of Shellfish Research 28:589-597. Cranford W (1999) The Sperm Whale’s Nose: Sexual Selection on a Grand Scale? Marine Mammal Science 15:1133–1157. Crespi-Abril AC, Morsan EM, Williams GN, Gagliardini DA (2013) Spatial distribution of Illex argentinus in San Matias Gulf (Northern Patagonia, Argentina) in relation to environmental variables: A contribution to the new interpretation of the population structuring. Journal of Sea Research 77:22–31. References 210
Croll DA, Marinovic B, Benson S, Chavez FP, Black N, Ternullo R, Tershy BR (2005) From wind to whales: trophic links in a coastal upwelling system. Marine Ecology Progress Series 289:117–130. Curtin S (2003) Whale-Watching in Kaikoura: Sustainable Destination Development? Journal of Ecotourism 2:173-195. D'Agrosa C, Lennert‐Cody CE, Vidal O (2000) Vaquita bycatch in Mexico's artisanal gillnet fisheries: driving a small population to extinction. Conservation Biology 14:1110-9. Dall’Olmo G, Gitelson AA, Rundquist DC, Leavitt B, Barrow T, Holz, JC (2005) Assessing the potential of SeaWiFS and MODIS for estimating chlorophyll concentration in turbid productive waters using red and near-infrared bands. Remote Sensing of Environment 96:176−187. Dalley DD, McClatchie S (1989) Functional feeding morphology of the euphausiid Nyctiphanes australis. Marine Biology 101:195–203. Danovaro R, Gambi C, Dell’Anno A, Corinaldesi C, Fraschetti S, Vanreusel A, Vincx M, Gooday AJ (2008) Exponential decline of deep-sea ecosystem functioning linked to benthic biodiversity loss. Current Biology 18:1–8. Dauby P, Poulicek M (1995) Methods for removing epiphytes from seagrasses: SEM observations on treated leaves. Aquatic Botany 52:217–228. Dave AC, Lozier MS (2015) The impact of advection on stratification and chlorophyll variability in the equatorial Pacific. Geophysical Research Letters 42: 4523–4531. Davenport SR, Bax NJ (2002) A trophic study of a marine ecosystem off southeastern Australia using stable isotopes of carbon and nitrogen. Canadian Journal of Fisheries and Aquatic Sciences 59:514- 530. Davidson AD, Boyer AG, Kim H, Pompa-Mansilla S, Hamilton MJ, Costa DP, Brown JH (2012) Drivers and hotspots of extinction risk in marine mammals. Proceedings of the National Academy of Sciences 109:3395-3400. Dawson SM (1990) Building the “CETOS” directional hydrophone. Available at: http://whaledolphintrust.org.nz/wp-content/uploads/Building-Directional-HPs.pdf. Dawson SM, Chessum CJ, Hunt PJ, Slooten E (1995) An inexpensive, stereo-photographic technique to measure sperm whales from small boats. Reports of the International Whaling Commission 45:431–436. Dayton PK, Tegner MJ, Edwards PB, Riser KL (1998) Sliding baselines, ghosts, and reduced expectations in kelp forest communities. Ecological Applications 8:309-22. de Boyer Montégut C, Madec G, Fischer AS, Lazar A, Iudicone D (2004) Mixed layer depth over the global ocean: An examination of profile data and a profile-based climatology. Journal of Geophysical Research 109:1–20. De Leo FC, Ogata B, Sastri AR, Heesemann M, Mihály S, Galbraith M, Morley MG (2018) High- frequency observations from a deep-sea cabled observatory reveal seasonal overwintering of Neocalanus spp. in Barkley Canyon, NE Pacific: insights into particulate organic carbon flux. Progress in Oceanography. De Leo FC, Smith CR, Rowden AR, Bowden DA, Clark MR (2010) Submarine canyons: hotspots of benthic biomass and productivity in the deep sea. Proceedings of the Royal Society B 277:2783- 2792. De Stephanis R, Cornulier T, Verborgh P, Sierra JS, Gimeno NP, Guinet C (2008) Summer spatial distribution of cetaceans in the Strait of Gibraltar in relation to the oceanographic context. Marine Ecology Progress Series 353:275-288. References 211
DeMaster DP, Fowler CW, Perry SL, Richlen MF (2001) Predation and Competition: the Impact of Fisheries on Marine-Mammal Populations Over the Next One Hundred Years. Journal of Mammalogy 82:641–651. DeNiro MJ, Epstein S (1977) Mechanism of carbon isotope fractionation associated with lipid synthesis. Science 197: 261-263. DeNiro MJ, Epstein S (1978) Influence of diet on the distribution of carbon isotopes in animals. Geochim Cosmochim Acta 42:495−506. DeNiro MJ, Epstein S (1981) Influence of diet on the distribution of nitrogen isotopes in animals. Geochim Cosmochim Acta 45:341−351. Derville S, Torres LG, Iovan C, Garrigue C (2018) Finding the right fit: Comparative cetacean distribution models using multiple data sources and statistical approaches. Biodiversity Research:1–17. DOI: 10.1111/ddi.12782. Diaz RJ, Rosenberg R (2008) Spreading dead zones and consequences for marine ecosystems. Science 321: 926-929. Díaz‐Gamboa RE, Gendron D, Busquets‐Vass G (2017) Isotopic niche width differentiation between common bottlenose dolphin ecotypes and sperm whales in the Gulf of California. Marine Mammal Science 34:440-457. DOC (2014) ‘Te Rohe o Te Whānau Puha’ Kaikōura Whale Sanctuary Restrictions Notice. Department of Conservation. Accessible at: http://www.doc.govt.nz/Documents/about- doc/role/legislation/kaikoura-whale-sanctuary-notice-2014.pdf Doniol-Valcroze T, Lesage V, Giard J, Michaud R (2012) Challenges in marine mammal habitat modelling: Evidence of multiple foraging habitats from the identification of feeding events in blue whales. Endanger Species Research 17:255–268. Douglas LA, Dawson SM, Jaquet N (2005) Click rates and silences of sperm whales at Kaikoura, New Zealand. Journal of the Acoustical Society of America 118:523–529. Duarte C, Navarro JM, Acuña K, Gómez I (2010) Feeding preferences of the sandhopper Orchestoidea tuberculata: the importance of algal traits. Hydrobiologia, 651:291-303. Dulvy NK, Rogers SI, Jennings S, Stelzenmüller V, Dye SR, Skjoldal HR (2008) Climate change and deepening of the North Sea fish assemblage: a biotic indicator of warming seas. Journal of Applied Ecology 45:1029-39. Dunn MR, Connell AM, Forman J, Stevens DW, Horn PL (2010) Diet of two large sympatric teleosts, the ling (Genypterus blacodes) and hake (Merluccius australis). PLoS One 5 e13647. Dunne JA, Williams RJ, Martinez ND (2004) Network structure and robustness of marine food webs. Marine Ecology Progress Series 273: 291-302. Edwards M, Richardson AJ (2004) Impact of climate change on marine pelagic phenology and trophic mismatch. Nature 430:881-884. Elith J, Graham CH, Anderson RP, Dudik M, Ferrier S, Guisan A, Hijmans RJ, Huettmann F, Leathwick JR, Lehmann A, Li J, Lohmann LG, Loiselle BA, Manion G, Moritz C, Nakamura M, Nakazawa Y, Overton J. Peterson AT, Phillips SJ, Richardson KS, Scachetti-Pereira R, Schapire RE, Soberón J, Williams S, Wisz MS, Zimmermann NE (2006) Novel methods improve prediction of species distributions from occurrence data. Ecography 29: 129-151. Embling CB, Gillibrand PA, Gordon J, Shrimpton J, Stevick PT, Hammond PS (2010) Using habitat models to identify suitable sites for marine protected areas for harbour porpoises (Phocoena phocoena). Biological Conservation 143:267–279. References 212
Epping E, van der Zee C, Soetaert K, Helder W (2002) On the oxidation and burial of organic carbon in sediments of the Iberian margin and Nazaré Canyon (NE Atlantic). Progress in Oceanography 52:399–431. Estes JA, Crooks K, Holt R (2001) Predators, ecological role. Encyclopedia of Biodiversity 4:857-878. S. Levin, ed., Academic Press, San Diego, CA. Estes JA, Demaster DP, Doak DF, Brownell RL, Williams TM (2006) Whales, Whaling and Ocean Ecosystems. University of California Press. Estes JA, Duggins DO (1995) Sea otters and kelp forests in Alaska: generality and variation in a community ecological paradigm. Ecological Monographs 65:75-100. Estes JA, JF Palmisano (1974) Sea otters: their role in structuring nearshore communities. Science 185:1058–1060. Estes JA, Riedman ML, Staedler MM, Tinker MT, Lyon BE (2003) Individual variation in prey selection by sea otters: patterns, causes, and implications. Journal of Animal Ecology 72:144–155. Estes JA, Terborgh J, Brashares JS, Power ME, Berger J, Bond WJ, Carpenter SR, Essington TE, Holt RD, Jackson JB, Marquis RJ (2011) Trophic downgrading of planet Earth. Science 333:301-6. Estrada M (1991) Phytoplankton assemblages across a NW Mediterranean front: changes from winter mixing to spring stratification. In: Ros JD, Prat N editors. Homage to Ramon Margalef; or Why is Such a Pleasure in Studying Nature: Oecologia Aquatica 10:157–185. Evans K, Hindell MA (2004) The diet of sperm whales (Physeter macrocephalus) in southern Australian waters. ICES Journal of Marine Science 61:1313-1329. Fais A, Aguilar Soto N, Johnson M, Pérez-González C, Miller PJO, Madsen PT (2015) Sperm whale echolocation behaviour reveals a directed, prior-based search strategy informed by prey distribution. Behavioral Ecology and Sociobiology 69:663–674. Fais A, Lewis TP, Zitterbart DP, Álvarez O, Tejedor A (2016) Abundance and Distribution of Sperm Whales in the Canary Islands: Can Sperm Whales in the Archipelago Sustain the Current Level of Ship-Strike Mortalities? PLoS One 11:1–16. Fanelli E, Papiol V, Cartes JE, Rumolo P, Brunet C, Sprovieri M (2011) Food web structure of the epibenthic and infaunal invertebrates on the Catalan slope (NW Mediterranean): evidence from δ¹³C and δ¹⁵N analysis. Deep-Sea Research Part I: Oceanographic Research Papers 58:98−109. Fanelli E, Papiol V, Cartes JE, Rumolo P, López-Pérez C (2013) Trophic webs of deep-sea megafauna on mainland and insular slopes of the NW Mediterranean: a comparison by stable isotope analysis. Marine Ecology Progress Series 490:199-221. Farmer NA, Baker K, Zeddies DG, Denes SL, Noren DP, Garrison LP, Machernis A, Fougères EM, Zykov M (2018) Population consequences of disturbance by offshore oil and gas activity for endangered sperm whales (Physeter macrocephalus). Biological Conservation 227:189-204. Farrugio H (2012) A refugium for the spawners of exploited Mediterranean marine species: the canyons of the continental slope of the Gulf of Lion. In: Mediterranean Submarine Canyons: Ecology and Governance, ed Würtz M (Gland; Málaga: IUCN) 45-48. Fenchel T (2008) The microbial loop–25 years later. Journal of Experimental Marine Biology and Ecology 366: 99-103. Fennell S, Rose G (2015) Oceanographic influences on Deep Scattering Layers across the North Atlantic. Deep-Sea Research Part I: Oceanographic Research Papers 105:132–141. Fenton GE, Ritz DA (1988) Changes in carbon and hydrogen stable isotope ratios of macroalgae and seagrass during decomposition. Estuarine, Coastal and Shelf Science 26: 429-436. References 213
Ferguson MC, Barlow J, Fiedler P, Reilly SB, Gerrodette T (2006b) Spatial models of delphinid (family Delphinidae) encounter rate and group size in the eastern tropical Pacific Ocean. Ecological Modelling 193:645–662. Ferguson MC, Barlow J, Reilly SB, Gerrodette T (2006a) Predicting Cuvier’s (Ziphius cavirostris) and Mesoplodon beaked whale population density from habitat characteristics in the eastern tropical Pacific Ocean. Journal of Cetacean Research and Management 7: 287-299. Fernandez-Arcaya U, Ramirez-Llodra E, Aguzzi J, Allcock AL, Davies JS, Dissanayake A, Harris P, Howell K, Huvenne VAI, Macmillan-Lawler M, Martín J, Menot L, Nizinski M, Puig P, Rowden AA, Sanchez F, Beld IMJ Van den (2017) Ecological Role of Submarine Canyons and Need for Canyon Conservation: A Review. Frontiers in Marine Science 4:1–26. Focken U, Becker K (1998) Metabolic fractionation of stable carbon isotopes: implications of different proximate compositions for studies of the aquatic food webs using δ13C data. Oecologia 115:337-343. Forcada J, Trathan PN (2009) Penguin responses to climate change in the Southern Ocean. Global Change Biology 15:1618-30. Forcada J, Trathan PN, Reid K, Murphy EJ (2005) The effects of global climate variability in pup production of Antarctic fur seals. Ecology 86:2408-2417. Forney KA (2000) Environmental Models of Cetacean Abundance: Reducing Uncertainty in Population Trends. Conservation Biology 14:1271-1286. Forney KA, Becker EA, Foley DG, Barlow J, Oleson EM (2015) Habitat-based models of cetacean density and distribution in the central North Pacific. Endangered Species Research 27:1-20. Forney KA, Ferguson MC, Becker EA, Fiedler PC, Redfern J V., Barlow J, Vilchis IL, Ballance LT (2012) Habitat-based spatial models of cetacean density in the eastern Pacific Ocean. Endangered Species Research 16:113–133. Forney KA, Southall BL, Slooten E, Dawson S, Read AJ, Baird RW, Brownell Jr RL (2017) Nowhere to go: noise impact assessments for marine mammal populations with high site fidelity. Endangered Species Research 8:391-413. France RL (1995) Carbon-13 enrichment in benthic compared to planktonic algae: foodweb implications. Marine Ecology Progress Series 124:307–312. France RL, Peters RH (1997) Ecosystem differences in the trophic enrichment of 13C in aquatic food webs. Canadian Journal of Fisheries and Aquatic Sciences 54:1255-1258. Franks PJS (1992) Sink or swim: accumulation of biomass at fronts. Marine Ecology Progress Series 82:1– 12. Friedlaender AS, Halpin PN, Qian SS, Lawson GL, Wiebe PH, Thiele D, Read AJ (2006) Whale distribution in relation to prey abundance and oceanographic processes in shelf waters of the Western Antarctic Peninsula. Marine Ecology Progress Series 317:297-310. Fry B (2006) Stable isotope ecology. Springer, New York, NY. Fry B, Sherr EB (1989) δ¹³C measurements as indicators of carbon flow in marine and freshwater ecosystems. In: Stable isotopes in ecological research (196-229). Springer, New York, NY. Gannier A, Praca E (2007) SST fronts and the summer sperm whale distribution in the north-west Mediterranean Sea. Journal of the Marine Biological Association of the United Kingdom 87:187–193. Gaskin DE (1968) Analysis of sightings and catches of sperm whales (physeter catodon L.) in the Cook strait area of New Zealand. New Zealand Journal of Marine and Freshwater Research 2:260-272. Gaskin DE (1973) Sperm whales in the Western South Pacific. New Zealand Journal of Marine and Freshwater Research 7:1-20. References 214
Gaskin DE, Cawthorn MW (1967) Diet and feeding habits of the sperm whale (Physeter catadon L.) in the Cook Strait region of New Zealand. New Zealand Journal of Marine and Freshwater Research 2:156-179. Gaskin DE, Cawthorn MW (1973) Sperm whales (Physeter catodon L.) in the Cook Strait Region of New Zealand: Some data on age, growth, and mortality. Norwegian Journal of Zoology 21:45–50. Gelman A (2008) Scaling regression inputs by dividing by two standard deviations. Statistics in Medicine 27: 2865-2873. Gero S, Whitehead H (2016) Critical Decline of the Eastern Caribbean Sperm Whale Population. PLoS One:1–11. Gibbs MT, Shaw AGP (2002) The use of AVHRR data in a pilot study for investigating and modelling regional ocean circulation. International Journal of Remote Sensing 23:199-204. Gilles A, Viquerat S, Becker EA, Forney KA, Geelhoed SCV, Haelters J, Nabe-Nielsen J, Scheidat M, Siebert U, Sveegaard S, van Beest FM, van Bemmelen R, Aarts G (2016) Seasonal habitat-based density models for a marine top predator, the harbor porpoise, in a dynamic environment. Ecosphere 7(6):e01367. 10.1002/ecs2.1367. Gillett NP, Kell TD, Jones PD (2006) Regional climate impacts of the Southern Annular Mode. Geophysical Research Letters 33(23). Giménez J, Ramírez F, Almunia J, Forero MG, de Stephanis R (2016) From the pool to the sea: Applicable isotope turnover rates and diet to skin discrimination factors for bottlenose dolphins (Tursiops truncatus). Journal of Experimental Marine Biology and Ecology 475:54-61. Ginzburg L, Jensen C 2004 Rules of thumb for judging ecological theories. Trends in Ecology & Evolution 19:121-126. Giorli G, Au WWL (2017) Combining passive acoustics and imaging sonar techniques to study sperm whales’ foraging strategies. The Journal of the Acoustical Society of America 142:1428-1431. Giorli G, Neuheimer A, Copeland A, Au WWL (2016) Temporal and spatial variation of beaked and sperm whales foraging activity in Hawaii, as determined with passive acoustics. The Journal of the Acoustical Society of America 140:2333–2343. Goericke R, Fry B (1994) Variations in marine plankton d13C with latitude, temperature, and dissolved CO2 in the world ocean. Global Biogeochemical Cycles 8:85–90. Gong DY, Wang SW (1999) Definition of Antarctic Oscillation Index. Geophysical Research Letters 26:459–462. Gons HJ (1999) Optical teledetection of chlorophyll a in turbid inland waters. Environmental Science & Technology 33:1127−1132. Goold JC, Jones SE (1995) Time and frequency domain characteristics of sperm whale clicks. The Journal of the Acoustical Society of America 98:1279-1291. Gordon JCD (1991) Evaluation of a method for determining the length of sperm whales (Physeter catodon) from their vocalizations. Journal of Zoology 224:301–314. Grace JB, Johnson DJ, Lefcheck JS, Byrnes JE (2018) Quantifying relative importance: computing standardized effects in models with binary outcomes. Ecosphere 9(6), e02283. Grady D (1982) The Perano Whalers of the Cook Strait. 1911-1964. A.H. and A.W. Reed LTD, Wellington. Graham BS, Koch PL, Newsome SD, McMahon KW, Aurioles D (201) Using isoscapes to trace the movements and foraging behavior of top predators in oceanic ecosystems. In: ‘Isoscapes’ (pp. 299-318). Springer, Dordrecht. References 215
Gregr EJ, Trites AW (2001) Predictions of critical habitat for five whale species in the waters of coastal British Columbia. Canadian Journal of Fisheries and Aquatic Sciences 58:1265–1285. Greig MJ, Gilmour AE (1992) Flow through the Mernoo Saddle, New Zealand. New Zealand Journal of Marine and Freshwater Research 26:155–165. Griffin B (1999) Sperm whale distributions and community ecology associated with a warm-core ring off Georges Bank. Marine Mammal Science 15:33–51. Growcott A, Miller B, Sirguey P, Slooten E, Dawson S (2011) Measuring body length of male sperm whales from their clicks: The relationship between inter-pulse intervals and photogrammetrically measured lengths. The Journal of the Acoustical Society of America 130:568- 573. Grubbs RD, Carlson JK, Romine JG, Curtis TH, McElroy WD, McCandless CT, Cotton CF, Musick JA (2016) Critical assessment and ramifications of a purported marine trophic cascade. Scientific Reports 6:20970. Grueber, C.E., Nagakawa, S., Laws, R.J. and Jamieson, I.G. (2011) Multimodel inference in ecology and evolution: challenges and solutions. Journal of Evolutionary Biology, 24: 699-711. Guerra M, Dawson SM, Sabadel AJM, Slooten E, Somerford T, Williams R, Wing L, Rayment WJ (2018) Effects of the Kaikōura earthquake on sperm whales (Physeter macrocephalus). New Zealand Aquatic Environment and Biodiversity Report, Fisheries New Zealand. Available from Fisheries New Zealand ([email protected]) or M. Guerra ([email protected]). Guerra M, Hickmott L, van der Hoop J, Rayment W, Leunissen E, Slooten E, Moore M (2017) Diverse foraging strategies by a marine top predator: sperm whales exploit pelagic and demersal habitats in the Kaikōura submarine canyon. Deep Sea Research Part I 128:98–108. Guest MA, Hirst AJ, Nichols PD, Frusher SD (2010) Multi-scale spatial variation in stable isotope and fatty acid profiles amongst temperate reef species: implications for design and interpretation of trophic studies. Marine Ecology Progress Series 410:25–41. Guinet C, Roux JP, Bonnet M, Mison V (1998) Effect of body size, body mass, and body condition on reproduction of female South African fur seals (Arctocephalus pusillus) in Namibia. Canadian Journal of Zoology 76:1418-24. Guinet C, Xing X, Walker E, Monestiez P, Marchand S, Picard B, Jaud T, Authier M, Cotté C, Dragon AC, Diamond E, Antoine D, Lovell P, Blain S, D’Ortenzio F, Claustre H (2012) Calibration procedures and first dataset of Southern Ocean chlorophyll a profiles collected by elephant seals equipped with a newly developed CTD-fluorescence tags. Earth Syst. Sci. Data 5:15–29. Guisan A, Edwards TC, Hastie T (2002) Generalized linear and generalized additive models in studies of species distributions: setting the scene. Ecological modelling 157:89-100. Guisan A, Zimmermann NE (2000) Predictive habitat distribution models in ecology. Ecological modelling 135: 147-186. Haberman KL, Ross RM, Quetin LB (2003) Diet of the Antarctic krill (Euphausia superba Dana): II. Selective grazing in mixed phytoplankton assemblages. Journal of Experimental Marine Biology and Ecology, 283:97-113. Hall LS, Krausman PR, MorrisonML (1997) The habitat concept and a plea for standard terminology. Wildlife Society Bulletin 173–182. Hall-Aspland SA, Rogers TL, Canfield RB (2005) Stable carbon and nitrogen isotope analysis reveals seasonal variation in the diet of leopard seals. Marine Ecology Progress Series 305:249-259. Halpern BS (2003) The impact of marine reserves: Do reserves work and does reserve size matter? Ecological Applications 13:117-137. References 216
Halpern BS, Selkoe KA, Micheli F, Kappel CV (2007) Evaluating and ranking the vulnerability of global marine ecosystems to anthropogenic threats. Conservation Biology 21:1301–1315 Hamilton LJ (2006) Structure of the Subtropical Front in the Tasman Sea. Deep Sea Research Part I: Oceanographic Research Papers 53:1989–2009. Hammond PS, Macleod K, Berggren P, Borchers DL, Burt L, Cañadas A, Desportes G, Donovan GP, Gilles A, Gillespie D, Gordon J (2013) Cetacean abundance and distribution in European Atlantic shelf waters to inform conservation and management. Biological Conservation 164:107- 22. Hansen JE, Martos P, Madirolas A (2001) Relationship between spatial distribution of the Patagonian stock of Argentine anchovy, Engraulis anchoita, and sea temperatures during late spring to early summer. Fisheries Oceanography 10:193–206. Harraway J (1993) Introductory statistical methods and the analysis of variance. University of Otago Press. Harrold C, Light K, Lisin S (1998) Organic enrichment of submarine-canyon and continental-shelf benthic communities by macroalgal drift imported from nearshore kelp forests. Limnology and Oceanography 43:669–678. Hastie GD, Wilson BEN, Wilson LJ, Parsons KM, Thompson PM (2004) Functional mechanisms underlying cetacean distribution patterns: hotspots for bottlenose dolphins are linked to foraging. Marine Biology 144:397-403. Hastie TJ & Tibshirani RJ (1990) Generalized additive models. Chapman & Hall, Boca Raton, FL. Hays GC 2003. A review of the adaptive significance and ecosystem consequences of zooplankton diel vertical migrations. Hydrobiologia 503:163–170. Hazen EL, Johnston DW (2010) Meridional patterns in the deep scattering layers and top predator distribution in the central equatorial Pacific. Fisheries Oceanography 19:427–33. Hazen EL, Jorgensen S, Rykaczewski RR, Bograd SJ, Foley DG, Jonsen ID, Shaffer SA, Dunne JP, Costa DP, Crowder LB, Block BA (2013b) Predicted habitat shifts of Pacific top predators in a changing climate. Nature Climate Change 3:234-238. Hazen EL, Nowacek DP, Laurent St L, Halpin PN, Moretti DJ (2011) The relationship among oceanography, prey fields, and beaked whale foraging habitat in the tongue of the ocean. PLoS One 6(4):e19269. Hazen EL, Suryan RM, Santora JA, Bograd SJ, Watanuki Y, Wilson RP (2013a) Scales and mechanisms of marine hotspot formation. Marine Ecology Progress Series 487:177–183. He S, Mazumdar S, Arena VC (2006) A comparative study of the use of GAM and GLM in air pollution research. Environmetrics 17:81-93. Heath RA (1971) Hydrology and circulation in central and southern Cook Strait, New Zealand. New Zealand Journal of Marine and Freshwater Research 5:178–199. Heath RA (1972a) Wind-derived water motion off the East Coast of New Zealand. New Zealand Journal of Marine and Freshwater Research 6:352-364. Heath RA (1972b) The Southland Current. New Zealand Journal of Marine and Freshwater Research 6:497- 533. Heath RA (1975): Oceanic circulation off the East Coast of New Zealand. New Zealand Oceanographic Institute, Memoir No. 55. Wellington, A. R. Shearer, Government Printer. Heath RA (1981) Oceanic fronts around southern New Zealand. Deep-Sea Research 28A:547–560. Heath RA (1985) A review of the physical oceanography of the seas around New Zealand. New Zealand Journal of Marine and Freshwater Research 19:79–124. References 217
Heithaus MR, Frid A, Wirsing AJ, Worm B (2008) Predicting ecological consequences of marine top predator declines. Trends in Ecology and Evolution 23:202–210. Hepburn CD, Pritchard DW, Cornwall CE, McLeod RJ, Beardall J, Raven JA, Hurd CL (2011) Diversity of carbon use strategies in a kelp forest community: implications for a high CO2 ocean. Global Change Biology 17:2488-2497. Herrera JM, Ploquin EF, Rodriguez-Perez J, Obeso JR (2014) Determining habitat suitability for bumblebees in a mountain system: a baseline approach for testing the impact of climate change on the occurrence and abundance of species. Journal of Biogeography 41:700–712. Hobson KA, Ambrose Jr WG, Renaud PE (1995). Sources of primary production, benthic-pelagic coupling, and trophic relationships within the Northeast Water Polynya: insights from δ13C and δ15N analysis. Marine Ecology Progress Series 1-10. Hobson KA, Piatt JF, Pitocchelli J (1994) Using stable isotopes to determine seabird trophic relationships. Journal of Animal Ecology 63:786–798. Hobson KA, Sease JL, Merrick RL, Piatt JF (1997) Investigating trophic relationships of pinnipeds in Alaska and Washington using stable isotope ratios of nitrogen and carbon. Marine Mammal Science 13:114–132. Hoegh-Guldberg O, Bruno JF (2010) The impact of climate change on the world’s marine ecosystems. Science 328: 1523-1528. Hooker SK, Cañadas A, Hyrenbach KD, Corrigan C, Polovina JJ, Reeves RR (2011) Making protected area networks effective for marine top predators. Endangered Species Research 13:203–218. Hooker SK, Iverson SJ, Ostrom P, Smith SC (2001) Diet of northern bottlenose whales inferred from fatty-acid and stable-isotope analyses of biopsy samples. Canadian Journal of Zoology 79:1442- 54. Hooker SK, Whitehead H, Gowans S (1999) Marine protected area design and the spatial and temporal distribution of cetaceans in a submarine canyon. Conservation Biology 13:592–602. Hucke-Gaete R, Moreno CA. Arata J (2004) Operational interactions of sperm whales and killer whales with the Patagonian toothfish industrial fishery off southern Chile. CCAMLR Science 11: 127- 140. Hussey NE, MacNeil MA, McMeans BC, Olin JA, Dudley SF, Cliff G, Fisk AT (2014) Rescaling the trophic structure of marine food webs. Ecology letters 17:239-250. Hutchings JA, Reynolds JD (2004) Marine fish population collapses: consequences for recovery and extinction risk. AIBS Bulletin 54:297-309. Huxel GR, Polis GA, Holt RD (2004) At the frontier of the integration of food web ecology and landscape ecology. In: Polis GA, Power ME, Huxel GR (eds) Food webs at the landscape level. University of Chicago Press, Chicago. Hyrenbach KD, Forney K, Dayton PK (2000) Marine protected areas and ocean basin management. Aquatic Conservation Marine and Freshwater Ecosystems 10:437–458. Iken K, Bluhm BA, Gradinger R (2005) Food web structure in the high Arctic Canada Basin: evidence from δ13C and δ15N analysis. Polar Biology 28:238-249. IPCC (2014) Fifth assessment report of the United Nations Intergovernmental Panel on Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III (eds Core Writing Team: Pachauri RK & Meyer L (IPCC 2014). IPCC, Geneva, Switzerland, 151 pp. Available at: http://ar5-syr.ipcc.ch/ipcc/ipcc/resources/pdf/IPCC_SynthesisReport.pdf Isojunno S, Ferndandes MC, Gordon J (2011) Effects of whale watching on underwater acoustic behaviour of sperm whales in the Kaikōura Canyon area. In: Effects of tourism on the References 218
behaviour of sperm whales inhabiting the Kaikōura Canyon. Eds: Markowitz TM, Richter C, Gordon J. Report for the Department of Conservation of NZ. Ito M, Minami H, Tanaka Y, Watanuki Y (2009) Seasonal and interannual oceanographic changes induce diet switching in a piscivorous seabird. Marine Ecology Progress Series 393:273–284. IUCN (2017) 2017 IUCN Red List of Threatened Species. www.iucnredlist.org (viewed 10-11-2017). Iverson SJ, Field C, Bowen WD, Blanchard W (2004) Quantitative fatty acid signature analysis: a new method of estimating predator diets. Ecology Monographs 74: 211–235. Jack L, Wing SR (2011) Individual variability in trophic position and diet of a marine omnivore is linked to kelp bed habitat. Marine Ecology Progress Series 443: 129-139. Jackson GD (1997) Age, growth and maturation of the deepwater squid Moroteuthis ingens (Cephalopoda: Onychoteuthidae) in New Zealand waters. Polar Biology 17:268-74. Jackson GD (2001). Confirmation of winter spawning of Moroteuthis ingens (Cephalopoda: Onychoteuthidae) in the Chatham Rise region of New Zealand. Polar Biology 24:97-100. Jackson GD, McKinnon JF, Lalas C, Ardern R, Buxton NG (1998) Food spectrum of the deepwater squid Moroteuthis ingens (Cephalopoda: Onychoteuthidae) in New Zealand waters. Polar Biology 20:56-65. Jackson GD, Shaw AG, Lalas C (2000) Distribution and biomass of two squid species off southern New Zealand: Nototodarus sloanii and Moroteuthis ingens. Polar Biology 23:699-705. Jackson GD, Wotherspoon S, McGrath-Steer BL (2005) Temporal population dynamics in arrow squid Nototodarus gouldi in southern Australian waters. Marine Biology 146:975-83. Jackson JB, Kirby MX, Berger WH, Bjorndal KA, Botsford LW, Bourque BJ, Bradbury RH, Cooke R, Erlandson J, Estes JA, Hughes TP (2001) Historical overfishing and the recent collapse of coastal ecosystems. Science 293:629-37. Jaquet N (1996) How spatial and temporal scales influence understanding of sperm whale distribution: a review. Mammal Review 26:51–65. Jaquet N, Dawson S, Douglas L (2001) Vocal behavior of male sperm whales: Why do they click? The Journal of Acoustical Society of America 109:2254–2259. Jaquet N, Dawson S and Slooten E (2000) Seasonal distribution and diving behaviour of male sperm whales off Kaikōura: foraging implications. Canadian Journal of Zoology 78:407-419. Jaquet N, Gendron D (2002) Distribution and relative abundance of sperm whales in relation to key environmental features, squid landings and the distribution of other cetacean species in the Gulf of California, Mexico. Marine Biology 141:591–601. Jaquet N, Gendron D, Coakes A (2003) Sperm whales in the Gulf of California: residency, movements, behavior, and the possible influence of variation in food supply. Marine Mammal Science 19:545–62. Jaquet N, Whitehead H (1996) Scale-dependent correlation of sperm whale distribution with environmental features and productivity in the South Pacific. Marine Ecology Progress Series 135:1-9. Jaquet N, Whitehead H, Lewis M (1996) Coherence between 19th century sperm whale distributions and satellite-derived pigments in the tropical Pacific. Marine Ecology Progress Series 145:1-10. Jefferson TA, Hung SK, Würsig B (2009) Protecting small cetaceans from coastal development: Impact assessment and mitigation experience in Hong Kong. Marine Policy 33:305-11. Johnson CM, Beckley LE, Kobryn H (2016) Crowdsourcing Modern and Historical Data Identifies Sperm Whale (Physeter macrocephalus) Habitat Offshore of South-Western. Frontiers in Marine Science 3:1–16. References 219
Johnson MW 1948 Sound as a tool in marine ecology, from data on biological noises and the deep scattering layer. Journal of Marine Research 7:443–458. Kaiser MJ, Attrill MJ, Jennings S, Thomas DN, Barnes DK (2005) Marine ecology: processes, systems, and impacts. Oxford University Press. Karczmarski L, Huang SL, Wong WH, Chang WL, S. Chan SCY, Keith M (2016) Distribution of a coastal delphinid under the impact of long-term habitat loss: Indo-Pacific humpback dolphins off Taiwan’s west coast. Estuaries and Coasts. doi:10.1007/s12337-016-0146-5. Karl JW, Heglund PJ, Garton EO, Scott JM, Wright NM, Hutto RL (2000) Sensitivity of species habitat relationship model performance to factors of scale. Ecological Applications 10:1690–1705. Kasuya T (1999) Examination of the reliability of catch statistics in the Japanese coastal sperm whale fishery. Journal of Cetacean Research and Management 1:109-122. Kawakami T (1980) A review of sperm whale food. Scientific Reports of the Whales Research Institute 32:199–218. Kernaléguen L, Arnould JP, Guinet C, Cherel Y (2015) Determinants of individual foraging specialization in large marine vertebrates, the Antarctic and Subantarctic fur seals. Journal of Animal Ecology 84:1081-1091. King JR, McFarlane G (2003) Marine fish life history strategies: applications to fishery management. Fisheries Management and Ecology 10:249-64. King R, Morgan B, Gimenez O, Brooks S (2009) Bayesian analysis for population ecology. Chapman and Hall/CRC. Koenigs C, Miller RJ, Page HM (2015) Top predators rely on carbon derived from giant kelp Macrocystis pyrifera. Marine Ecology Progress Series 537, 1e8. Kooyman GL, Castellini MA, Davis RW (1981) Physiology of diving in marine mammals. Annual Review of Physiology 43:343-56. Krahn MM, Herman DP, Matkin CO, Durban JW, Barrett-Lennard L, Burrows DG, Dahlheim ME, Black N, LeDuc RG, Wade PR (2007) Use of chemical tracers in assessing the diet and foraging regions of eastern North Pacific killer whales. Marine Environmental Research 63: 91-114. Kramer DL (1988) The behavioral ecology of air breathing by aquatic animals. Canadian Journal of Zoology 66:89–94. Krause-Jensen D, Duarte CM (2016) Substantial role of macroalgae in marine carbon sequestration. Nature Geoscience 9:737-742. Krumhansl K, Scheibling R (2012) Production and fate of kelp detritus. Marine Ecology Progress Series 467:281-302. Kutner M, Nachtsheim C, Neter J, Li W (2005) Building the regression model III: Remedial measures. In Applied Linear Statistical Models. New York: McGraw-Hill/Irwin Series, pp 421–480. Landwehr JM, Pregibon D, Shoemaker AC (1984) Graphical methods for assessing logistic regression models. Journal of the American Statistical Association 79:61–71. Larivière F (2001) A descriptive study of physical and biological processes influencing sperm whale distribution. MSc dissertation, University of Otago, New Zealand. Lavery TJ, Roudnew B, Gill P, Seymour J, Seuront L, Johnson G, Mitchell JG, Smetacek V (2010) Iron defecation by sperm whales stimulates carbon export in the Southern Ocean. Proceedings of the Royal Society of London B: Biological Sciences 277:3527-31. Lawrence JM, Armstrong E, Gordon J, Lusseau SM, and Fernandes PG (2016) Passive and active, predator and prey: using acoustics to study interactions between cetaceans and forage fish. ICES Journal of Marine Science 73:2075–2084. References 220
Le Bohec C, Durant JM, Gauthier-Clerc M, Stenseth NC, Park YH., Pradel R, Le Maho Y (2008) King penguin population threatened by Southern Ocean warming. Proceedings of the National Academy of Sciences 105:2493-2497. Lea MA, Guinet C, Cherel Y, Duhamel G, Dubroca L, Pruvost P, Hindell M (2006) Impacts of climatic anomalies on provisioning strategies of a Southern Ocean predator. Marine Ecology Progress Series 310:77–94. Leaman BM, Beamish RJ (1984) Ecological and management implications of longevity in some northeast Pacific groundfishes. International North Pacific Fisheries Commission Bulletin 42:85- 97. Leaper R, Cooke J, Trathan P, Reid K, Rowntree V, Payne R (2006) Global climate drives southern right whale (Eubalaena australis) population dynamics. Biology Letters 2:289-292. Leathwick JR, Elith J, Francis MP, Hastie T, Taylor P (2006) Variation in demersal fish species richness in the oceans surrounding New Zealand: an analysis using boosted regression trees. Marine Ecology Progress Series 321:267-281. Leduc D, Kingston A, Gibbs M, Swales A, Graham B, Nodder S, Rowden A (2018) Quantifying the transfer of terrestrial organic matter into the Kaikōura and Hokitika submarine canyons using bulk and compound-specific stable isotopes. Presentation at the New Zealand Marine Sciences Society conference. Napier, July 2018. Leduc D, Rowden AA, Bowden DA, Probert PK, Pilditch CA, Nodder SD (2012) Unimodal relationship between biomass and species richness of deep-sea nematodes: Implications for the link between productivity and diversity. Marine Ecology Progress Series 454:53–64. Leduc D, Rowden AA, Nodder SD, Berkenbusch K, Probert PK, Hadfield MG (2014) Unusually high food availability in Kaikoura Canyon linked to distinct deep-sea nematode community. Deep Sea Research Part II: Topical Studies in Oceanography 104:310-318. Lefcheck JS (2015) PiecewiseSEM: Piecewise structural equation modelling in R for ecology, evolution, and systematics. Methods in Ecology and Evolution 7:573–579. Leo FC De, Smith CR, Rowden AA, Bowden DA, Clark MR (2010) Submarine canyons: hotspots of benthic biomass and productivity in the deep sea. Proceedings of the Royal Society of London B: Biological Sciences 277:2783–2792. Leung ES, Chilvers BL, Nakagawa S, Moore AB, Robertson BC (2012) Sexual Segregation in Juvenile New Zealand Sea Lion Foraging Ranges: Implications for Intraspecific Competition, Population Dynamics and Conservation. PLoS One 7(9):e45389. Levin LA, Dayton PK (2009) Ecological theory and continental margins: where shallow meets deep. Trends in Ecology and Evolution 24: 606-617. Levin LA, LeBris N (2015) The deep ocean under climate change. Science 350:766–768. Levin LA, Michener R (2002) Isotopic evidence for chemosynthesis-based nutrition of macrobenthos: the lightness of being at Pacific methane seeps. Limnology and Oceanography 47:1336–1345. Levinton J (2001) Marine biology: Function, Biodiversity, Ecology. 2nd ed., Oxford University Press. Lin X, Zhang D (1999) Inference in generalized additive mixed models by using smoothing splines. Journal of the royal statistical society: Series b (statistical methodology) 61:381-400. Lindeman RL (1942) The trophic-dynamic aspect of ecology. Ecology 23: 399-418. Link J (2002) Does food web theory work for marine ecosystems? Marine Ecology Progress Series 230:1- 9. Lobo JM, Jiménez-Valverde A, Real R (2008) AUC: a misleading measure of the performance of predictive distribution models. Global Ecology and Biogeography 17:145–151. References 221
Lockyer C (1981) Estimates of growth and energy budget for the sperm whale, Physeter catodon. Food and Agricultural Organisation of the United Nations Fisheries Series 5:489-504. Lorrain A, Graham BS, Popp BN, Allain V, Olson RJ, Hunt BP, Potier M, Fry B, Galván-Magaña F, Menkes CE, Kaehler S (2015) Nitrogen isotopic baselines and implications for estimating foraging habitat and trophic position of yellowfin tuna in the Indian and Pacific Oceans. Deep Sea Research Part II: Topical Studies in Oceanography 113:188-98. Lorrain A, Savoye N, Chauvaud L, Paulet YM, Naulet N (2003) Decarbonation and preservation method for the analysis of organic C and N contents and stable isotope ratios of low- carbonated suspended particulate material. Analytica Chimica Acta 491:125-133. Lotze HK, Coll M, Magera AM, Ward-Paige C, Airoldi L (2011) Recovery of marine animal populations and ecosystems. Trends in Ecology and Evolution 26:595-605. Lotze HK, Lenihan HS, Bourque BJ, Bradbury RH, Cooke RG, Kay MC, Kidwell SM, Kirby MX, Peterson CH, Jackson JBC, Bay M (2006) Depletion, degradation, and recovery potential of estuaries and coastal seas. Science 312:1806–1809. Lourey MJ, Trull TW, Sigman DM (2003) Sensitivity of δ15N of nitrate, surface suspended and deep sinking particulate nitrogen to seasonal nitrate depletion in the Southern Ocean. Global Biogeochemical Cycles 17(3). Lovenduski NS, Gruber N (2005) Impact of the Southern Annular Mode on Southern Ocean circulation and biology. Geophysical Research Letters 32(11). Lusseau SM and Wing SR (2006) Importance of local production versus pelagic subsidies in the diet of an isolated population of bottlenose dolphins Tursiops sp. Marine Ecology Progress Series 321:283-293. MacAvoy SE, Carney RS, Fisher CR, Macko SA (2002) Use of chemosynthetic biomass by large, mobile, benthic predators in the Gulf of Mexico. Marine Ecology Progress Series 225:65–78. MacLeod CD (2009) Global climate change, range changes and potential implications for the conservation of marine cetaceans: a review and synthesis. Endangered Species Research 7:125- 136. MacLeod CD, Zuur AF (2005) Habitat utilization by Blainville’s beaked whales off Great Abaco, northern Bahamas, in relation to sea bed topography. Marine Biology 147:1–11. Macquart-Moulin C, Patriti G (1996) Accumulation of migratory micronekton crustaceans over the upper slope and submarine canyons of the northwestern Mediterranean. Deep-Sea Research Part I: Oceanography. 43:579–601. Magozzi S, Yool A, Vander Zanden HB, Wunder MB, Trueman CN (2017) Using ocean models to predict spatial and temporal variation in marine carbon isotopes. Ecosphere 8(5):e01763. Manel S, Williams HC, Ormerod SJ (2001) Evaluating presence–absence models in ecology: the need to account for prevalence. Journal of Applied Ecology 38:921-931. Marcoux M, Whitehead H, Rendell L (2007) Sperm whale feeding variation by location, year, social group and clan: evidence from stable isotopes. Marine Ecology Progress Series 333:309-314. Markaida U (2006) Population structure and reproductive biology of jumbo squid Dosidicus gigas from the Gulf of California after the 1997–1998 El Niño event. Fisheries Research 79: 28–37. Markel RW, Shurin JB (2015) Indirect effects of sea otters on rockfish (Sebastes spp.) in giant kelp forests. Ecology 96:2877-2890. Markowitz TM, Richter C, Gordon J (2011) Effects of tourism on the behaviour of sperm whales inhabiting the Kaikoura Canyon. Report to New Zealand Department of Conservation. References 222
Marshall GJ (2003) Trends in the Southern Annular Mode from observations and reanalyses. Journal of Climate 16:4134-4143. doi.org/10.1175/1520-0442(2003). Martin AR, Clarke MR (1986) The diet of sperm whales (Physeter macrocephalus) captured between Iceland and Greenland. Journal of the Marine Biological Association of the United Kingdom 66:779- 790. Mate BR, Irvine LM, Palacios DM (2016) The development of an intermediate‐duration tag to characterize the diving behavior of large whales. Ecology and evolution 7:585-95. Mate BR, Stafford KM, Ljungblad DK (1994) A change in sperm whale (Physeter macrocephalus) distribution correlated to seismic surveys in the Gulf of Mexico. Journal of the Acoustical Society of America 96:3268-3269. Matich P, Heithaus MR (2014) Multi‐tissue stable isotope analysis and acoustic telemetry reveal seasonal variability in the trophic interactions of juvenile bull sharks in a coastal estuary. Journal of Animal Ecology 83:199-213. Matich P, Heithaus MR, Layman CA (2011) Contrasting patterns of individual specialization and trophic coupling in two marine apex predators. Journal of Animal Ecology 80:294-305. Mauchline J, Gordon JDM (1991) Oceanic pelagic prey of benthopelagic fish in the benthic boundary layer of a marginal oceanic region. Marine Ecology Progress Series 74:109-115. McClain CR, Barry JP (2010) Habitat heterogeneity, disturbance, and productivity work in concert to regulate biodiversity in deep submarine canyons. Ecology 91: 964-976. McClelland JW, Montoya JP (2002) Trophic relationships and the nitrogen isotopic composition of amino acids in plankton. Ecology 83:2173–2180. McCullagh P, Nelder J (1989) Generalized Linear Models. Second edition. Chapman and Hall, London, UK McCulloch CE, Searle SH, Neuhaus JM (2008) Generalized, linear and mixed models, 2nd ed. Wiley, New York. McCutchan JH, Lewis WM, Kendall C, McGrath CC (2003) Variation in trophic shift for stable isotope ratios of carbon, nitrogen and sulfur. Oikos 102:378−390. McHugh CM, Ryan WBF, Hecker B (1992) Contemporary sedimentary processes in the Monterey Canyon-fan system. Marine Geology 107:35–50. McKinnon JF (2007) Aspects of the population biology of the southern arrow squid, Nototodarus sloanii, in southern New Zealand (Thesis, Doctor of Philosophy). University of Otago. Retrieved from http://hdl.handle.net/10523/544. McLeod RJ, Wing SR (2007) Hagfish in the New Zealand fjords are supported by chemoautotrophy of forest carbon. Ecology 88:809−816. McLeod RJ, Wing SR (2009) Strong pathways for incorporation of terrestrially derived organic matter into benthic communities. Estuarine, Coastal and Shelf Science 82:645-53. McMahon JJ, Summers WC (1971) Temperature effects on the developmental rate of squid (Loligo pealei) embryos. Biology Bulletin 141:561–567. Melzer E, Schmidt HL (1987) Carbon isotope effects on the pyruvate dehydrogenase reaction and their importance for relative C-13 depletion in lipids. Journal of Biological Chemistry 262:8159-64. Mendes S, Newton J, Reid RJ, Frantzis A, Pierce GJ (2007b). Stable isotope profiles in sperm whale teeth: variations between areas and sexes. Journal of the Marine Biological Association of the United Kingdom 87:621-627. References 223
Mendes S, Newton J, Reid RJ, Zuur AF, Pierce GJ (2007a) Stable carbon and nitrogen isotope ratio profiling of sperm whale teeth reveals ontogenetic movements and trophic ecology. Oecologia 151:605-615. Meyer MA, El-Sayed SZ (1983) Grazing of Euphausia superba Dana on natural phytoplankton populations. Polar Biology 1:193-197. Michener R, Lajtha K (2007) Stable Isotopes in Ecology and Environmental Science. Blackwell Publishing Ltd, Oxford, p. 565. Middelburg JJ, Nieuwenhuize J (1998) Carbon and nitrogen stable isotopes in suspended matter and sediments from the Schelde Estuary. Marine Chemistry 60:217-225. Miller BS, Dawson S, Vennell R (2013a) Underwater behavior of sperm whales off Kaikoura, New Zealand, as revealed by a three-dimensional hydrophone array. Journal of the Acoustical Society of America 134:2690–2700. Miller BS, Growcott A, Slooten E, Dawson SM (2013b) Acoustically derived growth rates of sperm whales (Physeter macrocephalus) in Kaikoura, New Zealand. Journal of the Acoustical Society of America 134:2438–45. Miller BS (2010) Acoustic growth rates and three-dimensional localisation of sperm whales (Physeter macrocephalus) in Kaikoura. Ph.D. thesis, University of Otago, New Zealand. Miller PI, Scales KL, Ingram SN, Southall EJ, Sims DW (2015) Basking sharks and oceanographic fronts: quantifying associations in the north‐east Atlantic. Functional Ecology 29:1099-109. Miller PJO, Johnson MP, Tyack PL (2004) Sperm whale behaviour indicates the use of echolocation click buzzes ‘creaks' in prey capture. Proceedings of the Royal Society of London B: Biological Sciences 271:2239–2247. Mills JA, Yarrall JW, Bradford-Grieve JM, Uddstrom MJ, Renwick JA, Merilä J (2008) The impact of climate fluctuation on food availability and reproductive performance of the planktivorous red-billed gull Larus novaehollandiae scopulinus. Journal of Animal Ecology 77:1129–1142. Mitchell JS, Mackay KA, Neil HL, Mackay EJ, Pallentin A, Notman P (2012) Undersea New Zealand, 1:5,000,000. NIWA Chart, Miscellaneous Series No. 92. Møhl B, Wahlberg M, Madsen PT, Miller LA, Surlykke A (2000) Sperm whale clicks: directionality and source level revisited. Journal of the Acoustical Society of America 107:638-648. Møhl B, Wahlberg M, Madsen PT, Heerfordt A, Lund A (2003) The monopulsed nature of sperm whale clicks, Journal of the Acoustical Society of America 114:1143–1154. Montgomery DC, Peck EA (1992) Introduction to linear regression analysis. New York: Wiley, 504 p. Moore JE, Barlow JP (2013) Declining abundance of beaked whales (Family Ziphiidae) in the California current large marine ecosystem. PLoS One 8(1):e52770. Moore BD, Lawler IR, Wallis IR, Beale CM, Foley WJ (2010) Palatability mapping: a koala's eye view of spatial variation in habitat quality. Ecology 91:3165-3176. Moore MJ, Zerbini AN (2017) Dolphin blubber/axial muscle shear: implications for rigid trans-dermal intra-muscular tracking tag trauma in whales. Journal of Experimental Biology jeb-165282. Moors-Murphy HB (2014) Submarine canyons as important habitat for cetaceans, with special reference to the Gully: A review. Deep Sea Research II: Topical Studies in Oceanography 104:6-19. Mountjoy JJ, Howarth JD, Orpin AR, Barnes PM, Bowden DA, Rowden AA, Schimel AC, Holden C, Horgan HJ, Nodder SD, Patton JR (2018) Earthquakes drive large-scale submarine canyon development and sediment supply to deep-ocean basins. Science advances, 4: eaar3748. References 224
MPI data release OTA2015-001. Summary of fisheries catch data from statistical areas 018 and 020 (offshore of Kaikōura), from January 1990 to January 2018. Supplied by New Zealand’s Ministry for Primary Industries (MPI) under Confidentiality Deed Number OTA2015-001. Murphy RJ, Pinkerton MH, Richardson KM, Bradford‐Grieve JM, Boyd PW (2001) Phytoplankton distributions around New Zealand derived from SeaWiFS remotely‐sensed ocean colour data. New Zealand Journal of marine and freshwater research 35:343-362. Mussi B, Miragliuolo A, Zucchini A, Pace DS (2014) Occurrence and spatio-temporal distribution of sperm whale (Physeter macrocephalus) in the submarine canyon of Cuma (Tyrrhenian Sea, Italy). Aquatic Conservation: Marine and Freshwater Ecosystems 24L:59-70. Myers RA, Worm B (2003) Rapid worldwide depletion of predatory fish communities. Nature 423:280– 283. Nakagawa S, Freckleton RP (2011) Model averaging, missing data and multiple imputation: a case study for behavioural ecology. Behavioral Ecology and Sociobiology 65:103−116. NASA Goddard Space Flight Center, Ocean Ecology Laboratory, Ocean Biology Processing Group (2018) Moderate-resolution Imaging Spectroradiometer (MODIS) Aqua erdMH1chla8day Data, NASA OB.DAAC, Greenbelt, MD, USA. doi: 10.5067/AQUA/MODIS/L3M/CHL/2018. Accessed on 17/5/2018. Nesis KN (1993) Cephalopods of seamounts and submarine ridges. In: Okutami T, O’Dor RK, Kubodera T (eds) Recent advances in fisheries biology. Tokyo University Press, Tokyo, pp 365–373. Neter J, Wasserman W, Kutner MH (1990) Applied linear statistical models. Regression, analysis of variance, and experimental design. Irwin, Homewood, USA Neumann DR (2001) Seasonal movements of short-beaked common dolphins (Delphinus delphis) in the north-western Bay of Plenty, New Zealand: Influence of sea surface temperature and El Niño/La Niña. New Zealand Journal of Marine and Freshwater Research 35:371–374. Newsome SD, Bentall GB, Tinker MT, Oftedal OT and others (2010b) Variation in δ¹³C and δ¹⁵N diet- vibrissae trophic discrimination factors in a wild population of California sea otters. Ecological Applications 20:1744−1752. Newsome SD, Clementz MT and Koch PL (2010a) Using stable isotope biogeochemistry to study marine mammal ecology. Marine Mammal Science 26:509-572. Newsome SD, Etnier MA, Monson DH, Fogel ML (2009b) Retrospective characterization of ontogenetic shifts in killer whale diets via δ13C and δ15N analysis of teeth. Marine Ecology Progress Series 374:229–242. Newsome SD, Tinker MT, Monson DH, Oftedal OT, Ralls K, Staedler MM, Fogel ML, Estes JA (2009a) Using stable isotopes to investigate individual diet specialization in California sea otters (Enhydra lutris nereis). Ecology 90:961–974. Nguyen HM, Rountrey AN, Meeuwig JJ, Coulson PG, Feng M, Newman SJ, Waite AM, Wakefield CB, Meekan MG (2015). Growth of a deep-water, predatory fish is influenced by the productivity of a boundary current system. Scientific Reports 5:9044. Nielsen JB, Nielsen F, Joergensen PJ, Grandjean P (2000) Toxic metals and selenium in blood from pilot whales (Globecephala melas) and sperm whales (Physeter catodon). Marine Pollution Bulletin 40:348-351. Nifong JC, Layman CA, Silliman, BR (2015) Size, sex and individual‐level behaviour drive intrapopulation variation in cross‐ecosystem foraging of a top‐predator. Journal of Animal Ecology 84:35-48. References 225
Nishiwaki N, Ohsumi S, Maeda Y (1963) Change of form in the sperm whale accompanied with growth. Scientific Reports. Whale Research Institute Tokyo 17:1–13. NIWA (2018) Summary of seasonal climate and weather over New Zealand since 2009. Available at https://www.niwa.co.nz/climate/seasonal-climate-outlook. Nodder SD, Pilditch CA, Probert PK, Hall JA (2003) Variability in benthic biomass and activity beneath the Subtropical Front, Chatham Rise, SW Pacific Ocean. Deep Sea Research Part I Oceanographic Research Papers 50:959–985. Nordstrom CA, Battaile BC, Cotté C, Trites AW (2013) Foraging habitats of lactating northern fur seals are structured by thermocline depths and submesoscale fronts in the eastern Bering Sea. Deep Sea Research Part II Topical Studies in Oceanography 88–89:78–96. Norris K, Harvey G (1972) A theory for the function of the spermaceti organ of the sperm whale (Physeter catodon) in Animal Orientation and Navigation. (Science and Technology Office, National Aeronautics and Space Administration, Washington, DC), pp. 397–417. Notarbartolo di Sciara G (2014) Sperm whales, Physeter macrocephalus, in the Mediterranean Sea: a summary of status, threats, and conservation recommendations. Aquatic Conservation 24:4–10. O’Toole MD, Lea MA, Guinet C, Schick R, Hindell M (2015) Foraging strategy switch of a top marine predator according to seasonal resource differences. Frontiers in Marine Science 1:10.3389. Ohkouchi N, Chikaraishi Y, Close HG, Fry B, Larsen T, Madigan DJ, McCarthy MD, McMahon KW, Nagata T, Naito YI, Ogawa NO (2017). Advances in the application of amino acid nitrogen isotopic analysis in ecological and biogeochemical studies. Organic Geochemistry 113: 150–174. Okutani T, Nemoto T (1964) Squids as the food of sperm whales in the Bering Sea and Alaskan Gulf. Scientific Reports. Whale Research Institute Tokyo 18:111–122. Orelien JG, Edwards LJ 2008. Fixed-effect variable selection in linear mixed models using R2 statistics. Computational Statistics and Data Analysis 52:1896–1907. Ostrom PH, Lien J, Macko SA (1993) Evaluation of the diet of Sowerby's beaked whale, Mesoplodon bidens, based on isotopic comparisons among northwestern Atlantic cetaceans. Canadian Journal of Zoology 71:858-61. Palanques A, Durrieu de Madron X, Puig P, Fabres J, Guillén J, Calafat A, Canals M, Heussner S, Bonnin J (2006) Suspended sediment fluxes and transport processes in the Gulf of Lions submarine canyons. The role of storms and dense water cascading. Marine Geology 234:43–61. Pandolfi JM, Bradbury RH, Sala E, Hughes TP, Bjorndal KA, Cooke RG, McArdle D, McClenachan L, Newman MJ, Paredes G, Warner RR, Jackson JB (2003) Global trajectories of the long-term decline of coral reef ecosystems. Science 301:955–958. Papastavrou V, Smith SC, Whitehead H (1989) Diving behavior of the sperm whale, Physeter macrocephalus, off the Galapagos Islands. Canadian Journal of Zoology 67: 839–846. Papiol V, Cartes JE, Fanelli F, Maynou (2012) Influence of environmental variables on the spatio- temporal dynamics of bentho-pelagic assemblages in the middle slope of the Balearic Basin (NW Mediterranean). Deep-Sea Research Part I 64:84-99. Papiol V, Cartes JE, Fanelli E, Rumolo P (2013) Food web structure and seasonality of slope megafauna in the NW Mediterranean elucidated by stable isotopes: relationship with available food sources. Journal of Sea Research 77:53-69. Pauly D, Christensen V, Dalsgaard J, Froese R, Torres F. (1998a) Fishing down marine food webs. Science 279: 860–863. Pauly D, Trites AW, Capuli E, Christensen V (1998b) Diet composition and trophic levels of marine mammals. ICES Journal of Marine Science 55:467-81. References 226
Pecl GT, Jackson GD (2008) The potential impacts of climate change on inshore squid: biology, ecology and fisheries. Reviews in Fish Biology and Fisheries 18:373-385. Pelletier L, Kato A, Chiaradia A, Ropert-Coudert Y (2012) Can thermoclines be a cue to prey distribution for marine top predators? A case study with little penguins. PLoS One 7(4):e31768. Peñaloza CL, Kendall WL, Langtimm C (2014) Reducing bias in survival under non-random temporary emigration. Ecological Applications 24: 1155–1166. Perga ME, Grey J (2010) Laboratory measures of isotope discrimination factors: comments on Caut, Angulo and Courchamp (2008, 2009). Journal of Applied Ecology 47:942-947. Perry AL, Low PJ, Ellis JR, Reynolds JD (2005) Climate change and distribution shifts in marine fishes. Science 308:1912-5. Peterson BJ, Fry B (1987) Stable isotopes in ecosystem studies. Annual Review of Ecology, Evolution and Systematics 18:293−320. Peterson BJ, Howarth R (1987) Sulfur, carbon, and nitrogen isotopes used to trace organic matter flow in the salt-marsh estuaries of Sapelo Island, Georgia. Limnology and Oceanography 32:1195– 1213. Phillips DL (2012) Converting isotope values to diet composition: the use of mixing models. Journal of Mammalogy 93(2): 342–352. doi:10.1644/11-MAMM-S-158.1 Phillips DL, Gregg JW (2001) Uncertainty in source partitioning using stable isotopes. Oecologia 127:171–179. Phillips KL, Jackson GD, Nichols PD (2001) Predation on myctophids by the squid Moroteuthis ingens around Macquarie and Heard Islands: stomach contents and fatty acid analyses. Marine Ecology Progress Series 215, 179-189. Pierce GJ, Boyle PR (2003) Empirical modelling of interannual trends in abundance of squid (Loligo forbesi) in Scottish waters. Fisheries Research 59:305-326. Pierce G, Boyle P (1991) A review of methods for diet analysis in piscivorous marine mammals. Oceanography and Marine Biology Annual Review 29:409–486. Pinnegar JK, Polunin NVC (1999) Differential fractionation of δ¹³C and δ¹⁵N among fish tissues: implications for the study of trophic interactions. Functional Ecology 13:225−231. Pinnegar JK, Polunin NVC (2000) Contributions of stable isotope data to elucidating food webs of Mediterranean rocky littoral fishes. Oecologia 122:399−409. Pinzone M, Budzinski H, Tasciotti A, Ody D, Lepoint G, Schnitzler J, Scholl G, Thomé J, Tapie N, Eppe G, Das K (2015) POPs in free-ranging pilot whales, sperm whales and fin whales from the Mediterranean Sea: Influence of biological and ecological factors. 142:185–196 Pirotta E, Matthiopoulos J, MacKenzie M, Scott-Hayward L, Rendell L (2011) Modelling sperm whale habitat preference: A novel approach combining transect and follow data. Marine Ecology Progress Series 436:257–272. Pirotta E, Thompson PM, Miller PI, Brookes KL, Cheney B, Barton TR, Graham IM, Lusseau D (2014) Scale‐dependent foraging ecology of a marine top predator modelled using passive acoustic data. Functional ecology 28:206-17. Pitman RL, Ballance LT, Mesnick SI, Chivers SJ (2001) Killer Whale Predation on Sperm Whales: Observations and Implications. Marine Mammal Science 17:494-507. Pollock, K. (1982). A capture-recapture design robust to unequal probability of capture. The Journal of Wildlife Management 46:752–757. References 227
Polunin NVC, Morales-Nin B, Pawsey WE, Cartes JE, Pinnegar JK, Moranta J (2001) Feeding relationships in Mediterranean bathyal assemblages elucidated by stable nitrogen and carbon isotope data. Marine Ecology Progress Series 220:13-23. Pomeroy LR, leB Williams PJ, Azam F, Hobbie JE (2007) The microbial loop. Oceanography 20:28-33. Popp BN, Graham BS, Olson RJ, Hannides CC, Lott MJ, López‐Ibarra GA, Fry B (2007) Insight into the trophic ecology of yellowfin tuna, Thunnus albacares, from compound‐specific nitrogen isotope analysis of proteinaceous amino acids. Terrestrial Ecology 1:173-190. Post D (2002) Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83:703−718. Post DM, Layman CA, Arrington DA, Takimoto G, Quattrochi J, Montana CG (2007) Getting to the fat of the matter: models, methods and assumptions for dealing with lipids in stable isotope analysis. Oecologia 152:179-189. Praca E, Gannier A, Das K, Laran S (2009) Modelling the habitat suitability of cetaceans: Example of the sperm whale in the northwestern Mediterranean Sea. Deep-Sea Research I 56:648–657. Puskaric S, Fowler SW, Miquel JC (1992) Temporal changes in particulate flux in the northern Adriatic Sea. Estuarine Coastal Shelf Science 35:267–287. Quetglás A, Carbonell A, Sánchez P (2000) Demersal continental shelf and upper slope cephalopod assemblages from the Balearic sea (North-Western Mediterranean). Biological aspects of some deep-sea species. Estuarine, Coastal and Shelf Science 50:739–749. Quillfeldt P, McGill RAR, Furness RW (2005) Diet and foraging areas of Southern Ocean seabirds and their prey inferred from stable isotopes: review and case study of Wilson’s storm-petrel. Marine Ecology Progress Series 295:295–304. R Development Core Team (2012) R: a language and environment for statistical computing. Version 2.15.0. R Foundation for Statistical Computing, Vienna, Austria. Available at: http://www.r- project.org/. Ramirez-Llodra E, Brandt A, Danovaro R, De Mol B, Escobar E, German C, Levin L, Arbizu P, Menot L, Buhl-Mortensen P, Narayanaswamy BE (2010) Deep, diverse and definitely different: unique attributes of the world's largest ecosystem. Biogeoscience 7:2851-2899. Ramos R, González-Solís J (2012) Trace me if you can: the use of intrinsic biogeochemical markers in marine top predators. Frontiers in Ecology and the Environment 10:258-266. Ramsay TO, Burnett RT, Krewski D (2003) The effect of concurvity in generalized additive models linking mortality to ambient particulate matter. Epidemiology 14:18-23. Rau GH, Ainley DG, Bengtson JL, Torres JJ, Hopkins TL (1992a) 15N/14N and 13C/12C in Weddell Sea birds, seals and fish: implications for diet and trophic structure. Marine Ecology Progress Series 84:1–8. Rau GH, Takahashi T, Desmarais DJ, Repeta DJ, Martin JH (1992b) The relationship between delta 13C of organic matter and [CO2(Aq)] in ocean surface water data from a JGOFs site in the northeast Atlantic Ocean and a model. Geochim Cosmochim Acta 56:1413–1419. Rau GH, Sweeney RH, Kaplan IR (1982) Plankton 13C:12C ratio changes with latitude: differences between northern and Southern Ocean. Deep-Sea Research 29:1035–1039. Rayment WJ, Dawson S, Webster T (2014) Breeding status affects fine-scale habitat selection of southern right whales on their wintering grounds. Journal of Biogeography 42:463-474. Rayment W, Webster T, Brough T, Jowett T, Dawson S (2018) Seen or heard? A comparison of visual and acoustic autonomous monitoring methods for investigating temporal variation in occurrence of southern right whales. Marine Biology 165:12. References 228
Read AJ, Drinker P, Northridge S (2006) Bycatch of marine mammals in U.S. and global fisheries. Conservation Biology 20:163–169. Redfern J V, Ferguson MC, Becker EA, Hyrenbach KD, Good C, Barlow J, Kaschner K, Baumgartner MF, Forney KA, Ballance LT, Fauchald P, Halpin P, Hamazaki T, Pershing AJ, Qian SS, Read A, Reilly SB, Torres L, Werner F (2006) Techniques for cetacean - habitat modeling. Marine Ecology Progress Series 310:271–295. Reeves RR, McClellan K, Werner TB (2013) Marine mammal bycatch in gillnet and other entangling net fisheries, 1990 to 2011. Endangered Species Research 20:71-97. Rendell L, Frantzis A (2016) Mediterranean sperm whales, Physeter macrocephalus: the precarious state of a lost tribe. Advances in Marine Biology 75:37-74. Rendell L, Whitehead H, Escribano R (2004) Sperm whale habitat use and foraging success off northern Chile: evidence of ecological links between coastal and pelagic systems. Marine Ecology Progress Series 275:289–295. Renison D, Boersma D, Martella MB (2002) Winning and losing: causes for variability in outcome of fights in male Magellanic penguins. Behavioural Ecology 13:462-466. Renwick J, Thompson D (2006) The southern annular mode and New Zealand climate. Water and Atmosphere 4:24-25. Reynolds RW, Smith TM, Liu C, Chelton DB, Casey KS, Schlax MG (2007) Daily high-resolution- blended analyses for sea surface temperature. Journal of Climate 20:5473-96. Rhinelander MQ, Dawson SM (2004) Measuring sperm whales from their clicks: stability of interpulse intervals and validation that they indicate whale length. The Journal of Acoustical Society of America 115:1826–1831. Rhodes JR, McAlpine CA, Zuur AF, Smith GM, Ieno EN (2009) GLMM applied on the spatial distribution of koalas in a fragmented landscape. In: Mixed effects models and extensions in ecology with R 2009 (pp. 469-492). Springer, New York, NY. Rice DW (1989) Sperm whale, Physeter macrocephalus Linnaeus, 1758. In: Ridgway SH, Harrison SR (eds) Handbook of marine mammals, Vol 4. Academic Press, London, p 177–233. Richards SA, Whittingham MJ, Stephens PA (2011) Model selection and model averaging in behavioural ecology: the utility of the IT-AIC framework. Behavioral Ecology and Sociobiology 65:77-89. Richter BE, Jones BA, Ezzell JL, Porter NL, Avdalovic N, Pohl C (1996) Accelerated solvent extraction: a technique for sample preparation. Analytical Chemistry 68:1033-39. Richter C, Dawson S, Slooten E (2006) Impacts of commercial whale watching on male sperm whales at Kaikoura, New Zealand. Marine Mammal Science 22:46–63. Ritz DA, Hosie GW, Kirkwood RJ (1990) Diet of Nyctiphanes australis Sars (Crustacea: Euphausiacea). Australian Journal of Marine and Freshwater Research 41:365–374. Roberts CM, Hawkins JP (1999) Extinction risk in the sea. Trends in Ecology and Evolution 14:241-246. Robinson J, Cavanaugh C (1995) Expression of form I and II Rubisco in chemoautotrophic symbioses: Implications for the interpretation of stable carbon isotopic values. Limnology and Oceanography 40:1496−1502. Rodhouse PG (2001) Managing and forecasting squid fisheries in variable environments. Fisheries Research 54:3-8. Rodhouse PG (2008) Large-scale range expansion and variability in ommastrephid squid populations: a review of environmental links. CalCOFI Rep., 49:83-9. References 229
Rodhouse PG (2013) Role of squid in the Southern Ocean pelagic ecosystem and the possible consequences of climate change. Deep Sea Research Part II: Topical Studies in Oceanography 95:129-138. Rodhouse PG, White MG, Jones MRR (1992). Trophic relations of the cephalopod Martialia hyadesi (Teuthoidea: Ommastrephidae) at the Antarctic Polar Front, Scotia Sea. Marine Biology 114:415-421. Roemmich D, Sutton PJH (1998) The mean and variability of ocean circulation past northern New Zealand: determining the representativeness of hydrographic climatologies. Journal of Geophysical Research 103: 13041–13054. Rolland V, Barbraud C, Weimerskirch H (2009) Assessing the impact of fisheries, climate and disease on the dynamics of the Indian yellow-nosed albatross. Biological Conservation 142:1084–1095. Roman J, Estes JA, Morissette L, Smith C, Costa D, McCarthy J, Nation JB, Nicol S, Pershing A, Smetacek V (2014) Whales as marine ecosystem engineers. Frontiers in Ecology and the Environment 12:377–385. Roman J, McCarthy JJ (2010) The whale pump: marine mammals enhance primary productivity in a coastal basin. PLoS One 5 e13255. Rooney N, McCann K, Gellner G, Moore JC (2006) Structural asymmetry and the stability of diverse food webs. Nature 442: 265−269. Ropelewski CF, Jones PD (1987) An extension of the Tahiti-Darwin Southern Oscillation Index. Monthly Weather Review 115:2161-2165. Röpke A, Nellent W, Piatkowski U (1993) A comparative study on the influence of the pycnocline on the vertical distribution of fish larvae and cephalopod paralarvae in three ecoligically different areas of the Arabian Sea. Deep Sea Research Part II 40:801-19. Rosenblatt A, Heithaus M, Mather M, Matich P, Nifong J, Ripple W, Silliman B (2013) The Roles of Large Top Predators in Coastal Ecosystems: New Insights from Long Term Ecological Research. Oceanography 26:156–167 Rosenzweig C, Karoly D, Vicarelli M, Neofotis P, Wu Q, Casassa G, Menzel A, Root TL, Estrella N, Seguin B, Tryjanowski P (2008) Attributing physical and biological impacts to anthropogenic climate change. Nature. 453:353-7. Ross PS, Dungan SZ, Hung SK, Jefferson TA, Macfarquhar C, Perrin WF, Riehl KN, Slooten E, Tsai J, Wang JY, White BN (2010) Averting the baiji syndrome: conserving habitat for critically endangered dolphins in Eastern Taiwan Strait. Aquatic Conservation: Marine and Freshwater Ecosystems. 20:685-94. Rossman S, Ostrom PH, Stolen M, Barros NB, Gandhi H, Stricker CA, Wells RS (2015) Individual specialization in the foraging habits of female bottlenose dolphins living in a trophically diverse and habitat rich estuary. Oecologia 178:415-25. Ruiz-Cooley RI, Engelhaupt DT, Ortega-Ortiz JG (2012) Contrasting C and N isotope ratios from sperm whale skin and squid between the Gulf of Mexico and Gulf of California: effect of habitat. Marine biology 159:151-164. Ruiz-Cooley RI, Gendron D, Aguíñiga S, Mesnick S, Carriquiry JD (2004) Trophic relationships between sperm whales and jumbo squid using stable isotopes of C and N. Marine Ecology Progress Series 277:275-283. Ruiz-Cooley RI, Koch PL, Fiedler PC, McCarthy MD (2014) Carbon and Nitrogen Isotopes from Top Predator Amino Acids Reveal Rapidly Shifting Ocean Biochemistry in the Outer California Current. PLoS ONE 9(10): e110355. References 230
Ryan JP, Chavez FP, Bellingham JG (2005) Physical-biological coupling in Monterey Bay, California: topographic influences on phytoplankton ecology. Marine Ecology Progress Series 287:23-32. Sabarros PS, Ménard F, Lévénez JJ, Tew-Kai E, Ternon JF (2009) Mesoscale eddies influence distribution and aggregation patterns of micronekton in the Mozambique Channel. Marine Ecology Progress Series 395, 101-107. Sagnol O, Richter C, Field LH, Reitsma F (2014b) Spatio-temporal distribution of sperm whales (Physeter macrocephalus) off Kaikoura, New Zealand, in relation to bathymetric features. New Zealand Journal of Zoology 41:234–247 Sagnol O, Richter C, Reitsma F, Field L (2014a) Estimating sperm whale (Physeter macrocephalus) daily abundance from a shore-based survey within the Kaikoura submarine canyon, New Zealand. New Zealand Journal of Marine and Freshwater Research 49:41–50. Saijo D, Mitani Y, Abe T, Sasaki H, Goetsch C, Costa DP, Miyashita K (2016) Linking mesopelagic prey abundance and distribution to the foraging behavior of a deep-diving predator, the northern elephant seal. Deep-Sea Research II: Topical Studies in Oceanography 140:163-170. Sallée JB, Speer KG, Rintoul SR (2010) Zonally asymmetric response of the Southern Ocean mixed- layer depth to the Southern Annular Mode. Nature Geoscience 3:273-279. Samarra FI, Vighi M, Aguilar A, Víkingsson GA (2017) Intra-population variation in isotopic niche in herring-eating killer whales off Iceland. Marine Ecology Progress Series 564:199-210. Santos MB, Pierce GJ, Boyle PR, Reid RJ, Ross HM, Patterson IAP, Hernandez-Garcia V (1999) Stomach contents of sperm whales Physeter macrocephalus stranded in the North Sea 1990–1996. Marine Ecology Progress Series 183:281-294. Sathyendranath S, Grant M, and 63 others (2018) ESA Ocean Colour Climate Change Initiative (Ocean_Colour_cci): Version 3.1 Data. Centre for Environmental Data Analysis, 04 July 2018. Scales KL, Miller PI, Embling CB, Ingram SN, Pirotta E, Votier SC (2014) Mesoscale fronts as foraging habitats: composite front mapping reveals oceanographic drivers of habitat use for a pelagic seabird. Journal of the Royal Society Interface 11:20140679. Schenker N, Gentleman JF (2001) On judging the significance of differences by examining the overlap between confidence intervals. The American Statistician 55:182-186. Schick RS, Kraus SD, Rolland RM, Knowlton AR, Hamilton PK, Pettis HM, Kenney RD, Clark JS (2013) Using hierarchical Bayes to understand movement, health, and survival in the endangered North Atlantic right whale. PloS one 8(6):e64166. Schiel DR, Lilley SA, South PM, Coggins JH (2016) Decadal changes in sea surface temperature, wave forces and intertidal structure in New Zealand. Marine Ecology Progress Series 548:77-95. Schlitzer R (2001) Ocean Data View (ODV), http://www.awi-bremerhaven.de/GEO/ODV Schmidt K, Atkinson A, Stübing D, McClelland JW, Montoya JP, Voss M (2003) Trophic relationships among Southern Ocean copepods and krill: some uses and limitations of a stable isotope approach. Limnology and Oceanography 48:277-289. Schreiber SJ, Bürger R, Bolnick DI (2011) The community effects of phenotypic and genetic variation within a predator population. Ecology 92:1582-1593. Scott BE, Sharples J, Ross ON, Wang J, Pierce GJ, Camphuysen CJ (2010) Sub-surface hotspots in shallow seas: fine-scale limited locations of top predator foraging habitat indicated by tidal mixing and sub-surface chlorophyll. Marine Ecology Progress Series 408:207-226. Scott BE, Sharples J, Wanless S, Ross ON, Frederiksen M, Daunt F (2006) The use of biologically meaningful oceanographic indices to separate the effects of climate and fisheries on seabird References 231
breeding success. In: Boyd IL, Wanless S, Camphuysen CJ (eds) Top predators in marine ecosystems: their role in monitoring and management. Cambridge University Press. Seber GAF (1982) The Estimation of Animal Abundance and Related Parameters, 2nd Edition. Charles and Griffin and Company Limited: UK. 654 p. Segar DA (2007) Introduction to ocean sciences. 2nd ed. Norton & Company, NY, USA. Seyboth E, Groch KR, Dalla Rosa L, Reid K, Flores PA, Secchi ER (2016) Southern right whale (Eubalaena australis) reproductive success is influenced by krill (Euphausia superba) density and climate. Scientific Reports 6: 28205. Shapiro SS, Wilk MB (1965) An analysis of variance test for normality (complete samples). Biometrika 52:591–611. Shaw AG and Vennell R (2000) Variability of water masses through the Mernoo saddle, South Island, New Zealand. New Zealand Journal of Marine and Freshwater Research 34:103-116. Shears NT, Babcock RC (2007) Quantitative description of mainland New Zealand’s shallow subtidal reef communities. Science for Conservation 280, New Zealand Department of Conservation. Shears NT, Bowen MM (2017) Half a century of coastal temperature records reveal complex warming trends in western boundary currents. Scientific Reports 7:14527. Shepard FP, Dill RF (1966) Submarine canyons and other sea valleys. Chicago: Rand McNally & Co. Shillinger GL, Swithenbank AM, Bailey H, Bograd SJ, Castelton MR, Wallace BP, Block BA (2011) Vertical and horizontal habitat preferences of post-nesting leatherback turtles in the South Pacific Ocean. Marine Ecology Progress Series 422:275-289. Simenstad CA, Estes JA, Kenyon KW (1978) Aleuts, sea otters, and alternate stable-state communities. Science 200:403–411. Similä T, Holst JC, Christensen I (1996) Occurrence and diet of killer whales in northern Norway: seasonal patterns relative to the distribution and abundance of Norwegian spring-spawning herring. Canadian Journal of Fisheries and Aquatic Sciences 53:769-779. Simmonds MP, Eliott WJ (2009) Climate change and cetaceans: concerns and recent developments. Journal of the Marine biological Association of the United Kingdom 89:203-10. Sims DW, Genner MJ, Southward AJ, Hawkins SJ (2001) Timing of squid migration reflects North Atlantic climate variability. Proceedings of the Royal Society of London B: Biological Sciences 268:2607-2611. Sing T, Sander O, Beerenwinkel N, Lengauer T (2015) Visualising the performance of scoring classifiers using ‘ROCR’. https://cran.r-project.org/web/packages/ROCR/index.html Skov H, Gunnlaugsson T, Budgell WP, Horne J, Nøttestad L, Olsen E (2008) Small-scale spatial variability of sperm and sei whales in relation to oceanographic and topographic features along the Mid-Atlantic Ridge. Deep-Sea Research II 55:254–268. Smith C, Whitehead H (1993) Variation in the feeding success and behaviour of Galapagos sperm whales (Physeter macrocephalus) as they relate to oceanographic conditions. Canadian Journal of Zoology 71:1991-1996. Smith CR, Hoover DJ, Doan SE, Pope RH, Demaster FC, Altabet MA (1996) Phytodetritus at the abyssal seafloor across 10° of latitude in the central equatorial Pacific. Deep-Sea Research 43:1309–1338. Smith LV, McMinn A, Martin A, Nicol S, Bowie AR, Lannuzel D, van der Merwe P (2013) Preliminary investigation into the stimulation of phytoplankton photophysiology and growth by whale faeces. Journal of experimental marine biology and ecology 446:1-9. References 232
Smith SC, Whitehead H (2000) The diet of Galapagos Sperm Whales, Physeter Macrocephalus, as indicated by fecal sample analysis. Marine Mammal Science 16:315-325. Smith JM, Parker GA (1976) The logic of asymmetric contests. Animal behaviour 24:159-175. Soldevilla MS, Wiggins SM, Hildebrand JA, Oleson EM, Ferguson MC (2011) Risso’s and Pacific white-sided dolphin habitat modelling from passive acoustic monitoring. Marine Ecology Progress Series 423:247-260. Somerford T (2018) Population dynamics and social structure of sperm whales in Kaikōura, New Zealand. MSc dissertation, University of Otago, New Zealand. Somes CJ, Schmittner A, Galbraith ED, Lehmann MF and others (2010) Simulating the global distribution of nitrogen isotopes in the ocean. Global Biogeochemical Cycles 24: GB4019. Soto KH, Trites AW, Arias-Schreiber M (2006) Changes in diet and maternal attendance of South American sea lions indicate changes in the marine environment and prey abundance. Marine Ecology Progress Series 312:277–290 Sprogis KR, Christiansen F, Wandres M, Bejder L (2017) El Niño Southern Oscillation influences the abundance and movements of a marine top predator in coastal waters. Global Change Biology 24:1085-1096. Starbuck A (1878) History of the American whale fishery from its earliest inception to the year 1876. Smithsonian libraries 128209. Steer MA, Moltschaniwskyj NA, Jordan AR (2003b) Embryonic development of southern calamary (Sepioteuthis australis) within the constraints of an aggregated egg mass. Marine and Freshwater Research 54:217–226. Stephens TA, Hepburn CD (2014) Mass-transfer gradients across kelp beds influence Macrocystis pyrifera growth over small spatial scales. Marine Ecology Progress Series 515:97-109. Stuart V (1986) Feeding and metabolism of Euphausia lucens (Euphausiacea) in the southern Benguela current. Marine Ecology Progress Series 30:117-125. Stuart V, Pillar SC (1990) Diel grazing patterns of all ontogenetic stages of Euphausia lucens and in situ predation rates on copepods in the southern Benguela upwelling region. Marine Ecology Progress Series 64:227-241. Suárez Jiménez RS, Hepburn CD, Hyndes GA, McLeod RJ, Taylor RB, Hurd CL (2015) Do native subtidal grazers eat the invasive kelp Undaria pinnatifida? Marine biology 162:2521-2526. Suárez-Seoane S, Osborne PE, Alonso JC (2002) Large-scale habitat selection by agricultural steppe bird in Spain: identifying species–habitat responses using generalized additive models. Journal of Applied Ecology 39:755–771. Sumich JL (1999) An introduction to the biology of marine life. WCB McGraw-Hill, New York, NY Sutton P (2001) Detailed structure of the Subtropical Front over Chatham Rise, east of New Zealand. Journal of Geophysical Research: Oceans, 106:31045- 31056. Sutton P (2003) The Southland Current: A Subantarctic current. New Zealand Journal of Marine and Freshwater Research 37:645-652. Sutton TT, Porteiro FM, Heino M, Byrkjedal I, Langhelle G, Anderson CI, Horne J, Søiland H, Falkenhaug T, Godø OR, Bergstad OA (2008) Vertical structure, biomass and topographic association of deep-pelagic fishes in relation to a mid-ocean ridge system. Deep-Sea Research II: Topical Studies in Oceanography. 55:161-84. Sweeting CJ, Polunin NV, Jennings S (2006) Effects of chemical lipid extraction and arithmetic lipid correction on stable isotope ratios of fish tissues. Rapid Communications in Mass Spectrometry 20:595-601. References 233
Symonds RE, Moussalli A (2011) A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion. Behavioral Ecology and Sociobiology 65:13−21. Takai N, Onaka S, Ikeda Y, Yatsu A, Kidokoro H, Sakamoto W (2000) Geographical variations in carbon and nitrogen stable isotope ratios in squid. Journal of the Marine Biological Association of the UK 80:675–684. Tallis H (2009) Kelp and rivers subsidize rocky intertidal communities in the Pacific Northwest (USA). Marine Ecology Progress Series 389: 85-96. Tasker ML, Camphuysen CJ, Cooper J, Garthe S, Montevecchi WA, Blaber SJ (2000) The impacts of fishing on marine birds. ICES journal of Marine Science 57:531-547. Taylor BL, Baird R, Barlow J, Dawson SM, Ford J, Mead JG, Notarbartolo di Sciara G, Wade P, Pitman RL (2008) Physeter macrocephalus. The IUCN Red List of Threatened Species. Teloni V, Mark JP, Patrick MJO, Peter MT (2008) Shallow food for deep divers: Dynamic foraging behavior of male sperm whales in a high latitude habitat. Journal of Experimental Marine Biology and Ecology 354:119–131. Teloni V, Zimmer W, Wahlberg M, Madsen P (2007) Consistent acoustic size estimation of sperm whales using clicks recorded from unknown aspects. Journal of Cetacean Research and Management 9:127–136. Tepsich P, Rosso M, Halpin PN, Moulins A (2014) Habitat preferences of two deep-diving cetacean species in the northern Ligurian Sea. Marine Ecology Progress Series 508:247-260. Thomas SM, Crowther TW (2015) Predicting rates of isotopic turnover across the animal kingdom: A synthesis of existing data. Journal of Animal Ecology 84: 861-870. Thomson JA, Heithaus MR, Burkholder DA, Vaudo JJ, Wirsing AJ, Dill LM (2012) Site specialists, diet generalists? Isotopic variation, site fidelity, and foraging by loggerhead turtles in Shark Bay, Western Australia. Marine Ecology Progress Series 453:213-226. Todd S, Ostrom P, Lien J, Abrajano J (1997) Use of biopsy samples of humpback whale (Megaptera novaeangliae) skin for stable isotope (δ13C) determination. Journal of Northwest Atlantic Fishery Science 22:71-76. Tønnessen JN, Johnsen AO (1982) The History of Modern Whaling. University of California Press, Berkeley and Los Angeles, CA, USA. Topping DT, Szedlmayer ST (2011) Site fidelity, residence time and movements of red snapper Lutjanus campechanus estimated with long-term acoustic monitoring. Marine Ecology Progress Series 437:183-200. Torres L, Smith T, Sutton P, MacDiarmid A, Bannister J (2011) Habitat use and distribution patterns of southern right whales and sperm whales discerned from spatial analyses of 19th century whaling records. Report prepared for Australian Marine Mammal Centre. NIWA Client Report No. WLG2011-52 Torres LG, Read A J, Halpen P (2008) Fine-scale habitat modelling of top marine predator: Do prey data improve predictive capacity? Ecological Applications 18:1702–1717 Torres LG, Smith TD, Sutton P, MacDiarmid A, Bannister J, Miyashita T (2013) From exploitation to conservation: Habitat models using whaling data predict distribution patterns and threat exposure of an endangered whale. Diversity and Distributions 19:1138–1152. Trathan P N, Brierley AS, Brandon MA, Bone DG, Goss C, Grant SA, Murphy EJ, Watkins JL (2003) Oceanographic variability and changes in Antarctic krill (Euphausia superba) abundance at South Georgia. Fisheries Oceanography 12:569–583. References 234
Trathan PN, Forcada J, Murphy EJ (2007) Environmental forcing and Southern Ocean marine predator populations: effects of climate change and variability. Philosophical Transactions of the Royal Society of London B: Biological Sciences 362:2351-65. Treacy SD, TW Crawford (1981) Retrieval of otoliths and statoliths from gastrointestinal contents and scats of marine mammals. The Journal of Wildlife Management 45:990–993. Trenberth KE (1997) The definition of El Niño. Bulletin of the American Meteorological Society 78:2771– 2777. Trites AW, Christensen V, Pauly D (1997) Competition between fisheries and marine mammals for prey and primary production in the Pacific Ocean. Journal of Northwest Atlantic Fishery Science 22:173-187. Tromp JJ, Jones PL, Symonds JE, Walker SP, Pope A, Pether SM, Afonso LO (2016) Effects of commercial diets and temperature on the growth performance and stress response of hapuku (Polyprion oxygeneios). Aquaculture 452:128-133. Trull TW, Armand L (2001) Insights into Southern Ocean carbon export from the δ13C of particles and dissolved inorganic carbon during the SOIREE iron release experiment. Deep-Sea Research II 48:2655–2680. Turvey ST, Risley CL (2006) Modelling the extinction of Steller's sea cow. Biology Letters 2:94-97. Twist BA (2015) Pecten novaezealandiae at its southern limit: habitat dynamics and ecology within Paterson Inlet, Stewart Island. MSc dissertation, University of Otago, New Zealand. Twist BA, Hepburn CD, Rayment WJ (2016) Distribution of the New Zealand scallop (Pecten novaezealandiae) within and surrounding a customary fisheries area. ICES Journal of Marine Science 73:384-393. Ullman DS, Cornillon PC (1999) Satellite‐derived sea surface temperature fronts on the continental shelf off the northeast US coast. Journal of Geophysical Research: Oceans 104:23459-78. Unger B, Bravo EL, Deaville R, Gröne A, Ijsseldijk LL, Leopold MF, Siebert U, Spitz J, Wohlsein P, Herr H (2016) Large amounts of marine debris found in sperm whales stranded along the North Sea coast in early 2016. Marine pollution bulletin, 112: 134-141. Urian K, Gorgone A, Read A, Balmer B, Wells RS, Berggren P, Durban J, Eguchi T, Rayment W, Hammond PS (2015) Recommendations for photo-identification methods used in capture- recapture models with cetaceans. Marine Mammal Science 31:298–321. van der Linde M (2009) Long-term monitoring of sperm whales in Kaikoura: data-management, abundance, and population dynamics. MSc dissertation, University of Otago, New Zealand. Van Hale R (2003) Stable Isotope Biogeochemistry of the Otago Continental Shelf. PhD dissertation, University of Otago, New Zealand. Van Hale R, Frew RD (2010) Rayleigh distillation equations applied to isotopic evolution of organic nitrogen across a continental shelf. Marine and Freshwater Research 61:369-378. Van Oevelen D, Soetaert K, García R, De Stigter HC, Cunha MR, Pusceddu A and Danovaro R (2011) Canyon conditions impact carbon flows in food webs of three sections of the Nazaré canyon. Deep Sea Research II: Topical Studies in Oceanography 58:2461-2476. Vander Zanden MJ, Clayton MK, Moody EK, Solomon CT, Weidel BC (2015) Stable isotope turnover and half-life in animal tissues: a literature synthesis. PLoS One 10(1), e0116182. Vander Zanden MJ, Rasmussen JB (2001) Variation in δ15N and δ13C trophic fractionation: implications for aquatic food web studies. Limnology and Oceanography 46:2061-2066. Vanderklift MA, Ponsard S (2003) Sources of variation in consumer-diet δ15N enrichment: a meta- analysis. Oecologia 136:169-182. References 235
Vaughan IP, Ormerod SJ (2005) The continuing challenges of testing species distribution models. Journal of Applied Ecology 42:720–730. Vetter EM, Dayton PK (1998) Macrofaunal communities within and adjacent to a detritus-rich submarine canyon system. Deep-Sea Research I 45:25−54. Vetter EM, Dayton PK (1999) Organic enrichment by macrophyte detritus, and abundance patterns of megafaunal populations in submarine canyons. Marine Ecology Progress Series 186:137−143. Vetter EM, Smith CR, Leo FC De (2010) Hawaiian hotspots: enhanced megafaunal abundance and diversity in submarine canyons on the oceanic islands of Hawaii. Marine Ecology 31:183–199. Vetter EW (1994) Hotspots of benthic production. Nature 372 47. Vidal EA, DiMarco FP, Wormorth JH, Lee PG (2002) Influence of temperature and food availability on survival, growth and yolk utilization in hatchling squid. Bulletin of Marine Sciences 71:915- 931. Vidal EA, Haimovici M, Hackbart VC (2010) Distribution of paralarvae and small juvenile cephalopods in relation to primary production in an upwelling area off southern Brazil. ICES Journal of Marine Science 67:1346-52. Villanueva R (2000) Effect of temperature on statolith growth of the European squid Loligo vulgaris during early life. Marine Biology 136:449–460. Villouta E, Chadderton WL, Pugsley CH, Hay C (2001) Effects of sea urchin (Evechinus chloroticus) grazing in Dusky Sound, Fiordland, New Zealand. New Zealand Journal of Marine and Freshwater Research 35(5):1007-1024. Vincent WF, Howard‐Williams C, Tildesley P, Butler E (1991) Distribution and biological properties of oceanic water masses around the South Island, New Zealand. New Zealand Journal of Marine and Freshwater Research 25:21-42. Vindenes Y, Engen S, Saether BE (2008) Individual heterogeneity in vital parameters and demographic stochasticity. The American Naturalist 171:455–467. Vitousek PM, Mooney HA, Lubchenco J, Melillo JM (1997) Human Domination of Earth’s Ecosystems. Science 277:494–499. von Biela VR, Newsome SD, Bodkin JL, Kruse GH, Zimmerman CE (2016) Widespread kelp-derived carbon in pelagic and benthic nearshore fishes suggested by stable isotope analysis. Estuarine, Coastal and Shelf Science 181: 364-74. Wada E, Hattori A (1991) Nitrogen in the sea: forms, abundances and rate process. CRC Press, FL. Wada E, Terazaki M, Kabaya Y, Nemoto T (1987) 15N and 13C abundances in the Antarctic Ocean with emphasis on the biogeochemical structure of the food web. Deep-Sea Research Part I– Oceanographic Research Papers 34:829–841. Wakeham SG, Lee C (1993) Production, transport, and alteration of particulate organic matter in the marine water column. In: Organic geochemistry (pp. 145-169). Springer US. Wakeham SG (1995) Lipid biomarkers for heterotrophic alteration of suspended particulate organic matter in oxygenated and anoxic water columns of the ocean. Deep-Sea Research I 42:1749–177. Waluda CM, Rodhouse PG (2006) Remotely sensed mesoscale oceanography of the Central Eastern Pacific and recruitment variability in Dosidicus gigas. Marine Ecology Progress Series 310:25-32. Waluda CM, Trathan PN, Rodhouse PG (1999) Influence of oceanographic variability on recruitment in the Illex argentinus (Cephalopoda: Ommastrephidae) fishery in the South Atlantic. Marine Ecology Progress Series 183:159-167. References 236
Wang C, Deser C, Yu J-Y, DiNezio P, Clement A (2017) El Niño and Southern Oscillation (ENSO): A review. In: Glynn PW, Manzello DP & Enochs IC (eds.), Coral reefs of the eastern tropical pacific: Persistence and loss in a dynamic environment (pp. 85–106). Dordrecht: Springer Waring GT, Hamazaki T, Sheehan D, Wood G, Baker S (2001) Characterization of beaked whale (Ziphiidae) and sperm whale (Physeter macrocephalus) summer habitat in shelf-edge and deeper waters off the Northeast U.S. Marine Mammal Science 17:703-717. Watkins WA, Daher MA, Fristrup KM, Howald TJ, Sciara GN Di (1993) Sperm whales tagged with transponders and tracked underwater by sonar. Marine Mammal Science 9:55–67. Watwood SL, Miller PJO, Johnson M, Madsen PT, Tyack PL (2006) Deep-diving foraging behaviour of sperm whales (Physeter macrocephalus). Journal of Animal Ecology 75:814–825. Weilgart LS, Whitehead H (1988) Distinctive vocalizations from mature male sperm whales (Physeter macrocephalus). Canadian Journal of Zoology 66:1931-1937. Weilgart L, Whitehead H (1993). Coda communication by sperm whales (Physeter macrocephalus) off the Galapagos Islands. Canadian Journal of Zoology 71:744-752. Werner EE, Mittelbach GG (1981) Optimal foraging: field tests of diet choice and habitat switching. American Zoologist 21:813–829. White EP, Ernest SKM, Kerkhoff AJ, Enquist BJ (2007) Relationships between body size and abundance in ecology. Trends in Ecology and Evolution 22:323-330. White GC, Burnham KP (1999) Program MARK: survival estimation from populations of marked animals. Bird study 46(sup1): 120–139. Whitehead H (1996) Variation in the feeding success of sperm whales: Temporal scale, spatial scale and relationship to migrations. Journal of Animal Ecology 65:429-438. Whitehead H (2002) Estimates of the current global population size and historical trajectory for sperm whales. Marine Ecology Progress Series 242:295–304. Whitehead H (2003) Sperm Whales: Social Evolution in the Ocean. University of Chicago Press. Whitehead H (2017) The sperm whale. In: Encyclopedia of Marine Mammals. 3rd edition. Eds Würsig B, Thewissen JGM, Kovacs KM. Whitehead H, Brennan S, Grover D (1992) Distribution and behaviour of male sperm whales on the Scotian Shelf, Canada. Canadian Journal of Zoology 70:912-918. Whitehead H, Christal J, Dufault S (1997) Past and distant whaling and the rapid decline of sperm whales off the Galapagos Islands. Conservation Biology 11:1387-1396. Whitehead H, Coakes A, Jaquet N, Lusseau S (2008) Movements of sperm whales in the tropical Pacific. Marine Ecology Progress Series 361:291-300. Whitehead H, Gordon J, Mathews EA, Richard, KR (1990) Obtaining skin samples from living sperm whales. Marine Mammal Science 6:316-326. Whitehead H, Papastavrou V, Smith SC (1989) Feeding success of sperm whales and sea-surface temperature off the Galapagos Islands. Marine Ecology Progress Series 201-203. Whitehead H, Rendell L (2004) Movements, habitat use and feeding success of cultural clans of South Pacific sperm whales. Journal of Animal Ecology, 73: 190-196. Würsig B, Gailey GA (2002) Marine mammals and aquaculture: conflicts and potential resolutions. Responsible Marine Aquaculture, pp 45-59. CAP International Press, New York. Williams PM, Gordon LI (1970) Carbon 13: carbon 12 ratios in dissolved and particulate organic matter in the sea. Deep-Sea Research 17:19-27. References 237
Williams RJ, Berlow EL, Dunne JA, Barabási AL, Martinez ND (2002) Two degrees of separation in complex food webs. Proceedings of the National Academy of Science in the USA, 99:12913– 12916. Wing SR, Beer NA, Jack L (2012) Resource base of blue cod Parapercis colias subpopulations in marginal fjordic habitats is linked to chemoautotrophic production. Marine Ecology Progress Series 466:205-214. Wing SR, McLeod RJ, Clark KL, Frew RD (2008) Plasticity in the diet of two echinoderm species across an ecotone: microbial recycling of forest litter and bottom-up forcing of population structure. Marine Ecology Progress Series 360:115–123. Wong SNP, Whitehead H (2014) Seasonal occurrence of sperm whales (Physeter macrocephalus) around Kelvin Seamount in the Sargasso Sea in relation to oceanographic processes. Deep-Sea Research Part I: Oceanographic Research Papers 91:10–16. Wood S, Scheipl F (2017) Generalized Additive Mixed Models using 'mgcv' and 'lme4'. Available at https://cran.r-project.org/web/packages/gamm4/index.html Wood SN (2003) Thin plate regression splines. Journal of the Royal Statististical Society B 65:95-114. Wood SN (2006) Generalized additive models: an introduction with R. Chapman & Hall/CRC, Boca Raton, FL Wood SN (2011) Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society 73:3–36. Wood SN (2016) Mixed GAM Computation Vehicle with GCV/AIC/REML Smoothness Estimation. Available at: https://cran.r-project.org/web/packages/mgcv/index.html Woodson CB, Litvin SY (2015) Ocean fronts drive marine fishery production and biogeochemical cycling. Proceedings of the National Academy of Sciences 112:1710-5. Woodward G, Ebenman B, Emmerson M, Montoya JM, Olesen JM, Valido A, Warren PH (2005) Body size in ecological networks. Trends in Ecology and Evolution 20:402-409. Worm B, Barbier EB, Beaumont N, Duffy JE, Folke C, Halpern BS, Jackson JBC, Lotze HK, Micheli F, Palumbi SR, Sala E, Selkoe KA, Stachowicz JJ, Watson R (2006) Impacts of biodiversity loss on ocean ecosystem services. Science 314:787–90. Worm B, Sandow M, Oschlies A, Lotze HK, Myers RA (2005) Global patterns of predator diversity in the open oceans. Science 309:1365-1369. Worm B, Tittensor DP (2011) Range contraction in large pelagic predators. Proceedings of the National Academy of Science USA 108:11942-11947. Xing X, Claustre H, Blain S, D’Ortenzio F, Antoine D, Ras J, Guinet C (2012) Quenching correction for in vivo chlorophyll fluorescence acquired by autonomous platforms: A case study with instrumented elephant seals in the Kerguelen region (Southern Ocean). Limnology and Oceanography: Methods 10:483–495. Yee TW, Mitchell ND (1991) Generalized additive models in plant ecology. Journal of Vegetation Science 2:587–602. Yen PPW, Sydeman WJ, Hyrenbach KD (2004) Marine bird and cetacean associations with bathymetric habitats and shallow-water topographies: implications for trophic transfer and conservation. Journal of Marine Systems 50:79–99. Yohannes E, Arnaud A, Béchet A (2014) Tracking variations in wetland use by breeding flamingos using stable isotope signatures of feather and blood. Estuarine, Coastal and Shelf Science 136:11- 18. References 238
Zaniewski AR, Lehmann A, Overton JM (2002) Predicting species spatial distributions using presence- only data: a case study of native New Zealand ferns. Ecological Modelling 157:261-280. Zupcic-Moore JR, Ruiz-Cooley RI, Paliza O, Koch PL, McCarthy MD (2017) Using stable isotopes to investigate foraging variation and habitat use of sperm whales from northern Peru. Marine Ecology Progress Series 579:201-212. Zuur AF, Ieno E, Walker NJ, Saveliev A, Smith GM (2009) Mixed effects models and extensions in ecology with R. Gail M, Krickeberg K, Samet JM, Tsiatis A, Wong W, editors. New York, NY: Spring Science and Business Media.