Content uploaded by Thomas Mattern
Author content
All content in this area was uploaded by Thomas Mattern on Apr 06, 2016
Content may be subject to copyright.
MARINE ECOLOGY OF OFFSHORE AND INSHORE FORAGING
PENGUINS: THE SNARES PENGUIN EUDYPTES ROBUSTUS AND
YELLOW-EYED PENGUIN MEGADYPTES ANTIPODES
THOMAS MATTERN
A thesis submitted for the degree of
Doctor of Philosophy
at the University of Otago, Dunedin, New Zealand
June 2006
What’s that on the television then? - Looks like a penguin. - No, no, no, I didn’t
mean what’s on the television set, I meant what programme. - Oh. (pause) It’s
been a long time there, now, has it? - It’s funny that penguin being there innit?
What’s it doing there? - Standing. - I can see that! - If it lays an egg, it will fall
down the back of the television set. - We’ll have to watch that. Unless it’s a male.
- Ooh, I never thought of that. - Yes, looks fairly butch. - Per’aps it’s from
next door. - Penguins don’t come from next door, they come from the Antarctic.
- Burma. - Why did you say Burma? - I panicked. - Oh. (pause) Perhaps it’s
from the zoo. - Which zoo? - How should I know which zoo? I’m not Doctor
bloody Bernowski. - How does Doctor Bernowski know which zoo it came
from? - He knows everything. - Oooh, I wouldn’t like that, that’d take all the
mystery out of life. (pause) Anyway, if it came from the zoo, it would have
‘property of the zoo’ stamped on it. - No it wouldn’t. ey don’t stamp animals
‘property of the zoo’. You can’t stamp a huge lion. - ey stamp them when
they’re small. - What happens when they moult? - Lions don’t moult. - No, but
penguins do. ere, I’ve run rings around you logically. - Oh, intercourse the
penguin! - (TV announcer) It’s just gone 8 o’clock and time for the penguin
on top of your television set to explode. (the penguin explodes) - How did
he know that was going to happen?! - It was an inspired guess. And now...
Monty Python’s Flying Circus - 22 - 24 November 1970
Abstract i
ABSTRACT
Seabirds have become adapted for foraging in an oceanic environment that can be highly
dynamic. Oceanographic processes determine the spatial distribution of seabird prey, while
seasonality often has a temporal infl uence on prey availability. In penguins, these factors are
refl ected in the different species’ foraging strategies. Penguins can broadly be categorized
as inshore foragers that live in subtropical to temperate regions and profi t from a stable food
supply throughout the year close to their breeding sites, and offshore foragers that breed in a
pelagic environment at higher latitudes where oceanographic processes and seasonality create
much more dynamic, temporally limited prey situations. In this light, offshore foragers can
be expected to be much more fl exible in their foraging behaviour so as to quickly respond to
changes in a dynamic marine environment, while inshore foragers are more likely to exhibit
predictable foraging patterns. I examined the foraging ecology of two New Zealand penguin
species – the offshore foraging Snares penguin Eudyptes robustus and the inshore foraging
Yellow-eyed penguin Megadyptes antipodes and how their foraging strategies refl ect an
adaptation to the marine environment they exploit.
Diet composition of breeding Snares penguins (incubation and early chick-guard) was
determined using the water-offl oading method. Before the chicks hatched, the penguins
generally brought little food back from their long foraging trips. During chick-guard, the
stomach contents comprised mainly of crustaceans (~55%), fi sh (~24%) and cephalopods
(~21%). However, the presence at times of many fi sh otoliths and squid beaks suggests that
the latter two prey classes may play an even more important role in the adults’ diet than the
simple percentages based on mass suggest. The penguins’ nesting routines were strongly
synchronised between the years and correlated with the onset of the spring planktonic bloom.
Using GPS data loggers and dive recorders I found that during the incubation phase, male
penguins that performed long (ca. 2 week) foraging trips exhibited a strong affi nity to forage
in the Subtropical Front some 200 km east of the Snares. At that stage (late mid-October)
the front featured elevated chlorophyll a concentrations, a pattern that can be observed every
year. Thus, it seems that the front represents a reliable and predictable source of food for the
Abstract ii
male penguins. After the males returned, the female penguins also performed long foraging
trips (<1 week) but never reached the front, primarily because they had to time their return to
the hatching of their chicks. After the chicks had hatched, the female Snares penguins were
the sole providers of food. At this stage, the penguins performed short foraging trips (1-3
days) and foraged halfway between the Snares and Stewart Island (ca. 70-90 km north of the
Snares), where nutrient-rich coastal waters fl ow eastwards to form the Southland Current.
The penguins concentrated their diving effort in these waters, underlining the importance
of the warm coastal waters as a food source for breeding Snares penguins. However, diving
behaviour between 2003 and 2004 differed with penguins searching for prey at greater depths
in the latter year. This underlines the Snares penguins’ behavioural fl exibility in response to a
changing marine environment.
The Yellow-eyed penguins as typical inshore foragers showed very consistent foraging
patterns at all stages. GPS logger deployments on penguins at Oamaru revealed that the
birds foraged almost exclusively at the seafl oor and targeted specifi c areas that featured
reefs or epibenthic communities. As a result, the penguins’ at-sea movements appeared
conservative and at times almost stereotypic. Nevertheless, a comparison of Yellow-eyed
penguins breeding on the adjacent Codfi sh and Stewart islands revealed a degree of plasticity
in the species’ foraging behaviour. Birds from Codfi sh Island extended their foraging ranges
considerably and switched from primarily bottom to mid-water foraging during the post-guard
stage of breeding. It seems likely that this switch is a result of enhanced feeding conditions
(e.g. increased prey abundance/quality) in an area further away from the island, but the time
required to get there renders this strategy not viable when chicks are small and need to be
guarded and fed on a daily basis. As such, the change of behaviour represents a traditional
pattern rather than a dynamic response to a sudden change in the marine environment. In
comparison, penguins from Stewart Island showed consistent foraging patterns during all
stages of breeding. Given the high levels of chick starvation on Stewart Island, the lack of
plasticity in foraging behaviour is surprising and might indicate that Yellow-eyed penguins
fi nd it diffi cult to react quickly to a sub-optimal food situation.
Abstract iii
Overall, it seems that Yellow-eyed penguins show a specialisation for a consistent benthic
environment and, thus, lack the behavioural fl exibility apparent in Snares penguins, which
fi nd their food in a changing pelagic marine environment.
Acknowledgments iv
ACKNOWLEDGMENTS
Honour to whom honour is due. Deepest thanks and appraisal to the best supervisor in the
world: Lloyd Davis. I don’t know how you managed to stay calm (and my friend) in all those
years. Not only was I the “most expensive student” you ever supervised, but often my work
was dogged by offi cial quarrels about permits, reconnaissance trips and device purchases.
As my supervisor, you repeatedly found yourself in the line of fi re, and had to fi ght battles
that no-one – least of all you – really needed. I can’t thank you enough for all those times
you covered my back, you’re a legend! Most importantly, however, your scientifi c guidance
always put me back on track when my argumentation got too carried away and – although it
might sound cheesy – it is always an honour to benefi t from your intellect.
All my love to Ursula Ellenberg! Almost everything I did during the last four years happened
side-by-side with you. I won’t even try to list all of your contributions during the genesis of
this thesis – from thought-inspiring discussions to last minute proof reading – I thank you for
all of that. On top of that you had to live with my moods when things went wrong, but always
gave me strength when I needed it most and never failed to remind me that our work is of
relevance beyond sciences’ ivory tower. It is priceless to be able to share the enthusiasm for
nature with a like-minded being. Side by side with his mother, Finn-Erik, my whirlwind of
a son… thank you so much for lightening up our lives (and for not driving your mum crazy
when daddy was away stalking penguins). Same applies to Hannah Lea the most recent
addition to our little family... there is nothing more soothing than 3 month old girl quietly
snoring on your lap while revising a PhD thesis...
Thanks heaps and more to Dave Houston. Although I have said it before, I say it again:
without you I wouldn’t be where I am. You sacrifi ced your spare-time so that we could go to
the Snares (and certainly put yourself professionally between a rock and a hard place in the
course of that). Until the end you sat on rainy beaches to chase Yellow-eyed penguins with
backpacks and to get yourself more than one bite mark. But most importantly… thank you for
being my best friend! And also thanks to Susan Houston for sharing her husband with me.
Further special thanks are due to Alvin Setiawan, for being a perfect student-colleague and
Acknowledgments v
great friend. It was always inspiring to discuss penguins with you. Thank you for helping
out repeatedly during fi eldwork. And, one thing for sure, I will never-ever forget that it was
you who hit the jackpot and caught the last Snares penguin equipped with an incredibly
expensive GPS logger – after 50 hours non-stop surveillance in ice rain on the Snares! You
really deserve your place in my personal hall of fame. And on the same line I want to mention
Katta Ludynia. Thanks for coming with us to the Snares and risking some of your own
devices, for your high spirits even in utterly miserable conditions and your never-ending smile
that is almost as wide as you are… err… tall. Further thanks to Lars-Gunnar Ellenberg for
contributing signifi cantly to the programming of analysis software and coming up with the
amazing otolith-measurement program! Special thanks to Herman Ellenberg for reading and
making valuable comments to improve some of the chapters
Next I want to thank my notoriously non-english speaking parents Horst and Agi Mattern.
Hotte, allerbesten Dank für deinen erneuten Pinguineinsatz im neuseeländischen Regenwald!
Ist dir vielleicht nicht aufgefallen, aber ohne dich hätte ich die Arbeit da unten nicht machen
können. Deshalb – auch wenn du’s nicht lesen kannst – Kapitel 7 ist mit dein Verdienst. Und
Mam… danke, dass du den Trainer ein weiteres Mal hast gehen lassen und Ursel so großartig
den Rücken freigehalten hast. Und Dank an euch beide für phänomenales Finn-Erik-sitten
während Ursel und ich wiedermal auf einer Vogelinsel herum kletterten.
. At the Department of Conservation I thank Kevin Pearce for looking out for penguins,
Graham Loh for diving on penguin tracks, Bruce McKinley for supporting the YEP logger
work, and Dean Nelson for refreshing my banding skills. I thank their colleagues down
South, most notably Brent Beaven, Phred Dobbins and Malcolm Rutherford for their
enthusiastic support of my work on Stewart and Codfi sh Island. I especially have to thank
Pete McClelland for biting the bullet and giving us the permission to work on the Snares
after all. A big thank you to the Yellow-eyed penguin trust (particularly Sue Murray and
David Blair) without whose support most of the Yellow-eyed penguin work would have been
impossible.
And fi nally, the biggest thanks of all to those marvellous Snares and Yellow-eyed penguins
that cooperated beyond expectation and allowed me to get a glimpse of their world.
Acknowledgments vi
Following I list chapter specifi c acknowledgements.
Chapter 2 - Original conference presentation co-authored by Katrin Ludynia & Stafen
Garthe (RTC Westcoast, Christian-Albrechts-University at Kiel, Germany), Dave Houston
(Department of Conservation, New Zealand) and Lloyd Spencer Davis (Department of
Zoology, University of Otago, New Zealand). We thank Ursula Ellenberg, Alvin Setiawan and
Kerry Weston for (at times very challenging) help in the fi eld. Special thanks are due to Gerrit
Peters from earth&OCEAN for developing the GPS loggers and unmatched product support.
Hermann Ellenberg provided valuable comments on early drafts of this paper. This research
was funded by a University of Otago research grant issued to LSD and a University of Otago
postgraduate scholarship granted to TM. KL was supported through Konrad-Adenauer-
Stiftung. The University of Kiel provided further fi nancial support.
Chapter 3 - Peer-review version of this chapter to be co-authored by Dave Houston
(Department of Conservation, New Zealand) and Lloyd Spencer Davis (Department of
Zoology, University of Otago, New Zealand). Thanks to Uwe Piatkowski (IFM-GEOMAR,
University of Kiel, Germany) for his help identifying cephalopod species and to Chris Lalas
(Dunedin, New Zealand) for advice and help with identifi cation of some otoliths. Very special
thanks to Ursula Ellenberg for help and support in the fi eld and beyond, and to Lars-Gunnar
Ellenberg for programming the otolith measuring software.
Chapter 4 - Peer-review version of this chapter to be co-authored by Dave Houston
(Department of Conservation, New Zealand) and Ursula Ellenberg & Lloyd Spencer Davis
(Department of Zoology, University of Otago, New Zealand). Lots of thanks to Alvin Setiawan
and Katrin Ludynia for help in the fi eld. We thank Gerrit Peters, earth&Ocean Technologies,
for his unlimited support during one of the fi rst fi eld applications of the newly developed GPS
loggers. Lars-Gunnar Ellenberg contributed to the development of the analysis software. We
thank Pete McClelland (Department of Conservation, Invercargill) for issuing the permits for
working on the Snares. This research was approved by the University of Otago Animal Ethics
Committee and complies with the current laws of New Zealand. The study was supported by
a University of Otago Postgraduate Scholarship to TM and a Otago Research Grant issued to
Acknowledgments vii
LSD.
Chapter 5 - Peer-review version of this chapter to be co-authored by Katrin Ludynia (RTC
Westcoast, Christian-Albrechts-University at Kiel, Germany), Dave Houston (Department of
Conservation, New Zealand) and Lloyd Spencer Davis (Department of Zoology, University
of Otago, New Zealand). I thank Ursula Ellenberg, Alvin Setiawan and Kerry Weston for
venturous help in the fi eld. Special thanks are due to Gerrit Peters from earth&OCEAN
for developing the GPS loggers and unmatched product support. I thank Pete McClelland
(Department of Conservation, Invercargill) for issuing the permits for working on the
Snares. This research was approved by the University of Otago Animal Ethics Committee
and complies with the current laws of New Zealand. The study was funded by a University
of Otago research grant issued to LSD and a University of Otago Postgraduate Scholarship
granted to TM. KL was supported through Konrad-Adenauer-Stiftung. The University of Kiel
provided further fi nancial support.
Chapter 6 - Peer-review version of this chapter to be co-authored by Dave Houston
(Department of Conservation, New Zealand) and Ursula Ellenberg & Lloyd Spencer Davis
(Department of Zoology, University of Otago, New Zealand). Very Special thanks have to be
expressed to Graeme Loh (Department of Conservation, Dunedin) for surveying the seafl oor
along foraging routes of Yellow-eyed penguins. I thank our helpers during the fi eld work:
Kevin Pearce (Department of Conservation, Oamaru) for monitoring nests and re-covering
elusive logger birds; Jim Caldwell provided access to and information about penguins he
knows for years; and Kevin Houston and Lars-Gunnar Ellenberg were a great help during the
deployment and recovery of devices. Further thanks are due to Gerrit Peters, earth&OCEAN
Technologies, Kiel, for outstanding product support. Herman Ellenberg provided valuable
comments on early drafts of this paper. We are particularly grateful to the Yellow-eyed
Penguin Trust for allowing the use of the Trust’s GPS loggers in the 2004/05 season. This
study was funded through a University of Otago Research Grant issued to LSD and an
University of Otago Postgraduate Scholarship granted to TM.
Acknowledgments viii
Chapter 7 - Peer-review version of this chapter to be co-authored by Ursula Ellenberg &
Lloyd Spencer Davis (Department of Zoology, University of Otago, New Zealand). This
research would not have been possible without the support of the Yellow-eyed penguin Trust,
who not only provided the GPS loggers but also funded most of the fi eld work and granted
additional fi nancial support to TM. We would like to thank numerous people who facilitated
this study and provided invaluable help and support. Horst Mattern, Lars-Gunnar Ellenberg,
Kathrin Englert were keen and able fi eld helpers on Stewart Island. Special thanks to Phred
Dobbins (Department of Conservation Stewart Island) for organising logistics on Stewart
Island and additional help in the bush, and to Malcolm Rutherford for giving us a hand and
being a great host on Codfi sh Island. We also thank Brent Beaven (DOC Stewart Island) for
his open-minded and enthusiastic support of this project. Pete McClelland (DOC Invercargill)
granted the permits and Jeremy Carroll gave us an introduction to Codfi sh Island. Special
thanks to Sandy King and Julie McInnes for fruitful discussion and uncomplicated exchange
of information and data. Additional funding came from a University of Otago Research grant
issued to LSD.
Table of contents ix
TABLE OF CONTENTS
Abstract i
Acknowledgments vi
Table of contents ix
Chapter 1 General Introduction: Foraging behaviour of 1
penguins and the infl uence of the marine environment
1.1 Foraging strategies in penguins 2
1.2 Infl uence of the marine environment 3
1.3 Foraging ecology of New Zealand penguins 4
1.4 Snares vs. Yellow-eyed penguin foraging - Aims of this thesis 5
1.5 Thesis outline 8
Chapter 2 How to get the most (or anything) out of GPS loggers: 10
a case study with Snares penguins
2.1 Introduction 11
2.2 Methods 13
2.2.1 Study site and species 13
2.2.2 GPS loggers and deployments 13
2.2.3 GPS basics and logger setup 15
2.2.4 Data analysis 16
2.3 Results 17
2.3.1 Data outcome and battery life 17
2.3.2 GPS performance and dive behaviours 18
2.4 Discussion 19
2.4.1 Deployment location and disadvantages of programmed start 19
2.4.2 GPS performance at sea 20
2.4.3 How to get the most out of GPS loggers 22
Tables & Figures 23-27
Table of contents x
Chapter 3 The diet of breeding Snares penguins Eudyptes robustus 28
3.1 Introduction 29
3.2 Methods 30
3.2.1 Stomach sampling 30
3.2.2 Sample processing and analysis 31
3.3 Results 32
3.3.1 Food load volume and importance of prey classes 32
3.3.2 Number of species 33
3.3.3 Importance of species in food load 33
3.3.4 Importance of species in accumulated material 34
3.3.5 Size of prey species 35
3.4 Discussion 35
3.4.1 Food composition and importance of prey species 35
3.4.2 Stomach content weights 37
3.4.3 Comparison with other crested penguins in New Zealand 38
Tables & Figures 41-45
Chapter 4 Infl uence of oceanography and seasonality on foraging 46
behaviour and nesting patterns of Snares penguins
Eudyptes robustus during the incubation stage
4.1 Introduction 47
4.2 Methods 49
4.2.1 Timing of fi eld work and study site 49
4.2.2 Nest attendance patterns 49
4.2.3 GPS loggers and Time Depth Recorders (TDR) 50
4.2.4 Oceanographic data 52
4.2.5 Data analysis 52
4.3 Results 54
4.3.1 Nest attendance patterns and foraging trip durations 54
4.3.2 At-sea movements 55
4.3.3 Diving behaviour 56
4.4 Discussion 58
4.4.1 Synchronous departure of the males 58
4.4.2 Foraging of male penguins and infl uence of oceanography 59
4.4.3 Foraging of female penguins 62
Tables & Figures 64-71
Table of contents xi
Chapter 5 Foraging ranges and spatial distribution of dive activity 72
in female Snares penguins Eudyptes robustus during
the chick-guard stage
5.1 Introduction 73
5.2 Methods 74
5.2.1 Data loggers 74
5.2.2 General data analysis 76
5.2.3 Spatial analysis of dive parameters 77
5.3 Results 78
5.3.1 Infl uence of device size on diving behaviour 78
5.3.2 At-sea movements 79
5.3.3 Spatial distribution of diving performance 79
5.3.4 Differences in diving behaviour between the years 80
5.4 Discussion 81
5.4.1 Logger impact on foraging behaviour 81
5.4.2 Foraging ranges of female Snares penguins 81
5.4.3 Comparison with other crested penguins 82
5.4.4 Infl uence of the oceanic environment 83
5.4.5 Different diving behaviour between years 84
Tables & Figures 86-94
Chapter 6 Conservative foraging routes and exclusive bottom 95
feeding behaviour in the Yellow-eyed penguin
Megadyptes antipodes
6.1 Introduction 96
6.2 Methods 98
6.2.1 GPS loggers and dive recorders 98
6.2.2 Logger deployment procedures 99
6.2.3 Data analysis 100
6.3 Results 101
6.3.1 Trip duration and foraging ranges 101
6.3.2 Patterns of at-sea movements 102
6.3.3 Diving behaviour 104
6.4 Discussion 105
6.4.1 Foraging ranges and trip types 105
6.4.2 Consistency of foraging patterns 106
6.4.3 Exclusive bottom feeders 107
6.4.4 Diving behaviour during different trip stages 109
6.4.5 Navigation by bottom features? 110
6.5 Conclusions 111
Tables & Figures 112-117
Table of contents xii
Chapter 7 Plasticity in foraging behaviour of Yellow-eyed penguins 118
Megadyptes antipodes from neighbouring breeding
locations, Stewart and Codfi sh Island, New Zealand
7.1 Introduction 119
7.2 Methods 121
7.2.1 Study sites 121
7.2.2 Data loggers 121
7.2.3 Data analysis 122
7.3 Results 123
7.3.1 Foraging patterns 124
7.3.2 Foraging parameters - Stewart Island vs. Codfi sh Island 124
7.3.3 Foraging parameters - chick-guard vs. post-guard 125
7.3.4 Aberrant foraging behaviour 126
7.3.5 Adult weights 127
7.4 Discussion 127
7.4.1 Relationship of foraging ranges and benthic habitat 127
7.4.2 Temporal or traditional plasticity of foraging behaviour? 129
7.4.3 The Stewart Island paradox: short ranges, heavy adults and 131
chick starvation
Tables & Figures 133-134
Chapter 8 General discussion 135
8.1 Summary 136
8.2 Conclusions 138
8.3 Future directions 139
References 141-154
Appendix Conference abstracts 155
Foraging behaviour of Snares crested penguins - a matter 156
of role allocations during breeding
5th Intl. Penguin Conference, Ushuaia, Tierra del Fuego, Argentina, Sep. 2004
Thugs and bullies - patterns of agression in Snares crested 157
penguins
5th Intl. Penguin Conference, Ushuaia, Tierra del Fuego, Argentina, Sep. 2004
How to get the most (or anything) out of GPS loggers: a case 158
study with Snares penguins
2nd Intl. Bio-logging Science Symp., University of St. Andrews, Scotland, June 2005
Warm-water penguins - why are Snares penguins (Eudyptes 159
robustus) doing better than other crested penguin species?
Oamaru Penguin Symp., Oamaru, New Zealand, June/July 2005
Fish and ships? Indications for substantial fi sheries inter- 160
actions of Yellow-eyed penguins Megadyptes antipodes
Oamaru Penguin Symp., Oamaru, New Zealand, June/July 2005
1. Penguin foraging behaviour and the marine environment 1
CHAPTER 1
GENERAL INTRODUCTION:
FORAGING BEHAVIOUR OF PENGUINS AND THE
INFLUENCE OF THE MARINE ENVIRONMENT
1. Penguin foraging behaviour and the marine environment 2
1.1 Foraging strategies in penguins
Penguins (Spheniscidae) are widely distributed in the cooler waters of the southern ocean
and are present on almost every Antarctic and Subantarctic island (Stonehouse 1975). They
constitute the majority of the avian biomass in the sub-polar and polar regions of the southern
hemisphere and as such are chief consumers of the food resources in the Southern Ocean
(Prevost 1981, Croxall & Lishman 1987). Given the crucial role the penguins play in the
marine ecosystem, it is not surprising that they represent one of the most studied southern
seabird family (e.g. Williams 1995).
However, until the early 1980’s most of the research focused on those aspects of the penguins’
biology that could be studied on land, such as breeding biology, behaviour and physiology
(e.g. Murphy 1936, Richdale 1951, Stonehouse 1975). This was primarily due to the logistic
diffi culties involved in collecting at-sea data on birds that spend most of their time in the
water submerged (Wilson 1995). As a result, information about the penguins’ marine ecology
was limited to what could be deduced, for example, from the food they brought ashore (e.g.
Warham 1975, Croxall & Prince 1980) or the time the penguins stayed away from the nest
(Wilson 1995). It was the development of miniaturised devices to record swimming speed and
dive depths in the early 1980s that made it possible to get a deeper insight into the penguins’
foraging ecology (Kooyman 2004, Wilson 2004).
Penguins can be broadly categorised into inshore and offshore foragers. Inshore foragers
are generally resident at their breeding sites throughout the year and undertake only short
foraging trips to sea. Offshore foragers, on the other hand, perform longer foraging trips
– particularly during the courtship and incubation period – and migrate away from their
breeding sites outside the breeding season (Croxall & Davis 1999). Whether a penguin species
adopts an inshore or offshore foraging strategy relates to the latitude at which it breeds.
Penguins living in polar and subpolar regions (e.g. crested penguins Eudyptes spp.) generally
employ an offshore foraging strategy and are absent from their breeding sites when oceanic
productivity is low in winter (Davis & Renner 2003). Conversely, penguin species breeding
at lower latitudes (e.g. Humboldt penguins Spheniscus humboldti) benefi t from a year-round
1. Penguin foraging behaviour and the marine environment 3
food availability at the breeding sites that is less infl uenced by seasonality. However, the
latitudinal distribution of inshore and offshore foraging strategy is not clear cut and fi xed
within a species. This is especially true when factors other than seasonality have an infl uence
on the availability of penguin prey. If prey is scarce at a certain location, a species which
elsewhere exhibits an inshore foraging strategy might display patterns that resemble those of
typical offshore foragers (see “Little Penguins” in Davis & Renner 2003, pp. 75-79).
1.2 Infl uence of the marine environment
Incidence of light is the key factor that determines the seasonal variation in productivity of
the oceans from the temperate to Polar Regions. However, while light delivers the energy
required to stimulate photosynthesis, nutrients and trace elements transported in the oceans
represent the building blocks for the accumulation of phytoplankton biomass which forms the
base of the oceanic food web (Mann & Lazier 2006). Without nutrients, primary production is
limited (e.g. Boyd et al. 1999). The availability and distribution of nutrients depends largely
on physical processes in the marine environment such as currents. This is especially relevant
for the open ocean where there is no nutrient infl ux from continental landmasses (Murphy et
al. 2001).
Almost a century ago, it was realized that the distribution of penguins correlates with the
presence of cool, oceanic currents that feature high oceanic productivity (Boubier 1919).
Murphy (1936) noted that certain current systems such as the Humboldt or Falkland current
systems west and east of South America, lead to upwelling of bottom waters. This process
transports nutrients from the deeper reaches of the ocean towards the surface, where it
fuels primary production and the availability of prey for seabirds. The interplay of oceanic
processes and seabird distribution was further refi ned with advances in technologies to
monitor oceanographic variables. From the 1980s onward, seabird ecologists started to
understand the infl uence of oceanic fronts (i.e. areas where water masses with different
properties like temperature or salinity meet) on seabird prey and ultimately the distribution
of seabirds (Hunt Jr & Schneider 1987). In this light, it is not surprising that in some penguin
1. Penguin foraging behaviour and the marine environment 4
species, foraging was found to be directed towards frontal systems (e.g. king penguins,
Charrassin & Bost 2001; royal penguins, Hull et al. 1997). Nevertheless, we are only
beginning to understand how foraging and, tied to that, the breeding behaviour of penguins is
infl uenced by processes in the marine environment.
1.3 Foraging ecology of New Zealand penguins
The New Zealand region represents a unique area to investigate how distribution, foraging
and breeding behaviour of penguins are interwoven with the oceanic ecosystem. Six different
penguin species breed in New Zealand territorial waters, two of which show all the hallmarks
of inshore foragers (Little penguin Eudyptula minor and the Yellow-eyed penguin Megadyptes
antipodes), while the remaining species are all typical offshore foragers (Snares penguins
Eudyptes robustus, Fiordland penguins E. pachyrhynchus, Erect-crested penguins E. sclateri
and Rockhopper penguins E. chrysocome). With the exception of the Little penguin, all of
the New Zealand penguin species, breed along the southern coast lines of the South Island,
on Stewart Island and on New Zealand’s Subantarctic Islands (Williams 1995), in regions,
therefore, that are dominated by important oceanographic features – namely the Subtropical
Front and the Subantarctic Front (Heath 1981). These features are likely to have a substantial
infl uence on the penguins’ marine ecology. Considering that penguin populations rise and fall
with the prey availability of the marine environment they are exploiting (e.g. Cunningham
& Moors 1994), a basic understanding of their foraging ecology is vital to assess their
population status and subsequently conservation measures.
Although penguin research has a long tradition in New Zealand (Warham et al. 1986), only
a few studies have focused on the penguins’ foraging biology. Seddon & van Heezik (1990)
examined the maximum diving depths of Yellow-eyed penguins, which was followed by
a more extensive study of Yellow-eyed penguin foraging ranges by Moore et al. (1995).
Foraging ranges and diving behaviour of Little penguins were studied for one season by
Mattern (2001) and most recently some information about the foraging ranges of Rockhopper
penguins was published by Sagar et al. (2005). Most of these studies provided descriptive
1. Penguin foraging behaviour and the marine environment 5
accounts that shed some light on basic facets of the penguins’ at-sea behaviour. Yet, the
underlying factors determining the penguins’ behaviour, i.e. the role of marine processes,
were only briefl y, if at all, considered. Certainly, the main reason for this shortcoming was the
technological limitations of the methods employed; limitations that have now been overcome
with the recent development of GPS based data loggers.
1.4 Snares vs. Yellow-eyed penguin foraging - Aims of the thesis
Considering that there are few data on the foraging behaviour in New Zealand penguins,
I set out to study the at-sea ecology of two New Zealand penguin species, the offshore
foraging Snares penguin, Eudyptes robustus, and the inshore foraging Yellow-eyed penguin,
Megadyptes antipodes. The two species not only employ different foraging strategies but live
in contrasting habitats. Snares penguins breed on the offshore Snares archipelago south of
New Zealand, while the Yellow-eyed penguins studied in the course of this thesis inhabit the
coastal regions of the New Zealand mainland and Stewart Island.
Little is known about Snares penguins and the only comprehensive published account dates
back to the early 1970s (Warham 1974). Nevertheless, the life history traits summarized in
Warham’s paper allow predictions about the Snares penguins’ offshore foraging strategy.
Snares penguins are absent from the Snares Islands outside the breeding period. Migration
outside the breeding period is a hallmark of offshore foraging penguins and primarily stems
from seasonal availability of prey in the marine environment (Croxall & Davis 1999).
Seasonality has the greatest effect at high trophic levels (Mann & Lazier 2006) and it can be
predicted that breeding Snares penguins target primarily planktonic prey. If so, the penguins’
foraging ecology must be greatly infl uenced by the oceanography around the Snares. The
Subtropical Front, which arches around the South and East of the island, represents a
dominant feature where nutrients accumulate and that supports elevated levels of primary
production (Heath 1981, Murphy et al. 2001) and as such might be targeted by the penguins.
Whether the front might be a viable destination for chick rearing penguins is questionable,
though, as Warham (1974) reports that the penguins are absent from the nest for relatively
1. Penguin foraging behaviour and the marine environment 6
short periods of time which makes it unlikely that the birds have enough time to travel the
200 km towards the front. Hence, the increasing oceanic productivity around the Snares must
provide the penguins with adequate food resources during the later stages of breeding.
In contrast to Snares penguins, Yellow-eyed penguins are resident at their breeding locations
throughout the year (Seddon & Darby 1990) and as such apparently unaffected by seasonality.
Not surprisingly their main prey has repeatedly been found to comprise primarily of benthic
fi sh species (van Heezik 1988, Moore & Wakelin 1997) and, thus, exploit trophic levels that
are considerably lower than what can be suspected for Snares penguins. Furthermore, their
preference for benthic prey suggests that the Yellow-eyed penguins’ foraging patterns might
be de-coupled from broad-scale oceanographic processes as the distribution of benthic fi sh
species is less affected by currents. Support for this prediction comes from a VHF telemetry
study which showed that penguins retained individual foraging sites that they revisited
at different times of the breeding season and in different years (Moore et al 1995). If this
information is considered in the light of the primarily benthic foraging, it can be predicted that
the Yellow-eyed penguins’ foraging behaviour is likely to be more infl uenced by the benthic
environment – or more precisely local benthic features – than any other oceanographic
feature.
The study of foraging behaviour in penguins requires the aid of sophisticated technology
(Kooyman 2004). Such technology was used to either determine foraging movements using
transmitting devices (i.e. VHF or satellite telemetry) or to record behaviour by fi tting data
loggers to the birds (e.g. to record dive depths) (e.g. Mattern 2001). Only in the past fi ve
years new devices became available – GPS data loggers – that not only record spatial and
behavioural information at the same time, but also are small and robust enough to be deployed
on penguins (Wilson 2004). The GPS technology differs fundamentally from the traditional
methods of telemetry, so this thesis starts off with a detailed description of the functionality
and the underlying technology of the GPS data loggers I used during the course of the
research (Chapter 2).
1. Penguin foraging behaviour and the marine environment 7
For an interpretation of penguin foraging data, it is essential to have at least a basic
knowledge about the species’ prey composition. While the diet of Yellow-eyed penguins
has been studied in great detail in the past (van Heezik 1988, Moore & Wakelin 1997), our
knowledge of the diet of Snares penguins is sketchy at best (Marchant & Higgins 1990,
Cooper et al. 1990). Therefore, I examined the diet composition of the penguins during the
incubation and early guard-stages of the breeding season 2002 (Chapter 3). This was followed
by deployments of GPS loggers and time-depth recorders (TDR) in the seasons 2003 and
2004 to examine foraging behaviour. In 2003, the research focussed on the relationship of
seasonal changes (i.e. primary production) in the marine environment to the Snares penguins’
foraging and nesting patterns during incubation. I examined how at-sea movements relate
to oceanographic features within range of the penguins (Chapter 4). The following year, the
emphasis was on the foraging behaviour of the female penguins that provide food for the
chicks while the males guard the nest (Warham 1974). At this stage, the females’ foraging
ranges are constrained by their offspring’s demand for food (e.g. Davis & Renner 2003),
which means that oceanographic features important during incubation are beyond reach
for the females. Nevertheless, I found that oceanographic variables infl uenced the foraging
patterns of the females (Chapter 5).
With the aid of the GPS data loggers I studied the foraging patterns of Yellow-eyed penguins
from Oamaru in great spatial detail and tried to assess to which degree the birds might have
adapted to a spatially conservative, target-oriented foraging strategy (Chapter 6). To assess
how fi xed such patterns are within the species, I studied the foraging behaviour of Yellow-
eyed penguins from Stewart and Codfi sh Island (Chapter 7). Penguins from these locations
show marked differences in reproductive success which seems to be related to differences
in prey availability despite the proximity of both locations. I used the results from these
deployments to assess the plasticity of the penguins’ foraging behaviour in the light of
apparently differing food situation at both locations.
1. Penguin foraging behaviour and the marine environment 8
1.5 Thesis outline
Each chapter forms the framework for a scientifi c paper and either has been, or will be,
submitted for publication in peer-review journals. Hence, there might be repetition of
information in a number of chapters.
CHAPTER 2, How to get the most (or anything) out of GPS loggers: a case study with Snares
penguins, represents a methods paper that provides detailed insight into the methodology and
the underlying technology of the new GPS data loggers that were essential for the study of
Snares and Yellow-eyed penguins’ foraging behaviour.
CHAPTER 3, The diet of breeding Snares penguins Eudyptes robustus, provides information
indispensable for the interpretation of any foraging data – the general prey composition of
Snares penguins. I discuss the fi ndings in light of what is known from other crested penguin
species with special regard to the New Zealand Rockhopper penguins which are believed to
be in substantial decline because of changes in prey availability (Thompson & Sagar 2002).
CHAPTER 4, Infl uence of Oceanography and Seasonality on Foraging Behaviour and
Nesting Patterns of Snares Penguins Eudyptes robustus, examines the synchrony of nesting
patterns in Snares penguins and how these might be a result of improving foraging conditions
due to seasonal changes in the marine environment. The chapter highlights the relationship
of the penguins’ at-sea movements during the incubation period, the presence of an
oceanographic front and the occurrence of patches of high oceanic productivity.
CHAPTER 5, Foraging ranges and spatial distribution of dive activity in female Snares
penguins Eudyptes robustus, describes the foraging behaviour of female Snares penguins
during the chick-guard stage, when foraging ranges are greatly limited because the chicks
need to be fed frequently. The geographic distribution of the females’ diving activity is
examined with special regard to the marine environment and provides data that underpins the
importance of New Zealand coastal waters for the penguins’ foraging direction.
CHAPTER 6, Conservative foraging routes and exclusive bottom feeding behaviour in the
1. Penguin foraging behaviour and the marine environment 9
Yellow-eyed penguin Megadyptes antipodes, gives new detailed insights into the foraging
behaviour of Yellow-eyed penguins. The chapter provides the fi rst quantitative and qualitative
analysis of the primarily benthic foraging behaviour of Yellow-eyed penguins. It describes
how the birds follow conservative and almost stereotypic foraging routes and shows that
conservative foraging is facilitated by navigating along bottom features during benthic dives.
CHAPTER 7, Plasticity in foraging behaviour of Yellow-eyed penguins Megadyptes antipodes
from neighbouring breeding locations, Stewart and Codfi sh Islands, New Zealand, examines
the foraging behaviour of Yellow-eyed penguins breeding at adjacent breeding sites and
underlines the plasticity of foraging behaviour within a penguin species that is apparent even
at neighbouring locations. The fi ndings are discussed with reference to the differences in
breeding success and penguin numbers at both sites.
CHAPTER 8, General discussion, provides a summary of the chapters, draws general
conclusions with regard to the difference and similarities in foraging patterns of Snares and
Yellow-eyed penguins and provides some recommendations for future research.
APPENDIX, Conference abstracts, lists titles and abstracts of conference presentations
presented during the course of this PhD.
2. How to get the most out of GPS loggers 10
CHAPTER 2
HOW TO GET THE MOST (OR ANYTHING) OUT OF GPS
LOGGERS: A CASE STUDY WITH SNARES PENGUINS*
* Paper presented at the second international symposium on the remote biological monitoring of
animals ‘Bio-logging II’, June 2005, University of St. Andrews, Scotland
2. How to get the most out of GPS loggers 11
2.1 Introduction
The study of seabird foraging is a challenging endeavour. As most seabird species are wide-
ranging, fast-moving and live in logistically relatively inaccessible areas, researchers have
to rely on information gathered by sophisticated and often expensive electronic aids. These
can generally be categorized as transmitting devices and data loggers. VHF and satellite
transmitters are primarily used to determine movement patterns (e.g. Weavers 1992, Culik
et al. 2000). Data loggers are the key element to measure behaviour (e.g. diving activity and
depths; Schiavini and Rey 2004), physiological parameters (e.g. heart and respiration rate,
meal size, etc.; Wilson et al. 2003, Wilson 2004) and physical parameters of the environment
(e.g. temperature; Charrassin et al. 2004). Until recently, studies tackling spatial movements
and behaviour at the same time had to use both methods either by combining transmitters
with data loggers (e.g. Hennicke 2001) or by deploying transmitters and loggers separately on
representative samples of a population (e.g. Mattern 2001). The development of data loggers
that are capable of determining geographic position via Global Positioning System satellites
(GPS) marked a successful amalgamation of both methodologies.
GPS loggers should not be confused with satellite transmitters. Satellite transmitters – often
referred to as platform terminal transmitters or PTTs – emit signals to satellites which
triangulate the transmitters’ geographic position; the data is then provided (against payment)
by the Argos System (Argos CLS, Toulouse, France, Wilson et al. 2002). In contrast, a GPS
logger comprises a receiving unit that itself uses signals transmitted from GPS satellites to
triangulate its position (Wilson 2004). GPS data are stored in an internal memory and can be
downloaded after recovery of the device and no major costs arise. In addition to the GPS unit,
the devices may carry the usual array of environmental sensors such as pressure transducers
or temperature sensors.
In wildlife research, GPS technology has been used for a number of years now, primarily on
large terrestrial animals such as wolf Canis lupus (Merrill and Mech 2000) and moose Alces
alces (Edenius 1997). With advancing technology the devices got smaller and lighter so that
2. How to get the most out of GPS loggers 12
they could be fi tted on birds like albatross (e.g. Weimerskirch 2002) and pigeon (e.g. von
Hünerbein et al. 2000, Lipp et al. 2004). Most recently, GPS data loggers became available
that were not only robust enough to be deployed on marine vertebrates but also suffi ciently
compact to be fi tted to penguins (Wilson 2004). However, the use of GPS loggers on diving
animals poses some challenges to the technology that are of less concern when used on
terrestrial animals or fl ying birds.
GPS signals are transmitted from satellites as radio waves and as such do not penetrate water
(Kaplan 1996). As a consequence, there is no GPS reception under water. This means that
when data loggers are deployed on penguins – or any diving animal – the GPS functionality
is limited to surface periods. Furthermore, position acquisition of the GPS unit is not
instantaneous but requires a certain amount of time (“time-to-fi x”). The time-to-fi x depends
on several factors and may range from a few seconds to several minutes. On penguins it is
obviously desirable to keep the time-to-fi x at a minimum to ensure that a position can be
determined before the animal dives again. Finally, the energy requirements of continuously
operating GPS units greatly limit the operation time of GPS loggers (Ryan et al. 2004).
Although the battery lifetime of GPS loggers can be considerably extended by programming
the device to regularly switch-off, such duty-cycling increases the time-to-fi x (G. Peters,
personal communication). In order to maximise the data outcome of GPS logger deployments
it is therefore essential to consider these issues in the context of the expected foraging
behaviour (e.g. foraging trip length, dive frequency, surface time).
Here, we give a brief overview of the underlying principle of GPS technology, summarise
our experience using GPS data loggers on Snares penguins Eudyptes robustus and discuss
behavioural and, thus, species related issues of this new technology.
2. How to get the most out of GPS loggers 13
2.2 Methods
2.2.1 Study site and species
We studied the foraging behaviour of Snares penguins using GPS loggers during three
consecutive seasons between 2002 and 2004. The Snares penguin breeds only on the Snares
(S48°01’, E166°36’), a small subantarctic island group covering a total land area of less than
3.6 km² approximately 120 km south of New Zealand’s South Island (Fig. 2.1). Information
about the Snares penguin’s foraging ecology was virtually non-existent A reason for this
was that the technology to determine at-sea movements was either unsuitable or lacked the
accuracy needed. VHF-telemetry was not an option as there are no adequate vantage points
for receiving stations on the Snares and the small size of the island would have increased
triangulation errors to an unacceptable degree (see Zimmerman and Powell 1995). Satellite
telemetry with its infrequent fi xes and poor accuracy (Wilson et al. 2002) would have been
only reasonable to deploy when the penguins were leaving on longer foraging trips, i.e. during
the over-wintering period or the incubation stage of breeding.
Field work was carried out between October and November of each year covering the late
incubation and the chick-guard stages of breeding. Our main study site was colony A3, with
ca. 1200 pairs one of the largest colonies on the Snares’ main island. The penguin colony is
located on a large clearing in otherwise dense forest (mostly tree daisy Olearia lyalli), and
connected to the coast by a penguin path that runs under dense forest cover for some 500 m
before emerging at Station Cove (Fig. 2.1).
2.2.2 GPS loggers and deployment
In 2002, we used prototype GPS loggers from Sirtrack (Sirtrack Ltd., Havelock North,
New Zealand). However, after successful trials on land, none of the devices collected any
GPS data when deployed on eight penguins. As a consequence, we used GPS-TDlog data
loggers (earth&OCEAN Technologies, Kiel, Germany) in the following years 2003 and
2. How to get the most out of GPS loggers 14
2004. The design of the GPS-TDlog is described in detail by Ryan et al. (2004) so that
only a brief summary of its features will be given here. The GPS-TDlog (dimensions:
L100xW48xH24 mm, mass: ca. 70 g) comprises a GPS receiver as well as high precision
depth (resolution: ~0.1 m) and temperature (resolution: ~0.005 K) sensors. GPS and sensor
data are stored in non-volatile 2 Mb Flash memory at programmable intervals. GPS data
consist of a timestamp, geographic coordinates (lat/lon), corresponding HDOP (horizontal
dilution of precision, a measure of GPS accuracy, see Dessault et al. 2001) and horizontal
speed over ground between fi xes. Only fi xes with HDOP <12 are stored which translates to
a position error <20 m for most position fi xes (Ryan et al. 2004). The device is powered by a
CR123A lithium photo battery and enclosed in a streamlined aramide fi bre/epoxy-composite
housing with an O-ring-sealed cap that allows easy battery exchange and data retrieval.
In 2003, we deployed GPS-TDlogs on four males and two females leaving on long trips
during incubation. During chick guard, when only the female penguins forage (Warham
1974), a total of eight birds were fi tted with GPS-loggers. In the following season 2004, two
females were equipped during incubation (no males were deployed due to late start of fi eld
work). During chick guard that year, 16 females were fi tted with loggers.
In 2003, nine birds were fi tted with loggers in the colony, but this practise was found to be
detrimental due to GPS limitations (see discussion) as well as aggressive reactions of the
equipped birds’ mates towards the device. The remaining three logger deployments in 2003
and, with the exception of the two incubating birds, all deployments in 2004 occurred in
Station Cove where we could ensure optimal GPS reception shortly before the birds launched.
For that, we marked females in the colony with a small spot of water soluble stock dye
sprayed on their breast. After marking birds the penguin traffi c coming out of the forest at
Station Cove was observed and marked birds were captured for logger deployment on their
way to the sea. For device recovery, penguins were either re-captured shortly after landing in
Station Cove or if encountered on the nest during daily monitoring in the colony.
The entire handling procedure lasted on average 13 minutes. To minimize stress, the penguins’
heads were covered with a dark cloth hood. All devices were attached to the penguins’ lower
2. How to get the most out of GPS loggers 15
back using black TESA-tape (TESA tape, No. 4651, Baiersdorf AG, Hamburg, Germany)
following Wilson et al. (1997). When attached, the devices were entirely covered with 6-7
overlapping strips of tape leaving only the device’s posterior sealing cap with the sensors free.
During incubation, loggers were deployed for the duration of one trip. During chick guard, the
loggers were recovered after 48 to 76 hours during which some birds performed two foraging
trips. Some birds avoided recapture for up to six days.
2.2.3 GPS basics and logger setup
An important factor before the deployment of GPS loggers is the programming of the GPS
unit. For this, one needs to be familiar with some of the basic terms and underlying principles
of the Global Positioning System.
GPS data are constantly transmitted as radio waves from satellites in earth’s orbit. Each
satellite transmits precise time according to its internal clock as well as information about
its exact orbit, the “ephemeris data”. For the successful triangulation of a position fi x, the
logger’s GPS receiver needs valid ephemeris data from at least three satellites that are in view
(Kaplan 1996). Environmental features, such as topography or vegetation block reception
from some satellites and, thus, reduce the number of satellites in view to calculate a position
fi x. While continuously operating GPS loggers follow the changes of the satellite constellation
and update ephemeris data in real-time, GPS loggers that operate intermittently have to
reconfi rm the satellite data after waking up. During a logger’s sleep phase known satellites
(i.e. satellites for which valid ephemeris data is stored in the logger’s memory) might have
moved behind obstacles or the horizon, while new satellites have come into view. Depending
on how many known satellites are still in view after a logger wakes up, it might be necessary
to download GPS data from new satellites before a position can be determined. The validity
of ephemeris data after a sleep period is referred to as “GPS mode”. In “hot mode”, the
logger has valid data for at least four satellites currently in view, which allows the devices to
quickly determine its position (for GPS-TDlog between 6-22 s, earth&OCEAN Technologies)
while at the same time download ephemeris data from new satellites. In “warm mode” the
2. How to get the most out of GPS loggers 16
logger needs to download ephemeris data from new satellites fi rst before a position can be
determined. Since GPS data is transmitted at low data rates, it takes at least 30 s to download
a whole ephemeris data set from a satellite (Zogg 2002). Any signal disruption (e.g. short
dive) increases the download time. Consequently, once fallen back into “warm mode”, the
time-to-fi x of a logger is prolonged considerably. Deploying loggers that sample continuously
reduces the risk of the device falling back into warm mode, but comes at the cost of greatly
reduced battery lifetime (for GPS-TDlogs ca. 12 hours; Ryan et al. 2004).
For long trips (7-14 days, i.e. during incubation, mid- to late October), GPS loggers were
set to record environmental (i.e. sensor) data at 5 s intervals. The GPS unit was programmed
to wake-up every 20 min (intermittent mode) during periods of inactivity, but to operate in
pressure controlled mode (“upon resurfacing”) during periods of dive activity. The pressure
control overrides the programmed sampling intervals and switches the device on after
every dive. In hindsight, it would have been better to de-activate the pressure control as this
operation mode reduced the battery lifetime so that it was unlikely to obtain data for entire
foraging trips. However, at this stage we had no experience with the GPS loggers and opted
for a setting that would be most likely to yield GPS data.
For short trips (1-3 days, i.e. during chick-guard, from early November), we used a 1 s sensor
interval while the GPS unit was set to 2 min intervals with activated pressure control. In
one case, a GPS logger was programmed to operate in the energy-consuming continuous
mode, where the GPS unit was never switched off and positions were stored every 1 s. When
deployed in the colony, GPS loggers were set to start sampling at 4 am the next morning. All
devices deployed in Station Cove, were started before the penguins were captured to ensure
that the GPS unit had successfully determined its position by the time the penguins left the
island.
2.2.4 Data analysis
Dive data were analysed with custom written software in Matlab 6.5 (Release 13, Mathworks,
Inc., Natick, MA, USA). Dive analysis encompassed identifi cation of dive events and
2. How to get the most out of GPS loggers 17
calculation of basic dive parameters for every dive (e.g. duration, max depth, surface interval
between dives). Dive events were only accepted if depths >1 m were registered to compensate
for erratic pressure fl uctuations when the birds were at the surface. Mean±SD are given for
normally distributed data otherwise median and range were used as summarizing fi gure.
2.3 Results
2.3.1 Data outcome and battery life
Table 2.1 gives an overview of the deployments in both years and the respective data
outcome. Of eight deployments of GPS-TDlogs on Snares penguins in the incubation stage
of breeding (four males in 2003, and two females in each 2003 and 2004), three deployments
resulted in GPS and sensor data for the fi rst 2-3 days of the foraging trips. The battery life
during these deployments ranged between 3 and 5 days. However, in all three cases the
programmed logger start-up did not coincide with the birds’ departure from the island. The
penguins left only between 26 h to 60 h (median: 26 h) after the devices started recording,
which encompassed 28-55% of the loggers’ total operating time before the loggers were
exhausted. One male did not leave its nest at all and the device was recovered fi ve days after
the programmed logger start. Deployments of GPS-TDlogs on two incubating females in
each 2003 and 2004 were unsuccessful. One female failed to return from her long foraging
trip, two devices were recovered waterlogged and irreparably damaged and one device
malfunctioned due to damage to its circuit board.
In both years, we fi tted GPS loggers to 24 females with chicks that left on a total of 27 short
foraging trips (Table 2.1). In 2003, eight females fi tted with loggers left on a total of 11
foraging trips. Complete sets of GPS and sensor data were recorded for fi ve trips performed
by fi ve different birds. The six remaining trips performed by three individuals yielded only
sensor data. On these three birds, the GPS units recorded only position fi xes when the
penguins were on the nest; one of the units was programmed to store GPS fi xes continuously.
Overall the battery life that year ranged between 12.3 h (for the continuously logging device)
2. How to get the most out of GPS loggers 18
and 64.9 h.
In 2004, 16 females fi tted with GPS loggers performed a total of 17 foraging trips.
Unfortunately one device malfunctioned, so data are available for 16 trips. Of these, GPS and
sensor data were obtained for 14 trips, two penguins returned with sensor data only which
was due to a malfunction of the loggers’ GPS units. On fi ve trips the batteries were exhausted
before the birds had returned to the island, but the data recorded still covered substantial parts
of the trips (Table 2.1). The overall battery lifetime ranged between 26-64 h (median 43.3 h).
2.3.2 GPS performance and dive behaviour
The likelihood of a GPS fi x acquisition was related to the time a penguin spent at the surface
between dives. Generally, if a penguin spent less than 20 s at the surface, the loggers were
unable to determine and store a position fi x (Fig. 2.2a). With increasing surface time, the
occurrence and, thus, likelihood of a GPS fi x acquisition increased, and surface times of
45 s or longer generally had a chance of fi x acquisition of 50% or more. However, when
the penguins were active (i.e. during the day between 5 am-10 pm) only 7% of all observed
surface intervals (n = 28922) were 45 s or longer whereas the majority (55%) was shorter
than 20 s (Fig. 2.2b). The frequency of occurrence of longer surface intervals, varied between
birds. This is refl ected in their individual median surface times that range between 11-35 s
(average median surface time: 18±6 s, n = 19 birds). The surface interval itself was related
to individual dive behaviour. Penguins that mainly performed short and shallow dives also
exhibited shorter surface intervals (Pearson correlation – surface time vs. dive time: r = 0.600,
p = 0.018; surface time vs. dive depth:. r = 0.627, p = 0.012). Therefore, it is not surprising
that the number of GPS fi xes varied greatly between deployments. Overall the number of
fi xes recorded during single foraging trips ranged from 23 to 816 fi xes. With regard to total
trip length this corresponds to a hourly fi x rate 1.4 fi xes*h-1 to 12.1 fi xes*h-1 (mean: 7.2±3.2
fi xes*h-1) . However, this fi x rate did not refl ect the true temporal distribution of GPS fi xes
over the course of a foraging trip. During all deployments, the maximum time interval
between two successive daytime GPS fi xes ranged from 0.3 h to 13.8 h (median: 6.2 h). In the
worst case GPS fi xes were stored only at night (Fig. 2.3, top graph).
2. How to get the most out of GPS loggers 19
2.4 Discussion
GPS-loggers are a huge advance for tracking penguins. This technology does not require
the laborious efforts of VHF telemetry and delivers position fi xes of foraging seabirds with
an accuracy far superior to that of satellite telemetry (Wilson 2004). Nevertheless, this
new technology still has its limitations. Especially the short battery life and varying GPS
performance can reduce data outcome and quality, particularly when used on diving animals.
These limitations are largely rooted in the principle technology of the Global Positioning
System. In the case of the Snares penguin the loggers’ performance depended to a great
extent on dive behaviour. In order to make the most of GPS logger deployments it is therefore
necessary to identify and attempt to mitigate the effects of site and species-specifi c factors that
contribute to a deterioration of GPS performance.
2.4.1 Deployment location and disadvantages of programmed start
Fitting loggers in the penguin colony had several disadvantages. Particularly during the
incubation stage, it was diffi cult to accurately predict the penguins’ departure time. As a
result some of the loggers started operating more than a day before the penguins actually
left the island and, thus, valuable battery time was lost on land. One deployment even failed
completely as the penguin did not leave its nest for more than a week after the logger started
operating. This illustrates the disadvantages of a programmed logger start.
Apart from delayed departures, deploying loggers in the colony had other disadvantages.
Of fi ve colony deployments on females in 2003, only two resulted in at-sea GPS data, while
the other three only produced position fi xes at the nest sites. During both of the successful
deployments, the loggers stored a fi x shortly before the penguin entered the water and were,
thus, in “hot mode” (see Methods). That three loggers failed to obtain GPS fi xes at sea
indicates a poor GPS performance which was likely to be the result of the loggers’ GPS units
being in the detrimental “warm mode” when the penguins left the island. This, in turn, was
probably a result of the location of the logger deployments. In the colony the surrounding
2. How to get the most out of GPS loggers 20
topography and vegetation limited GPS reception. Furthermore, the penguins’ behaviour
and incubating postures at the nest further might have contributed to limitations of GPS
performance. Standing upright and/or facing the back towards the partner can result in a
detrimental alignment of the GPS antenna which also reduces GPS reception. This situation
often occurred when the mates changed incubation duties and the penguin with logger did
not leave the colony afterwards. During the 45 min walk through the forest between colony
and sea (see Fig. 2.1), GPS reception was practically non-existent because of signal shielding
forest canopy and the vertical alignment of the GPS antenna on a walking penguin. The
negative effects of colony deployments were largely mitigated by equipping birds shortly
before they left the island. The three Station Cove deployments in 2003 all returned at-sea
GPS data. In 2004, when all chick-guard deployments happened in Station Cove only two
deployments did not return at-sea GPS data due to a logger malfunction.
Logger deployment shortly before a penguin left on a foraging trip had two major advantages.
Firstly, it took the guesswork out of programmed logger starts and effectively reduced logging
time “lost” on land. Secondly, starting the logger when the penguin was about to leave the
island allowed us to check via computer that the loggers were in “hot mode” before the
devices were fi tted. Thus, the GPS units were able to acquire position fi xes in the shortest time
possible when the penguins launched on their trips.
2.4.2 GPS performance at sea
The GPS functionality is limited to the periods a penguin spends at the surface. To ensure a
position is determined before the animal dives again, a short time-to-fi x is imperative. If a
device fails repeatedly to update its satellite data and acquire a fi x during successive surface
intervals, the likelihood of known satellites moving out of range and the GPS unit falling back
into “warm mode” increases. Once this has happened, a longer period of undisrupted GPS
reception is required in order to update the satellite data.
According to the GPS-TDlog manual, loggers with intermittent setup are estimated to require
between 6 s and 22 s to acquire a fi x in “hot mode”. However, these values derive from
2. How to get the most out of GPS loggers 21
tests on land and as such could not take into account erratic surfacing patterns of foraging
penguins. Frequently, the Snares penguins performed short and shallow dives over long
periods (Fig. 2.3, top graph) which meant that surface breaks between dives were often too
short for successful GPS fi x acquisition. Penguins that dived deeper exhibited longer surface
intervals and, thus, their GPS unit had a higher chance to successfully determine a position
before the next dive. Depending on the proportion of shorter surface intervals during a trip,
the performance of the GPS logger varied between deployments.
Besides the length of surface intervals other factors further infl uence GPS performance at sea.
Figure 2.4 shows frequency distributions of surface interval lengths of two male penguins
during incubation 2003. Although male T32 (lower graph) showed longer surface intervals
between dives, the number of surface intervals that resulted in GPS fi xes was considerably
lower than what was observed in male T13 (top graph). As a result over the course of 2.4 d
only 127 GPS fi xes were recorded for T32, whereas the logger on T13 stored a total of 478
GPS fi xes during 2.6 d of operation. The reason for such different GPS performance is most
likely to be a result of several factors. Firstly, surface intervals determined from dive data do
not necessarily mean that a penguin is resting. As dive events were only accepted when the
sensor detected depths of one meter or more, immediate sub-surface dives (<1 m) were not
considered during analysis and were part of a surface interval. Other surface behaviour of
penguins such as bathing or preening often involves rolling or dipping which means that the
device is frequently submerged which greatly affects GPS reception. Even resting penguins
might tilt their body when at the surface (Healy et al. 2004) so that the GPS antenna might
be in a suboptimal alignment even though the penguin is inactive. Other non-behavioural
factors also might have contributed to lower GPS performance. The adhesive tape we used
to attach the devices tends to become soggy in sea water over time. Particularly in the area
of the GPS antenna, the moist tape can act as barrier for GPS signals (G. Peters, personal
communication). Other factors, such as temporary satellite constellation or wave height,
further affect GPS performance at sea.
2. How to get the most out of GPS loggers 22
2.4.3 How to get the most out of GPS loggers
On diving animals, it is necessary to make sure that a GPS device is in the best possible
condition when its bearer goes to sea. While the ways of achieving this might be different
from location to location, there are basic rules of thumb that can be followed.
The programmed logger start-up should only be used when the departure time is predictable to
avoid unnecessary waste of battery time on land and to ensure that the logger is able to store
a position fi x before the animal enters the water. In this light, the location of the deployment
should be considered prior to the logger attachment to ensure that the GPS unit has the best
possible reception between deployment site and sea.
The device should be attached so that the internal GPS antenna’s alignment allows the best
possible view of the sky when the animal is at sea. Furthermore, any additional coverage of
a device’s antenna (i.e. tape, glue etc.) should be kept at a minimum to avoid unnecessary
shielding of the GPS signal.
The loggers should be programmed according to the expected foraging activity. In order to
cover longer time periods (>3 d), a time-based duty cycling routine is the key to extending
battery life to the expected trip duration. In this case, intervals between logger activation
should not be too long to reduce the risk of the logger falling back into “warm mode”. For
male Snares penguins on long trips, for example, a 15-20 minute duty-cycled, non-pressure
controlled GPS unit would probably have covered entire two week foraging trips. For short
trips (i.e. deployment period <3 d) pressure-based duty cycling of the device’s GPS unit is
the preferable option as this ensures best GPS performance and greatly minimizes the risk
of “warm mode”. With activated pressure control, any time-based GPS duty cycling is only
in effect when the animal is inactive and the GPS performance is least affected by diving
behaviour. It is therefore recommended to use low timed-based sampling rates (15-30 min) for
the GPS unit when pressure control is activated.
2. How to get the most out of GPS loggers 23
Table 2.1. Overview of GPS logger deployments and data outcome on Snares penguins in the breeding seasons 2003 and 2004. Trip duration and battery life
are given as Median (range). For logger setup see Methods (2.3). Data sets either comprised GPS & sensor data or sensor data only when the GPS unit failed to
store position fi xes at sea. Depending on trip duration and/or state of the battery upon departure, some data sets were incomplete, i.e. covered only a proportion
of the entire foraging trip.
Data outcome
GPS & sensor only sensor
Breeding stage Year sex No. of
deployments
No. of
trips
Trip duration
(h)
Battery life
(h)
complete
trips
incomplete
trips
(coverage)
complete
trips
incomplete
trips
(coverage)
Incubationa2003 male 4 3 244.8
(189.9-367.2)
93.4
(83.1-109.2)
03
(17%-27%)
--
Chick-guard 2003 female 8 11 29.3
(11.9-39.5)
30.8
(12.3-64.9)
50 33
(36%-51%)
2004 female 16 16b31.6
(13.1-72.7)
43.3
(26.1-64.3)
10 4
(77%-86%)
11
(43%)
atwo females were equipped in each 2003 and 2004 but the loggers were lost or damaged bone penguins performed two trips, one device malfunctioned
2. How to get the most out of GPS loggers 24
The Snares
Stewart Island
New Zealand
47°00’
48°00’
167°00’ 168°00’ 169°00’
EAST
SOUTH
Station Cove
Colony A3
200m
N
Figure 2.1. Overview of New Zealand and the Snares. Inset detail map shows the location of the
study colony A3 (grey area) and the path (black line) connecting the colony with the penguins’
departure point from the island in Station Cove. Strong grained area represents dense tree daisy forest
(no GPS reception), light grained area indicate coastal fringe without forest cover.
2. How to get the most out of GPS loggers 25
0
10
20
30
40
50
60
70
5 101520253035404550556065>
Surface interval (s)
Proportion of surface intervals
with GPS fixes (%)
n = 28922
0
5
10
15
20
5 101520253035404550556065>
Surface interval (s)
Frequency of occurence (%)
Figure 2.2. Frequency of surface intervals (i.e. time spent at the surface between dives) and
proportion of surface intervals with at least one GPS fi x in Snares penguins. Graphs were compiled
from pooled daytime data (5 am-10 pm) recorded with GPS loggers on three male and 19 female
penguins during incubation and chick-guard 2003 and 2004.
2. How to get the most out of GPS loggers 26
Figure 2.3. GPS performance as a result of diving behaviour during short trips performed by two
female Snares penguins. Top map gives example of a deployment with poor GPS data quality, bottom
map shows a trip with good quality GPS data. Position fi xes acquired are night (2200-0500 hrs) are
indicated by fi lled markers, hollow markers represent daytime fi xes (0500-2200 hrs). Arrows give
travel direction. Inset graphs represent hourly means of surface interval between dives (top), dive
depth (middle) and distribution of GPS fi xes (bottom) over the course of the respective foraging trip.
Filled bars indicate night time hours.
166°30’E166°40’E166°50’E
47°50’S
47°55’S
48°00’S
48°05’S020km
Time of day (h)
17 20 23 2 5 6 11 14
no. of fixes
0
15
30
dive depth (m)
100
50
0
surface time (s)
0
40
80
020km
no. of fixes
0
15
30
Time of day (h)
18 21 0 3 6 9 12 15 18
dive depth (m)
100
50
0
surface time (s)
0
40
80
166°20’E166°30’E166°40’E166°50’E
47°50’S
47°55’S
48°00’S
48°05’S
2. How to get the most out of GPS loggers 27
0
20
40
60
80
100
120
140
160
180
0 102030405060708090100>
Surface time (s)
Number of observations (n)
0
20
40
60
80
100
120
140
160
180
0 102030405060708090100>
Surface time (s)
Number of observations (n)
surface intervals with GPS fix
surface intervals without GPS fix
Male - T13
Male - T32
Figure 2.4. Frequencies of surface intervals without (grey bars) and with GPS fi xes (black bars) in
two male Snares penguins on foraging trips during incubation 2003. Total number of surface intervals
analysed were n = 1228 for top graph and n = 1146 for bottom graph.
3. Diet of breeding Snares penguins 28
CHAPTER 3
THE DIET OF BREEDING SNARES PENGUINS
EUDYPTES ROBUSTUS
3. Diet of breeding Snares penguins 29
3.1 Introduction
The Snares penguin Eudyptes robustus breeds only on the subantarctic Snares island group
some 200 km south of New Zealand’s South island. The species is one of the four crested
penguin species occurring in New Zealand waters and like all eudyptids is considered
a typical offshore forager (Croxall & Davis 1999, Davis & Renner 2003). Although the
species’ general biology has been described by Stonehouse (1971) and Warham (1974), little
information on the Snares penguins’ ecology has emerged in the past three decades. Yet the
fact that the Snares penguin population seems to be thriving makes the ecology of this species
particularly interesting as it might help to understand the underlying factors determining the
success or failure of other subantarctic penguin populations where declines have been the
norm (Ellis 2005).
The number of Rockhopper penguins breeding on Campbell Island are believed to have
declined by more than 90% in the last fi ve decades, possibly due to sea surface temperature-
related changes in prey availability (Cunningham & Moors 1994, Thompson & Sagar 2002).
In contrast to this, the Snares penguin population is considered stable (Williams 1995, Amey
et al. 2001) and numbers, in fact, appear to have increased between the 1960 and 1980s
(Warham et al. 1986). This suggests that if there have been any changes in abundance of the
Snares penguins’ principle prey, these changes either have had no effect on the penguins or
might have even improved the penguins’ situation.
The baseline information necessary to investigate the apparent success of the Snares penguins
and to compare their ecology with other less successful crested penguin species, is the
composition the Snares penguins’ diet. From accidental spillages at meal times Warham
(1974) deduced that cephalopods and crustaceans were the most important food components.
During dissection of dead Snares penguin chicks, it was found that the chicks’ food comprised
primarily of crustaceans, but also included some squid while fi sh seemed to play a minor
role (Cooper et al. 1990). However, due to their gross methodologies, neither of these
3. Diet of breeding Snares penguins 30
studies allowed for a more detailed analysis of the Snares penguins’ prey composition or a
determination of the importance of its main prey species.
In order to provide some baseline information on the Snares penguins’ diet, the stomach
contents of adult penguins returning from foraging trips were examined in detail during the
late incubation and early chick-guard stage of the breeding season 2002/03.
3.2 Methods
3.2.1 Stomach sampling
Stomach contents were collected at the main penguin landing site in Station Cove, North East
Island, the Snares (S48°09’, E166°36’), during late afternoon (1500-1800 hours) between 21
October 2002 and 8 November 2002. Breeding status of arriving birds could not be assessed
but penguins that did not loiter on shore after landing were chosen under the assumption
that they were more likely to be parents hurrying to feed chicks, whereas non-breeders were
unlikely to be as hurried. Suitable individuals were selected from groups of penguins heading
towards the pathways leading into the forest and to the colonies. A total of 24 adult Snares
penguins were captured with a butterfl y net shortly after landing.
After capture, bill measurements were taken to determine sex morphometrically following
Warham (1974) and then the bird was weighed in a cloth bag, so as to assess body condition.
The birds were relieved of their stomach contents by water-offl oading (Wilson 1984). Birds
were fl ushed until clear water indicated complete retrieval of stomach contents. No bird was
fl ushed more than three times; recovery breaks between fl ushes ranged between one and three
minutes. The entire stomach sampling process (i.e. from capture to release) lasted between 14
and 20 minutes.
Between 21 and 24 October, nine male penguins were sampled. Judging from their body girth
and mass (mean weight: 3.9±0.2 kg, n=9) the penguins were returning from long foraging
trips. The birds had very few identifi able items in their stomachs, which did not allow any
3. Diet of breeding Snares penguins 31
generalisations to be made about the males’ diet composition or determination of prey
quantities. As a result, further sampling of males returning from the long incubation-period
foraging trips was deemed unjustifi able. Between 02 and 08 November, when chicks had
hatched in most nests, stomach samples were collected from 15 female penguins. Only males
heavier than 3.5 kg (mean weight: 3.9±0.2 kg, n = 9) and females weighing more than 2.5 kg
(mean weight: 2.7±0.1 kg, n = 15) were considered suitable for sampling.
3.2.2 Sample processing and analysis
Following Ridoux (1994), each sample was divided into the “food load” (fresh fraction e.g.
entire specimens, fl esh remains of fi sh and squid) and “accumulated material” or hard-part
remains (i.e. squid beaks, fi sh otoliths). The food load represents the nourishing component
of food transferred to the chicks (i.e. the chicks’ diet), while accumulated material provides
additional information on fi sh and squid and hints at prey species that might be more relevant
to the adult penguins.
The food load was sorted for prey class, namely: crustaceans, cephalopods and fi sh. After
sorting, water was decanted from the subsamples before the material was transferred onto
fi lter paper to extract further excess liquid. Filter papers were replaced if saturated, until
water ceased to permeate from the material into the paper. The wet weight of each subsample
was determined to the nearest 0.1 g using an electronic bench scale, before the material was
transferred into storage containers with 99% ethanol until further analysis. Subsamples of the
food load were sorted for identifi able taxa and components that were beyond identifi cation.
Prey taxa were identifi ed using published keys (Crustaceans, Kirkwood 1982; Cephalopods,
Roper et al. 1969 and Lu & Ickeringill 2002; Fish, Paulin et al. 1989). Identifi ed prey taxa
were separated from the subsample and weighed after removal of excess liquid. Entire
specimens of a given taxon were measured to determine standard body length according to
Ridoux (1994) and individual wet weight.
Crustaceans (krill) represented a considerable portion of most samples and the number of
individual krill in a sample was large. As a consequence, a random crustacean subsample
3. Diet of breeding Snares penguins 32
(~3-4 g, ca. 100-150 individuals) was separated from the drained material and identifi ed in
detail. Wet weight for individual crustaceans was <0.1 g, lower than the resolution of the
bench scale.
Accumulated material provided additional information on cephalopod and fi sh species
that were taken by the penguins but not necessarily present or identifi able in the food load.
Otoliths were identifi ed using Hecht (1987) and Lalas (1983), differentiated for taxon and
sorted into pairs to determine number of individual prey items. For one fi sh species allometric
equations were available in the literature (red cod, Fea et al. 1999). To determine standard fi sh
length and weight using this equation, red cod otoliths dimensions were measured digitally.
For that high resolution photos of otolith pairs were taken together with a calibration ruler
and then analysed in custom written digital-dimensioning software that determines maximal
length and width of each otolith to the nearest 0.01 mm (L.-G. Ellenberg and T. Mattern,
unpublished data). Using the software had the advantage that all otoliths of a sample could
be measured simultaneously in a standardized way which reduced workload and chance of
measuring errors. Squid beaks were identifi ed, sorted and measurements required for size
and weight estimations were made following Lu & Ickeringill (2002). Average weights and
lengths are given as mean±standard deviation unless indicated otherwise.
3.3 Results
3.3.1 Food load volume and importance of prey classes
The nine males returning from long-term trips all returned with little material in their
stomachs. One penguin regurgitated nothing but bile. The food load masses of the other eight
birds averaged at 20.9±20.1 g. The pooled weight of these samples amounted to 188.5 g only
60 g of which represented identifi able material (Table 3.1). Three samples contained some
crustaceans in different states of digestion that contributed only 0.6 g (<1%) to the pooled wet
weight. Fleshy remains of cephalopod were present in fi ve vomits (5.4 g or 9% of pooled food
3. Diet of breeding Snares penguins 33
loads). Fish remains were found in fi ve males and made up the largest part of the identifi able
food load (55.2 g or 92%).
Two females fl ushed on 2 November 2002 (i.e. during the chick guard stage) had only
digested and unidentifi able material in their stomachs which suggests that they were not
returning to feed chicks. Food load masses obtained from the remaining 13 females averaged
82.4±46.3 g (Table 3.1). Crustaceans were present in all 13 food loads, and cephalopod and
fi sh remains were found in 12 of these. For the females, crustaceans were the most important
prey class (301.2 g or 55% of pooled food loads) followed by fi sh (131.4 g or 24%) and squid
(115.0 g or 21%, Table 3.1). One sample contained two small salps.
3.3.2 Number of species
From food loads and accumulated material, a total of 24 different species from 23 families
were identifi ed (Table 3.2). Nyctiphanes australis was by far the most frequently occurring
crustacean species and was present in 13 of the 15 sampled females. Equally dominant in the
cephalopod class was arrow squid Nototodarus sloani which was present as either identifi able
fresh remains (1 sample) or beaks (17 samples) in the 24 samples. Other cephalopods species
that frequently occurred were warty squid Morotheutis ingens (beaks in 7 samples) and violet
squid Histioteuthis atlantica (beaks in 5 samples). The most frequently occurring fi sh species
was the benthic long-snouted pipefi sh Leptonotus norae which was present in the food load of
six males and seven females but also was found in the accumulated materials of an additional
three females. The pelagic redbait Emmelichthys nitidus was present as an intact specimen in
the food load of one female but was also found in the accumulated material of another seven
females. Red cod Pseudophycis bacchus was only found in the accumulated materials of one
male and six females, while Lanternfi sh Electrona sp. was only present in the fresh fraction of
six samples.
3.3.3 Importance of species in food load
Figure 3.1 summarises the importance of the different prey species in the identifi able
3. Diet of breeding Snares penguins 34
portion of the pooled food loads. In the males, pipefi sh L. norae seems to be the dominant
prey. However, this is primarily due to the fact that one male had a total of 74 pipefi sh in
its stomach – most of which were fully intact indicating that the bird ingested them shortly
before coming ashore –, and contributed 46.9 g of the 51.2 g of pipefi sh remains in the pooled
food loads.
In the females, N. australis contributed the largest portion of the identifi able material in the
stomach samples (301 g or 54.9% of 547.6 g of pooled food loads) and was found in 13
females. Redbait constituted 38.9 g (7.1%) to the identifi able food load. However, this was
the wet weight of one single intact specimen from one female which represented the only
fresh, identifi able remains of redbait found in all samples. Pipefi sh contributed considerably
less to the fresh fraction of the samples (18.1 g or 3.3%) but was present in seven females.
Identifi able remains of squid were scarce with arrow squid and Brachioteuthis sp. only
representing 1.9% (10.6 g) of the food load (Fig. 3.1). However, wet weight of unidentifi able
squid remains that were found in nine females added up to 57.3 g or 10.5% of the pooled food
loads. Similarly, unidentifi ed fi sh remains recovered from 10 different females weighed in at
40.3 g (7.4%).
3.3.4 Importance of species in accumulated material
The accumulated materials from all 24 samples are dominated by two species (Fig. 3.2).
Overall, 498 otolith pairs or singles without counterpart were extracted from all samples; 118
of these were too eroded to be identifi ed. Otolith pairs of red cod amounted to a total of 228
individuals in seven different samples and number of otolith pairs per sample ranged from 2
to 135 individuals (median: 19 individuals). Overall, red cod otoliths represented 60% of all
identifi ed otolith pairs. Although pipefi sh otoliths were not as numerous as red cod, redbait or
thornfi sh ear bones, they were nevertheless present in six birds, three of which did not feature
fresh pipefi sh in their food loads (Fig. 3.2).
A total of 617 individual squid beaks were isolated from 18 of the 24 stomach samples.
Similar to red cod, arrow squid beaks were the most numerous hard-part remains (Fig. 3.2).
3. Diet of breeding Snares penguins 35
A total of 293 arrow squid beaks were extracted which represented 64.8% of all identifi able
squid beaks. 164 beaks (26.6% of all beaks) were too eroded for identifi cation.
3.3.5 Size of prey species
Measured and estimated sizes and weights of the principal prey items are listed in Table 3.3.
The mean mantle lengths of squid indicated that primarily juvenile stages were being taken
by Snares penguins. The estimated sizes of red cod (mean standard length: 33.4±12.3 mm)
represented the pelagic small larval and post-larval stages of this species. The sizes
determined for redbait, thornfi sh and pipefi sh also indicated that primarily juvenile stages of
these species were targeted by the penguins.
3.4 Discussion
3.4.1 Food composition and importance of prey species
The main component of food brought ashore by Snares penguins was a single species of krill,
Nyctiphanes australis. This is consistent with what had been suggested by Warham (1974) and
the overview given in Cooper et al. (1990). Considering the dominance of this species in the
Snares penguins’ diet, the biomass of N. australis occurring in the waters around the Snares
must be considerable. Dense swarms of this species are indeed a common occurrence even
close inshore to the island and Snares penguins are not the only seabirds that benefi t from
krill swarms (Fenwick 1978). N. australis is an important food component of several seabird
species breeding on the Snares such as sooty shearwater Puffi nus griseus (Cruz et al. 2001)
or red-billed gulls Larus novaehollandiea (Jillett 2003). At the Snares it has been observed
that different seabird species prey simultaneously on single Nyctiphanes swarms that are
transported to the surface by upwelling turbulences over submerged rocks and reefs around
the Snares (Fenwick 1978). This phenomenon, however, raises the question whether penguins
only benefi t from tight euphausids assemblages close inshore or whether the planktonic
crustaceans are also targeted when the penguins forage further away from the island. Krill
3. Diet of breeding Snares penguins 36
is generally digested more quickly than bigger prey items such as squid and fi sh (Jackson &
Ryan 1986) and, therefore, it seems likely that the large amounts of N. australis found in the
food loads were taken closer to the island. The intake of large amounts of krill shortly before
landing will mask any evidence of krill intake further offshore (e.g. more digested material).
However, in an offshore environment zooplankton consumes a substantial portion of the
primary production and as such represents a fundamental element in the food web (e.g. Walsh
& McRoy 1986, Bradford-Grieve et al. 2003). The abundance and distribution of zooplankton
is strongly infl uenced by and often associated with permanent hydrographical features (e.g.
fronts), so that there is a correlation between such features and the foraging behaviour of top-
level predators like seabirds (e.g. Hunt 1990, Schneider 1990). Considering that the Snares
are in fact located in the productive subtropical convergence zone and in the vicinity of the
Subtropical Front (Murphy et al. 2001), makes it likely that the offshore foraging behaviour of
Snares penguins is indeed oriented towards zooplankton assemblages and associated fi sh and
squid species.
In comparison to crustaceans, squid and fi sh seem to play only a minor role for the nutrition
the chicks and comprised a relatively small portion in the food loads (21% and 24% of
total identifi able wet weight, Table 1). In some samples, squid beaks were numerous and
represented 78 individuals in one case which might indicate that squid is a more important
food for the adults. Squid beaks may remain intact in seabird stomachs for up to 30-50 days,
(Furness et al. 1984, Jackson & Ryan 1986) which makes it diffi cult to quantify a penguin’s
squid intake per foraging trip and, thus, to assess of importance of cephalopods in the Snares
penguins diet. In terms of species composition, clearly the most important cephalopod in
the Snares penguins’ diet is the southern arrow squid Nototodarus sloani. This cephalopod
is particularly numerous in the area of the subtropical convergence (Smith et al. 1987) and,
therefore, can be assumed to be abundant in waters around the Snares. It is targeted by a wide
range of pelagic vertebrates (e.g. penguins, van Heezik 1990a; albatross, James & Stahl 2000;
fur seals, Harcourt et al. 2002) and is also an important species for commercial fi sheries.
Apart from areas off the east coast of New Zealand’s South Island and around the subantarctic
Auckland Island shelf, squid fi sheries are active along the subtropical convergence zone
3. Diet of breeding Snares penguins 37
(Smith et al. 1987, NABIS 2006). This suggests that there is some potential for interactions
between this fi shery and Snares penguins.
Like squid, fi sh also comprised a relatively low portion of the food loads and, thus, represents
a minor component in the food for the chicks. In seabirds, small fi shes like pilchards or
lantern fi shes are fully digested within a day or less (Furness et al 1984, Jackson & Ryan
1986). Fish taken by Snares penguins generally are post-larval juveniles that are small
(Table 3.3) and it can be assumed that digestion rates similar to those reported for other
seabirds apply for penguins too. Considering that the female Snares penguins during chick-
guard on average stay at sea for 30-40 hours (Chapter 4), fi sh prey caught during the fi rst
half of the trip is likely to be mostly digested by the time the penguins return to the island
and, hence, only evident in the accumulated materials. Judging from the presence of at times
large quantities of intact otoliths in the samples, fi sh seems to play a more important part for
the adults during chick-guard. The most commonly found otoliths in the food loads were
red cod and redbait. Redbait is a planktivorous mid-water species common in New Zealand
waters (Paulin et al. 1989) and as such likely to be associated with prey patches that Snares
penguins might exploit when targeting euphausiids. Although red cod is primarily a bottom
dwelling species, its larval and juvenile stages are pelagic and occur in open waters over the
continental shelf (Habib 1973). Commonly found in the fresh fraction of the samples were
long-snouted pipefi sh and thornfi sh which were sometimes recovered in an intact or little
digested state suggesting that the penguins must have taken them only shortly before landing.
Both species are typical benthic fi sh that primarily occur on rocky habitat (Paulin et al. 1989)
and, therefore, must have been taken along the Snares coast.
3.4.2 Stomach content weights
The food loads recovered from the Snares penguins (mean weight: males – 20.9±20.1 g;
females – 82.4±46.3 g) was relatively low when compared to what has been reported for
the closely related Fiordland penguins Eudyptes pachyrhynchus during the post guard stage
(mean weights: 348±330 g, van Heezik 1989). However, the stage of breeding has a great
3. Diet of breeding Snares penguins 38
infl uence on size of food load in Rockhopper and Royal penguins breeding on Macquarie
Island (Hindell 1988a, Hindell 1988b), with food mass brought to the chicks being highest
during the late chick-guard and the early post-guard stages of breeding. During this study,
samples were mostly taken during early chick-guard (peak hatching period 2002: 27 October
– 3 November; see Chapter 4) and the weight of the food loads are comparable to weight
ranges given for stomach contents of Rockhopper penguins at the same stage (Hindell 1988a).
At this stage, chicks are still small and require lower quantities of food than later on which
allows female Snares penguins to invest primarily in the replenishment of own body reserves
used up during the long fast of the incubation period.
3.4.3 Comparison with other crested penguins in New Zealand
The Snares penguin is one of four crested penguin species in New Zealand. Like the
Rockhopper penguin Eudyptes chrysocome and the Erect-crested penguin E. sclateri, the
Snares penguin breeds only on offshore islands in New Zealand’s subantarctic (Davis &
Renner 2003). The Fiordland penguin, on the other hand, breeds along the New Zealand
mainland and Stewart Island but is nevertheless considered an offshore forager (Croxall
& Davis 1999). Detailed studies on diet composition have so far only been conducted in
Fiordland penguins (van Heezik 1989, van Heezik 1990a).
The main prey species reported for Fiordland penguins are remarkably similar to what has
been found in this study, although the importance of the prey classes differs (van Heezik
1989). Cephalopods (mainly N. sloani and M. ingens) made up most of the Fiordland
penguins diet (85% of the reconstituted food mass), followed by crustaceans (13%, primarily
N. australis) and fi sh (2%). However, it should be noted that the prey class proportions
derived primarily from size estimations from accumulated material (i.e. squid beaks and
otoliths) rather than the fresh fraction, so that the importance of cephalopods might be
exaggerated (van Heezik 1989). Just as in Snares penguins, the sizes of the Fiordland
penguins’ cephalopod and fi sh prey also represented primarily juvenile and larval stages
which might suggest association of penguin prey with zooplankton occurrences.
3. Diet of breeding Snares penguins 39
There is no information about the food of Erect-crested penguins other than that the penguins
feed on crustaceans and cephalopods (Marchant & Higgins 1990). Rockhopper penguins
breeding on New Zealand’s Campbell Island seem to have a surprisingly different diet than
the predominantly planktivorous Snares penguins and crested penguins in general (see Cooper
et al. 1990). According to information given in Merchant & Higgins (1990), Rockhopper
penguins from Campbell Island predominantly foraged for fi sh (91% of all prey items), while
the number of crustaceans (~8%) and cephalopods (~2%) were low. It has been argued that
this “unusual” preference for fi sh is probably a result of a dietary shift due to changes in
the marine environment (Cunningham & Moors 1994). However, stable isotope analysis of
Rockhopper penguin feathers found no evidence for a diet change and instead showed that an
overall reduction of the penguin’s prey led to massive declines of Rockhopper penguins on
Campbell Island in the last 50 years (Thompson & Sagar 2002).
The Campbell Island Rockhopper penguins’ preference for fi sh stands in contrast with the
Snares penguins’ preference for krill. However, these dietary differences refl ect the disparity
of the marine environments each species forage in. Campbell Island lies centrally in cool,
subantarctic waters that are known to be iron-limited and, thus, show low phytoplankton
production (Boyd et al. 1999) which also limits the biomass of zooplankton, particularly
grazing species such as krill (e.g. Ward et al. 2005). As a result of this, Rockhopper penguins
are part of a long food web and fi sh seem to be the most available penguin prey at Campbell
Island (Bradford-Grieve et al. 2003). Fish stocks are greatly affected by sea surface
temperatures with increasing temperatures often resulting in population declines (e.g. Noto
and Yasuda 1999, Kawasaki 2002). In the light of increasing sea surface temperatures around
Campbell Island (Cunningham and Moors 1990), a decimation of local fi sh populations and
connected to that Rockhopper penguins seem likely.
Contrasting this, the Snares are located in warm and nutrient-rich waters north of the
Subtropical Front that feature high phytoplankton concentrations (Murphy et al. 2001) and,
thus, support high zooplankton biomass which is directly available to top-level predators like
seabirds (Jillett 2003). The high abundance of crustacean swarms and associated fi sh and
squid even within close range of the Snares underlines this fact (Fenwick 1978). Therefore, it
3. Diet of breeding Snares penguins 40
seems likely that the secret to the Snares penguins’ success is to a large degree a result of the
location of the species breeding place. Therefore, it seems likely that the secret to the Snares
penguins’ success is rooted in oceanic productivity that is considerably higher in Tasman Sea
waters when compared to the subantarctic waters around Campbell Island.
3. Diet of breeding Snares penguins 41
Table 3.1. Frequency of occurrence of different prey classes in stomach samples, food
loads and weight proportion of prey classes on food loads of nine male (incubation, 21-24
October 2002) and 15 female Snares penguins (early chick-guard, 04-08 November 2002).
Males Females
N = 9 N = 15
Frequency of occurrence (n)
Crustaceans 313
Cephalopods 812
Salps -1
Fish 612
Mean food load (g) 20.9±20.1 82.4±46.3
range 0 - 56.8 17.4 - 163.0
Pooled food load (g) 188.5 1235.5
Identifi able portion of pooled food load (g) 60.0 547.6
Proportion in identifi able food load
Crustaceans <1 % 55 %
Cephalopods 9 % 21 %
Salps - <1 %
Fish 92 % 24 %
3. Diet of breeding Snares penguins 42
Table 3.2. Frequency of occurrence of prey species of Snares penguins during late incubation and early chick-guard (21 October – 08 November 2002).
Species are grouped in main prey classes and ordered according to frequency. Occurrence of species was determined from fresh (all prey classes) and
accumulated material (cephalopod beaks and fi sh otoliths) in a total of 24 stomach samples, three of these contained only digested and unidentifi able material.
Prey Class
Common name
Species Family Males
(N = 9)
Females
(N = 15)
Total
(N = 24)
n% n% n%
Crustaceans Nyctiphanes australis Euphausiacea 2 22.2 13 86.7 15 62.5
Euphausia sp. Euphausiacea 1 11.1 2 13.3 3 12.5
unidentifi able Decapoda 1 11.1 1 6.7 2 8.3
Cephalopods
Arrow squid Nototodarus sloani Ommastrephidae 8 88.9 10 66.7 18 75.0
Warty squid Morotheutis ingens Onychoteuthidae 2 22.2 5 33.3 7 29.2
Violet squid Histioteuthis atlantica Histioteuthidae 1 11.1 4 26.7 5 20.8
Pelagic octopus ?Ocythoe tuberculata Ocythoidae 2 22.2 2 13.3 4 16.7
Brachioteuthis sp. Brachioteuthidae - - 3 20.0 3 12.5
?Enoploteuthis galaxias Enoploteuthidae - - 1 6.7 1 4.2
Mastigotheutis sp. Mastigoteuthidae - - 1 6.7 1 4.2
Taonius sp. Cranchidae - - 1 6.7 1 4.2
Salps
unidentifi able Salpidae - - 1 6.7 1 4.2
Fish
Long-snouted pipefi sh Leptonotus norae Syngnathidae 6 66.7 9 60.0 15 62.5
Redbait Emmelichthys nitidus Emmelichthyidae - - 8 53.3 8 33.3
Red cod Pseudophycis bacchus Moridae 1 11.1 6 40.0 7 29.2
Lanternfi sh Electrona sp. Myctophidae 1 11.1 5 33.3 6 25.0
Thornfi sh Bovichtus psychorolutes Bovichthyidae 1 11.1 4 26.7 5 20.8
Blue warehou Sariolella brama Centrolophidae - - 3 20.0 3 12.5
Rock fi sh Acanthoclinus sp. Acanthoclinidae - - 1 6.7 1 4.2
Silverside Argentina elongata Argentinidae 1 11.1 - - 1 4.2
Conger eel Conger sp. Congridae - - 1 6.7 1 4.2
Opalfi sh Hemerocietes pauciradiatus Percophidae - - 1 6.7 1 4.2
Blue moki Latridopsis ciliaris Latrididae - - 1 6.7 1 4.2
Hoki Macruronus novaezelandiae Merlucciidae - - 1 6.7 1 4.2
3. Diet of breeding Snares penguins 43
Table 3.3. Measured and estimated sizes and weights of principal prey species of Snares penguins (N = 24).
a random sub sample from pooled items b equation for Nototodarus gouldi c no otoliths equation available, one fresh specimen
UHL upper hood length OL Otolith length, size estimation from allometric equation F intact specimen
Measurements
(mm)
Measured/estimated
body length (mm)
Measured/estimated
body mass (g)
n mean±sd range mean±sd range mean±sd range
Crustaceans
Nyctiphanes australisF250a- - 12.9±2.6 8-20 <0.1 -
Cephalopods
Arrow squidUHL,b 100a2.3±1.6 0.6-10.8 53.2±19.1 32.6-157.6 no adequate equation
Warty squidUHL 49 2.8±1.5 0.6-6.1 no adequate equation
Violet squidUHL 25 7.0±2.4 2.8-10.6 46.4±18.7 13.5-81.1 31.5±23.6 2.7-90.3
Fish
Red codOL 228 2.4±0.5 1.7-6.7 33.4±12.3 19.6-152.9 0.6±2.9 0.2-32.6
RedbaitF,c 1 - - 169.0 - 38.9 -
Thornfi shF16 - - 36.3±1.3 34.0-38.0 0.4±0.2 0.2-0.6
Long-snouted pipefi shF68 - - 93.4±21.9 50-135 0.4±0.1 0.3-0.9
3. Diet of breeding Snares penguins 44
Figure 3.1. Composition of prey species in pooled food loads (i.e. fresh fraction of stomach
samples) from male (N = 9) and females (N = 15) Snares penguins during late incubation and early
chick-guard 2002. Total mass of pooled food loads pooled was 188.5 g for males and 1235.5 g for
females. Material too digested to allow identifi cation of at least prey class was excluded from graph
compilation.
0 102030405060
N. australis
unident. squid
Pipef ish
unident. fish
Redbait
Thornfish
Lante rnf ish
Arr o w qu i d
Pelagic octopus
Brachiotheutis sp.
Conger eel
Rockfish
Euphausia sp.
Decapod
wet weight (g)
males
females
300
N=9
pooled, identifiable food load: 60.0 g
N=15
pooled, identifiable food load: 547.6 g
3. Diet of breeding Snares penguins 45
Figure 3.2. Composition of accumulated prey items (i.e. otoliths/squid beaks) recovered from Snares
penguin stomachs (N = 24) during late incubation and early chick-guard 2002.
020406080100120
Red cod
Redbait
Thornfish
Pipefish
Moki
Wharehou
Hoki
Opalfish
unidentifiable
220
number of prey items (n)
FISH
0 20 40 60 80 100 120 140 160
Arrow squid
Warty squid
Violet squid
Brachiotheutis sp.
?Enoplotheutis galaxias
Mastigotheutis sp.
Taonius sp.
unidentifiable
280 300
number of prey items (n)
CEPHALOPODS
4. Infl uence of Oceanography and Seasonality on Snares penguins 46
CHAPTER 4
INFLUENCE OF OCEANOGRAPHY AND SEASONALITY
ON FORAGING BEHAVIOUR AND NESTING PATTERNS OF
SNARES PENGUINS EUDYPTES ROBUSTUS DURING THE
INCUBATION STAGE
4. Infl uence of Oceanography and Seasonality on Snares penguins 47
4.1 Introduction
The distribution of life in the world’s oceans is a product of dynamic physical processes.
Currents and oceanographic features such as fronts play the most important part in the
distribution of nutrients and, linked to that, primary production (Chang & Gall 1998).
Phytoplankton is an essential determinant of the abundance of zooplankton and, therefore,
higher trophic levels of the food web (Bradford-Grieve et al. 2003). Besides oceanographical
factors, seasonality strongly infl uences primary production, which is particularly important at
higher latitudes. Primary production, e.g. accumulation of phytoplankton biomass, depends on
solar energy, which is reduced during the winter months and, as a result, leads to fl uctuations
of the phytoplankton concentration with the seasons (e.g. Murphy et al. 2001). Phytoplankton
is the basis of the marine food web so that the seasonality affecting the primary production is
bound to have an effect on the consumers at higher trophic levels.
Penguins are important consumers in the marine environment (Croxall & Lishman 1987). The
foraging ecology of penguins often refl ects the environmental conditions that determine the
abundance and distribution of food. This is, for example, particularly apparent in primarily
planktivorous species like most crested penguins (Eudyptes spp.). The annual breeding cycles
of most crested penguins are highly synchronised and coincide with the seasonal increase in
food availability and day length (Williams 1995). At the same time, the distribution of the
penguins’ prey is often linked to the presence of oceanic fronts, making these an attractive,
spatially predictable source of food for penguins (e.g. Hull et al. 1997, Tremblay & Cherel
2003).
The Snares penguin Eudyptes robustus is endemic to the small Snares island group some
200 km south of the New Zealand mainland. As with other crested penguins, the timing of
breeding is similar between years (Warham 1975, Williams 1995). The nesting patterns of
Snares penguins are well structured and predictable, especially during the incubation stage.
Egg-laying in late September is followed by a two week period during which both partners
stay at the nest. In mid-October, the breeding males leave the island (the ‘exodus’) for a long
(ca. 2 weeks, Warham 1974) foraging trip. Upon the males’ return in late October, the
4. Infl uence of Oceanography and Seasonality on Snares penguins 48
females leave the nest to forage and return after about one week at sea, when chicks hatch
in early November. The key events associated with incubation seem to occur over relatively
short timeframes (e.g. all eggs hatch within a fi ve day period; Warham 1974), which is
indicative of a high level of synchrony. While the synchrony might be enhanced by social
interactions (e.g. Fishman & Stone 2006), environmental factors such as food supply are
likely to play an equal if not more important role (Davis & Renner 2003).
The incubation stage (September-November) coincides with the onset of the spring bloom
of phytoplankton in New Zealand waters (Murphy et al. 2001). One of the main components
of the Snares penguins’ diet is the euphausiid Nyctiphanes australis (see Chapter 3) which
feeds predominantly on phytoplankton (Ritz et al. 1990) and, thus, provides a direct link
between the Snares penguins’ foraging and marine primary production. However, during the
incubation stage (late September & early October), the spring bloom is just beginning so that
oceanic productivity closer to the Snares is still low (Murphy et al. 2001). Accordingly it is
predicted that the penguins have to forage further away in marine areas where the bloom is
already advanced. One such area and an important oceanographic feature in relative proximity
of the Snares is the Subtropical Front (STF). At the front, warm, saline subtropical Central
Tasman Waters (CTW) and cool, less saline subantarctic waters (SAW) converge (Heath
1985, see Fig. 4.1). Such convergence zones create mixing processes that have been found to
accumulate planktonic prey of seabirds (Schneider 1990).
The purpose of this study was to examine the interactions between Snares penguins and the
oceanic environment and their infl uence on nesting patterns. For this I examined the penguins’
foraging ranges and diving behaviour with GPS dive loggers, and at the same time recorded
nest attendance patterns. I made the following predictions: (1) the synchrony in the departure
of male penguins in mid-October correlates with the onset of the spring phytoplankton
bloom, (2) the male penguins travel east towards the Subtropical Front where phytoplankton
concentrations can be expected to be highest at the time, and (3) before the chicks hatch the
female penguins benefi t from the advanced spring bloom which facilitates shorter foraging
trip times and, hence, foraging ranges.
4. Infl uence of Oceanography and Seasonality on Snares penguins 49
4.2 Methods
4.2.1 Timing of fi eld work and study site
Nest attendance patterns and foraging behaviour of incubating Snares penguins were studied
during the breeding seasons 2002 (no foraging data, see Chapter 2), 2003 and 2004 on the
main island of the Snares group, North-East Island. All observations and logger deployments
occurred in the second-largest Snares penguin colony (A3) that comprises about 1200
breeding pairs (Amey et al. 2001). This colony was chosen not only for ease of accessibility
but also because some basic information on nest attendance patterns for the same colony was
available in Warham (1974).
In 2002 and 2003, the research team arrived on the Snares on 6 October and 9 October,
respectively, which permitted timing the onset and end of the exodus of breeding males in
colony A3. In 2004, fi eld work commenced on 17 October, after males had already left the
island. Duration of expeditions ranged from fi ve weeks in 2002 to six weeks in 2003 and
2004.
4.2.2 Nest attendance patterns
To determine nest attendance patterns, observation plots in colony A3 were established. In
2002, data were obtained for one plot comprising 42 nests, while in the following two years,
three observation plots were established in different areas of the colony encompassing 40
to 60 nests per plot. In total, 154 nests were monitored in 2003, and 166 nests in 2004. For
complete nest attendance pattern analysis only nest that successfully hatched chicks were
considered (2002: 28 nests, 2003: 126 nests; 2004: 109 nests). Each observation plot was
assigned to a single observer who conducted two- to six-hour long observations every day
(usually between 1100 and 1700 hours). During observations, nest status and sex of attending
adults were recorded. Additionally, nest relief times of breeding adults as well as other
behavioural activities in the plot were noted. Marking penguins was not permitted so that
4. Infl uence of Oceanography and Seasonality on Snares penguins 50
identifi cation of adults had to be achieved by passive means only. Male Snares penguins have
a markedly heavier bill than females (Warham 1974) so that the sex of each mate in a pair
could be determined visually when both partners were together at the nest. Individual features,
such as melanisms or black splotches on a penguin’s breast, overall body condition or scars
were also used to identify birds. The return of a male to its nest after foraging was easy to spot
as males were considerably fatter than the incubating females and soon produced numerous
scat marks radiating from the nest bowl. Females generally return during daylight hours, and
especially after their long incubation stage foraging trips generally spend the rest of the day
at several hours on their nests – even if their eggs had not yet hatched (Warham 1974, own
observations). Thus, the likelihood of females returning and departing during the absence of
the observer was minimal.
4.2.3 GPS loggers and Time Depth Recorders (TDR)
Foraging and diving behaviour were studied in 2003 and 2004, using two types of data
loggers. GPS data loggers (GPS-TDlog, earth&OCEAN Technologies, Kiel, Germany;
dimensions: L100xW48xH24 mm, mass: ~70 g) record information of the penguins’ at-sea
movements and dive behaviour. The devices contain a GPS receiving unit that determines
accurate geographical position (position error <10 m) from signals from orbiting satellites
of the Global Positioning System (U.S. Department of Defence, USA) (see Chapter 2). The
device also contains environmental sensors that recorded dive depth (resolution: ~0.1 m)
and ambient water temperature (resolution: ~0.005°K). However, during all deployments,
the temperature sensors failed and recorded no data. GPS and sensor data were stored with a
precise date and timestamp in the device’s internal non-volatile fl ash memory and had to be
downloaded to a computer after device recovery. The loggers’ sampling regime was freely
programmable and was set up to record a GPS position after each dive, and to store sensor
readings at 5 s intervals. Since acquisition of a GPS fi x takes between 25 and 30 seconds (see
Chapter 2), no position could be recorded when a penguin stayed at the surface for shorter
intervals between dives.
4. Infl uence of Oceanography and Seasonality on Snares penguins 51
Time Depth Recorders or TDRs (MK9, Wildlife Computers, Redmond, WA, USA;
dimensions: L67xW17xH17mm, mass: ~30 g) were used to study diving behaviour too.
The TDRs contain a pressure transducer to determine dive depth (resolution: 0.5 m) and a
temperature sensor (resolution: 0.05°C). The TDRs are also freely programmable and were set
up to record dive depth and temperature at 5 s intervals, similar to the GPS logger sampling
regime. Additionally, the device features a wet/dry sensor that triggered the TDR to sample
only when wet (i.e. during dives) and stopped sampling, while still keeping track of the time,
when dry (i.e. at the surface). This greatly reduced battery consumption and allowed data to
be recorded for complete long-term foraging trips.
The devices were attached with adhesive tape (Tesa-tape method, Wilson et al. 1997) to the
penguins’ lower backs to reduce drag (Bannasch et al. 1994). All loggers were deployed at
colony A3. The penguins were captured at the nest while their mates were incubating; the
birds were then carefully transferred out of the nesting area. Before logger deployment, the
penguins were weighed to the nearest 50 g in a cloth bag using spring balances. During the
attachment procedure the penguin’s head was covered with a cloth hood to reduce stress. After
successful attachment, the penguin was carried back into the colony and released 5 m from
its nest site. The entire handling time (i.e. capture, measurements, deployment and release)
ranged between 12 to 18 minutes. All loggers were recovered after the penguins’ long-term
foraging trips. Birds were recaptured either at the penguins’ main landing site, which is
located next to the research huts in Station Cove, or when the birds returned to their nest in
the colony.
In 2003, four males were equipped with GPS loggers and three males with TDRs before
they left on long-term trips. The number of deployments was limited by the number of
devices available for the study. This meant that the devices had to be recovered and data
downloaded before they could be re-deployed on females. Three of the males returned too
late for re-deployment, so that only four females could be fi tted with two GPS and two TDRs,
respectively, before they left on long trips. Both GPS logger deployments on females failed
– one bird returned with a water logged device while the other female did not return to the
colony. In 2004, an attempt was made to collect additional GPS logger data on two females
4. Infl uence of Oceanography and Seasonality on Snares penguins 52
before the chicks had hatched, but both deployments failed again, this time due to malfunction
of one device and water logging of the other.
4.2.4 Oceanographic data
The foraging tracks of penguins were analysed with regard to sea surface temperatures
(SST, °C) to help identify water masses, and chlorophyll a concentration (ChlA, mg/m³) as
a measure of primary production. Both parameters were assessed from satellite ocean colour
data recorded by NASA’s Moderate-resolution Imaging Spectroradiometer (MODIS/Aqua)
programme (http://oceancolor.gscf.nasa.gov). Depending on the satellite’s data coverage,
which is affected by cloud cover, either weekly (8-day) or monthly data sets were used.
The data are available as Level-3 Standard Mapped Image (SMI) that give average ChlA
concentration and SST in global, equal-area cells with spatial resolutions of 4x4 km (Barbini
et al. 2005). Water masses were identifi able from SST gradients, with warmer Central Tasman
Water (CTW) having temperatures >10° C, the Subtropical Front (STF) ranging between 9°
and 10° while cool Subantarctic Waters (SAW) were <9° C.
4.2.5 Data analysis
All analysis of GPS, dive and MODIS data was done with custom written software
(T. Mattern, unpublished data). GPS data were used to linearly extrapolate the penguins’
foraging tracks from all recorded fi xes, to calculate travel distance as the sum of linear
distances between consecutive GPS fi xes, and to determine the furthest linear distance
between a position fi x and the island. The tracks were then analysed with regard to
oceanographic parameters that occurred along a penguin’s extrapolated travel route by
determining the relative time a penguin spent within a cell of the MODIS SMI. That way,
daily averages of ChlA concentration and, in the case of birds equipped with GPS loggers,
SST of the sea areas visited, were determined for each bird. The TDRs’ temperature readings
at the surface fl uctuated depending on the weather situation (e.g. sunshine resulted in higher
temperature readings), so that sea surface temperature was determined from temperature data
4. Infl uence of Oceanography and Seasonality on Snares penguins 53
recorded during each dive at depths between 5 and 10 m. The TDR temperature readings
served as indicator for watermass.
From dive data, basic parameters were determined such as dive time, dive depth and duration
of post-dive interval the penguins spent at the surface. Three different dive phases were
distinguishable: the descent, the bottom phase and the ascent (Wilson 1995). The start of the
bottom phase was defi ned as the time when the penguin’s descent rate (i.e. vertical velocity)
became less than 0.2 m/s after a continuous descent. Accordingly, the end of the bottom phase
was marked by an increase in vertical velocity >0.2 m/s followed by a continuous ascent to
the surface. The time between start and end of the bottom phase was defi ned the bottom time.
Descent and ascent rates were calculated from the depth change and transit time during the
ascent or descent phase.
Each dive was analysed with regard to the depth reached during the preceding dive.
Consecutive dives reaching similar depths were defi ned as repeated maximum depth (RMD)
dives. Similarity of depths was accepted when a dive’s maximum depth was within 10%
of the previous dive’s maximum depth (following Tremblay & Cherel 2000). Under the
assumption that after a penguin that has located a prey patch at a specifi c depth it returns to
a similar depth during consecutive dives (i.e. dive bout, cf. Wilson 1995), RMD dives are
more likely to represent feeding rather than searching behaviour. During the analysis of the
dive data it was found that Snares penguins often performed travelling dives, i.e. dives with a
u-shaped profi le without any signifi cant vertical undulations while at the same time covering
large distances in a short time. On such dives the birds sometimes reached depths of up to
20m. To fi lter out travelling behaviour, only dives deeper than 20 m were included in the
RMD analysis.
Additionally, diving effi ciency (bottom time/[dive time + post-dive interval]; Ydenberg &
Clark 1989) and diving effort (dive time/[dive time + post-dive interval]) were calculated
for each dive. Dive events could only be identifi ed when the penguins dived deeper than
3 m. Furthermore, only dives lasting 20 s or more (i.e. with a minimum of four data points)
were accepted for analysis. This was because the pressure transducer in the TDRs showed
4. Infl uence of Oceanography and Seasonality on Snares penguins 54
considerable fl uctuations close to the surface. For comparative reasons, GPS logger data were
treated similarly, although surface fl uctuations did not occur.
All statistical analysis was carried out in Minitab 14 (Minitab Inc, State College, PA, USA).
Data were tested for normality using the Kolmogorov-Smirnov test. Comparisons were made
using two-tailed t-test and one-way ANOVA followed by Tukey’s post-hoc comparison.
Comparisons between groups of birds (e.g. males vs. females) were made using individual
means for each bird to avoid pseudo replication. Statistical signifi cance was accepted at the
α < 0.05 level. Averages are given as mean ± standard deviation unless indicated otherwise.
4.3 Results
4.3.1 Nest attendance patterns and foraging trip durations
The departure of breeding males on their long post-laying foraging trip was a highly
synchronous event. The exodus was heralded in colony A3 by the departure of the fi rst
individuals on 11 October 2002 and 10 October 2003, respectively (Fig. 4.2). In both years,
the vast majority of breeding males left the colony within a four day time window (11-15
October) that was identical in both years. Quite possibly, the exodus in 2004 occurred in a
similar time frame as – with one exception – no breeding males were present in the colony
by the time nest monitoring started on 17 October, and return dates of males that year were
similar to those of 2002 and 2003. The foraging trip lengths determined from nest attendance
observations in 2002 (mean trip length: 11± 2 d, range: 8-17 d; n = 29) and 2003 (11±2 d,
range: 6-21 d, n = 126) did not differ (t152 = 0.819; p = 0.414). The return of the males was
somewhat less synchronized than departure. Nevertheless, 75% of the males arrived back at
their nests within four days between 22 and 26 October in all three seasons (Fig. 4.2).
The departure of the females was closely related to the return of their mates (Fig. 4.2).
Usually a female would leave the nest shortly after her mate’s return unless the male arrived
4. Infl uence of Oceanography and Seasonality on Snares penguins 55
back in the colony in the late afternoon. In this case, females generally departed early the
next morning. Median departure dates of females ranged between 24 and 25 October in all
three years. Foraging trip durations of the females varied between years. Mean trip durations
were comparable in 2002 (5.5±2.1 d, range: 1-9 d) and 2003 (6.1±2.4 d, range: 1-11 d) but
were signifi cantly longer in 2004 (8.2±2.7 d, range: 1-16 d; ANOVA with Tukey’s post-hoc:
F1,2 = 23.98, p < 0.001). Despite these differences, the return of the females coincided with
egg hatching in all years (Fig. 4.2). Median return and hatching date in 2002 and 2003 was 30
October. In 2004, the median date of return was 01 November, the median hatch date was a
day later (02 November).
4.3.2 At-sea movements
The deployments of four GPS loggers on male Snares penguins on long-term trips during
incubation in 2003 resulted in GPS/dive data sets for three of them; the fourth male did not
leave its nest for fi ve days and the device was recovered without any at-sea data. The three
other birds left the island between the 15 and 16 October, respectively (Table 4.1). The data
recorded before the loggers’ batteries were exhausted encompassed ca. three days (mean
operation time: 2.2±0.2 d). This relates to one-third to one-fi fth of the complete trip durations
(mean trip duration: 12.0±4.4 d). During the loggers’ operation time, the penguins foraged
an average 158.2± 59.7 km away from the Snares and covered distances of up to 226.3 km
(Table 4.1). All three penguins travelled due east from the island (Fig. 4.3 map). Two of the
birds crossed into the deeper waters beyond the shelf edge of the Snares Rise in the evening of
their second day at sea. The third male changed its easterly course while still in the shallower
waters (<200 m). The horizontal speeds of all three birds decreased while dive depth increased
as the trip duration progressed and was lowest during the last day of logger operation (Fig.
4.3a&c). During their third day at sea, two birds foraged in waters of the subtropical front
(STF, sea surface water temperature < 10°C, Fig. 4.3b, compare with Fig. 4.4 map) that
featured relatively high productivity (ChlA concentration ~0.3 mg/m³, Fig. 4.3d). The third
penguin’s course change also coincided with a patch of high primary production similar to
the conditions at the front but was still some 80 km short of the cooler waters of the STF
4. Infl uence of Oceanography and Seasonality on Snares penguins 56
(Fig. 4.3 map, b&d).
The deployments of three time depth recorders (TDR) on male Snares penguins produced
sensor data for complete foraging trips. The three birds all left on the same day (14.10.2003)
and stayed at sea for 8.8 to 13.4 days (mean trip duration: 11.3±2.5 d). While the TDRs did
not record any spatial information, the temperature data give some indications about the birds’
general movements at sea. All three penguins foraged in waters >10°C during their fi rst two
to three days at sea (Fig. 4.4, top graph). The temperature profi les of two birds then dropped
markedly to surface temperatures of <10°C indicating that the birds entered the cooler waters
of the STF to the east or south of the Snares (Fig. 4.4 map). Both penguins stayed at the front
for most of their time at sea (6.1 of 11.1 days and 4.3 of 8.8 days, respectively) and returned
to the island within three days after leaving the cooler water. The third male equipped with a
TDR foraged for 13.4 days and stayed in waters >10°C all the time. Nevertheless, during its
fi rst few days at sea the bird foraged in cool waters (10-11°C) which shows that it must have
got close to the front (compare Fig. 4.4, top graph, light grey line, with Fig. 4.4 map).
After the return of the three males, two of the TDRs were re-deployed on incubating females
leaving on long foraging trips. Both females left their nests on the 26 and 27 October,
respectively. They both foraged for 4.1 days, a period considerably shorter than that of
the males. During their entire time at sea both females stayed in waters >11°C (Fig. 4.4,
middle graph). Considering the distribution of the isotherms compiled from satellite sea
surface temperature data, the females must have foraged north to north-east of the Snares
(Fig. 4.4 map).
4.3.3 Diving behaviour
Although diving behaviour varied amongst individuals, we found no differences when data
were tested with regard to device type. Basic dive parameters did not differ signifi cantly
(t-test of means during fi rst three days at sea; max depth: GPS – 45.0±11.0 m, n = 3, TDR
– 51.7±4.7 m, n = 3, t5 = -0.39, p = 0.725; dive time: 108.1±18.6 s vs. 125.3±9.7 s, t5 = -1.42,
p = 0.251; diving effi ciency: 0.27±0.04 vs. 0.26±0.01, t5 = 0.80, p = 0.508; diving effort:
4. Infl uence of Oceanography and Seasonality on Snares penguins 57
0.77±0.01 vs. 0.755±0.01, t5 = 1.91, p = 0.151). Therefore, dive data of all six males were
deemed comparable regardless of device type.
The diving behaviour of males refl ected the gradual change from travelling to foraging
behaviour during the fi rst three days at sea, regardless of the birds’ destinations (Fig. 4.3c).
Most dive parameters differed signifi cantly between the fi rst and third day at sea (Table
4.2). Higher transit rates to greater depths combined with a signifi cantly longer bottom time
resulted in slightly higher diving effi ciency during the third day at sea, which further indicates
a shift from travelling to prey searching/feeding behaviour. At the same time, diving effort
and effi ciency remained similar which indicates no signifi cant changes in energy expenditure
between the days.
The daily means of dive parameters determined for three males with TDRs over the entire
duration of their foraging trips showed strong correlations with the sea surface temperature.
The duration of a dive cycle (i.e. dive time, bottom phase and post-dive interval) was shorter
in warmer CTW (Figs 4.5a, c, e). The birds dived deeper at the front (SST <10°C, Fig. 4.5b),
showed a higher frequency of RMD dives (Fig. 4.5 d) but an overall lower diving effort (Fig.
4.5f). The frequencies of dive depths show a marked bimodality for the two penguins foraging
at the STF (Fig. 4.6 left graph, grey bars). 23% of all dives were less than 20 m deep. Half of
the dives at the STF were deeper than 85 m. Although the birds showed similar bimodal depth
frequencies when foraging in CTW, 30% of the dives were shallower than 20 m and only 26%
were deeper than 85 m (Fig. 4.6 left graph, black bars). The dive depths were much more
evenly distributed along the entire depth spectrum in the male that remained in CTW during
its entire foraging trip (Fig. 4.6 middle graph).
In contrast, the two females showed a strong preference for dives in the upper 20 m (54% of
all dives) (Fig. 4.6, right graph). Consequently, dive behaviour of the females differed from
that of males (Table 4.3). Dive times of females were signifi cantly shorter than those of in
males and same was true for bottom times. With the exception of descent and ascend rates that
were similar, most other dive parameters also differed considerably between sexes. Although
statistical signifi cance could not always be confi rmed, presumably as a consequence of small
4. Infl uence of Oceanography and Seasonality on Snares penguins 58
sample sizes, the differences between the sexes nevertheless represent an obvious trend (Table
4.3).
4.4 Discussion
In all three years, Snares penguins exhibited nest attendance patterns that were highly
synchronized. The synchrony of some key events of breeding – primarily the timing of male
exodus and, to a lesser extent, the return of the males and departure of the females – was also
remarkably similar between the years. This interannual synchrony suggests that breeding
patterns are strongly infl uenced by daylength which in turn also determines the onset of the
phytoplankton spring bloom (Murphy 2001) and, thus, the availability of the penguins’ prey.
4.4.1 Synchronous departure of the males
During the fi rst two years, the male exodus ranged around the same median date (13 October).
The departure of the males was highly synchronous and the colony was practically devoid
of breeding males within 5 days. The timing of the exodus also was consistent with historic
records. Warham (1974) reported that in colony A3 the departure of the males “was almost
completed by 15 October 1972”. Nest attendance patterns and the date of return of males in
2004 suggest that the exodus most likely occurred around the same date as in the previous
years (Fig. 4.2). Furthermore, the synchrony of the male exodus does not seem to be restricted
to birds from the same colony as other colonies nearby were also observed to empty out at
the same time with a concomitant increase in numbers of males departing from Station Cove
landing.
The highly synchronous departure of the males has also been reported in Erect-crested
penguins (Eudyptes sclateri). Davis and Renner (2003) found that the male Erect-crested
penguins left the colony within a three day period independently of their egg-laying dates, and
suggest that the synchrony of the males’ exodus might minimize the probability of aggressive
assaults on lone females that could result in nest failure. While social interactions indeed
4. Infl uence of Oceanography and Seasonality on Snares penguins 59
are believed to facilitate breeding synchrony in seabirds (e.g. Fetterolf & Dunham 1985,
Waas 1988, Waas et al. 2000), social stimuli might elucidate intra-colonial synchrony but do
not explain why the males’ exodus seems to occur around a specifi c day of the month (13
October) each year.
The inter-annual – and perhaps inter-colonial – synchrony of the departure of incubating male
Snares penguins correlates strongly with date, which suggests photoperiod as the primary
trigger for the well-timed exodus. Birds are photosensitive and daylength is known to induce
hormonal and subsequently behavioural responses (Nicholls et al. 1988). For penguin species
living in temperate and polar regions, photosensitivity facilitates the synchronisation of
reproduction with seasonal changes in the environment (Cockrem 1995), most importantly
food availability within range of the breeding location (Williams 1995).
4.4.2 Foraging of male penguins and infl uence of oceanography
A likely explanation for the synchronisation of male Snares penguins’ departure with
photoperiod is the onset of the phytoplankton spring bloom in the waters around the Snares in
October each year (see detailed description in Murphy et al. 2001). Increased phytoplankton
biomass often correlate with high zooplankton abundance (Krell et al. 2005) which
subsequently makes phytoplankton-rich areas at sea particularly interesting for planktivorous
top level predators like whales and seabirds (Bradford-Grieve et al. 2003). Considering that
crested penguins feed predominantly on planktonic crustaceans (i.e. euphausiids Williams
1995, Davis & Renner 2003, see Chapter 3) a relationship between chlorophyll a (ChlA)
concentration (as a measure of phytoplankton abundance) and foraging behaviour seems
likely (e.g. Tremblay & Cherel 2003).
Due to subtle interactions of factors limiting primary production in New Zealand’s
subantarctic, phytoplankton biomass is not evenly distributed in the waters around the Snares
(Murphy et al. 2001). Apart from localised occurrences of high ChlA concentrations (Banse &
English 1997), the Subtropical Front (STF, Fig. 4.4) represents a sea region with probably the
most predictable primary production within range of the Snares. The STF is known to feature
4. Infl uence of Oceanography and Seasonality on Snares penguins 60
elevated ChlA concentrations throughout the year (Murphy et al. 2001), which was also the
case during the period of this study (Fig. 4.3). Such properties render the STF a reliable food
source for seabirds in general (compare the “Snares Islands hotspot” in Waugh et al. 2002)
and are likely to make the front a tempting destination for male Snares penguins too.
Indeed, the front was the supposed destination of four of the six males equipped with
data loggers. The GPS data recorded on one male that did not reach the front, shows that
the bird nevertheless travelled towards the front until it reached a patch of elevated ChlA
concentration. Similarly, the temperature readings of one TDR bird that did not enter
waters <10°C suggest that it must have foraged close to the front before orienting back into
warmer Central Tasman Water (CTW, Fig. 4.3). Supporting this, all birds’ dive behaviour
during the fi rst days at sea refl ects primarily travelling behaviour. Dive times and depths
increased considerably in all birds between the fi rst and the third day at sea (Table 4.2) and
indicate a gradual shift from shallower travelling dives to deeper dives consistent with prey
searching behaviour. This suggests signifi cant benefi ts in terms of fi nding food must accrue
to those penguins that travel to the STF. This suggests that the chance of fi nding prey at the
STF justifi es the effort required to travel there. Conversely, it implies that the likelihood of
encountering productive areas within CTW – and, thus, closer to the Snares – was probably
considerably lower than at the STF. In this light, it is interesting that two of the males with
GPS loggers apparently “ignored” an area of high productivity that was of obvious interest for
the third male (Fig. 4.3). A possible explanation for this might be that the fi rst two penguins
passed through this area almost 48 hours earlier than the third male and it is conceivable that
a more favourable prey situation developed in the time between the transit of the fi rst two
penguins and the arrival of the third bird.
The diving behaviour of the male penguins equipped with TDRs was a function of the sea
surface temperature (Fig. 4.5). The penguins dived deeper and longer at the STF than when
they were in CTW. In the cooler waters of the front, a majority of dives occurred in dive bouts
during which the birds returned to similar depths of previous dives (RMD dive frequency
>60% at the front, Fig. 4.5). As a result, the frequencies of dive depths observed in two males
4. Infl uence of Oceanography and Seasonality on Snares penguins 61
foraging at the STF are distinctly bimodal (Fig. 4.6). More than 50% of all dives recorded at
the front were deeper than 85 m, suggesting a primary exploitation of prey patches at greater
depths. Another, less pronounced peak is apparent in the upper 20 m (~30% of all dives, Fig.
4.6) which can be attributed to travelling behaviour and shallow prey searching/feeding.
The bimodality is most likely a result of the downwelling mechanisms that are in play at
the STF which transport nutrients and plankton to greater depths (Nodder & Gall 1998).
This downward transport at the front is restricted by a vertical temperature gradient which
is strongest at depths between 100 and 200m (Morris et al. 2001). Hence, the temperature
gradients produce a horizontal as well a vertical barrier at the front which, therefore, acts as
catchment for the penguins’ planktonic prey.
The bimodality in depth frequencies was apparent, although less pronounced, when both
penguins foraged in warmer waters (Fig. 4.6). It is possible that the frequency of dives to
depths >65 m stem from diving activity close to and, therefore, still infl uenced by the front. In
contrast to this, the male that foraged in CTW only, showed a much greater diversity of dive
depths (Fig. 4.6). Unlike the birds that foraged in the STF, the penguin did not concentrate its
foraging efforts at any particular depth classes but utilised the entire water column to forage.
In CTW, the occurrence of productive patches is less defi ned than at the front and depends
largely on interactions of such factors as local nutrient availability and wind infl uence
(Murphy et al. 2001).
Although there are apparent differences in dive behaviour of males visiting the STF and males
that remain in CTW, the data do not allow conclusions with regard to the success of either
foraging strategies. The TDR-equipped bird foraging in the CTW stayed considerably longer
at sea than the birds that foraged at the STF (13.4 vs. 8.8 and 11.1 days). Conversely, the
penguin with GPS logger that apparently did not reach the STF returned earlier than the other
two GPS birds (7.9 vs. 10.2 and 15.3 days, Table 4.1).
The timing of the males’ return was overall still remarkably similar between all three years
and appears to be independent of hatching (Fig. 4.2). However, when compared to the
synchrony of the exodus, the males’ timing of return varied considerably more between
4. Infl uence of Oceanography and Seasonality on Snares penguins 62
individuals, quite possibly as a result of different foraging success (Davis & Renner 2003). In
this light, it can be assumed that body condition is the most important factor to infl uence the
males’ decision to return to the colony.
4.4.3 Foraging of female penguins
In comparison to the males, the females face a different situation when they leave for their
long-term foraging trips. Probably most importantly, the females’ time at sea is limited by the
need to return to the nest around the time of hatching to feed their chicks (Davis & Renner
2003). Considering that the females’ average foraging trip durations ranged between 5-6 days
in the fi rst two seasons, travelling to the STF to forage is an unlikely strategy. Accordingly,
the temperature profi les of two females fi tted with TDRs in 2003 show that the penguins
foraged solely in warmer CTW to the north or north-east of the Snares (Fig. 4.4). Temporally
detailed ChlA data is not available for the second half of October 2003 because of cloud cover
during the satellite passes. So, it is not possible to directly relate the estimated female foraging
area to oceanic productivity. However, considering that the ChlA concentration around the
Snares tends to increase rapidly from October on each year (Murphy et al. 2001), it is possible
that the females benefi ted from accelerated productivity closer to the islands.
Overall, the two females fi tted with TDRs showed higher diving activity and performed
considerably shallower dives than males (Table 4.3, Figure 6). A penguin’s chance to
encounter prey patches increases with distance travelled (Wilson & Wilson 1990). Given the
timeframe to forage, a combination of travelling and prey searching behaviour appears to be
a viable strategy for the females. This is also supported by the fact that the females exhibited
a much lower proportion of RMD dives, indicating that dive bouts targeting prey patches
at certain depths must have been brief and interspersed by shallower travelling episodes
(Table 4.3).
In 2004, the foraging situation for the females was somewhat different. While the males’
return dates were similar to the previous years, the chicks hatched about two days later
(Fig. 4.2) and the females could stay at sea for longer. The fact, the foraging trips were
indeed signifi cantly longer indicates that the females nest attendance patterns and, thus, time
4. Infl uence of Oceanography and Seasonality on Snares penguins 63
available for foraging is linked to egg status. This, in turn, makes it likely that other factors
such as hormonal mechanisms (e.g. Davis & Renner 2003, Massaro 2004) play a more
important role for the females’ return dates and hence foraging trip length than daylength and,
consequently, oceanography.
4. Infl uence of Oceanography and Seasonality on Snares penguins 64
Table 4.1. Basic foraging parameters of three male Snares penguins equipped with GPS loggers
on long-term foraging trips during incubation in 2003. Due to limited battery life of GPS devices,
maximum distance from island and distance travelled relate to the time of logger operation rather than
entire foraging trips. Considering the much longer duration of the trips, it is likely that at least distance
travelled represents a gross underestimation of the true distance covered by the penguins on their trips.
Bird ID T13 T14 T32
Date of departure 15.10.2003 16.10.2003 15.10.2003
Logger operation time (days) 2.9 2.7 3.1
Total trip duration (days) 15.3 7.9 10.2
Maximum distance from island (km) 215.2 96.1 163.3
Distance travelled (km) 226.3 132.5 187.9
ANOVA
Day 1 Day 2 Day 3 F2,6 p
Number of dives 270±130 a261±26 a339±132 a0.96 0.406
Descend rate (m/s) 0.7±0.1 a0.8±0.2 a,b 1.1±0.1 b11.82 0.001*
Max depth (m) 34.4±7.0 a48.4±15.0 a,b 63.8±9.7 b10.59 0.001*
Bottom time (s) 27. 9±2.6 a34.3±5.6 a48.3±7.6 b20.55 <0.001*
Ascend rate (m/s) 0.7±0.1 a0.8±0.1 a,b 0.9±0.1b4.33 0.033*
Dive time (s) 95.6±10.5 a115.4±24.3 a,b 140.8±14.3 b10.20 0.002*
Post-dive interval (s) 26.4±5.0 a34.4±10.6 a,b 40.5±5.7 b5.29 0.018*
Diving effi ciency° 0.25±0.02 a 0.27±0.04 a 0.27±0.02 a 0.81 0.463
Diving effort°° 0.77±0.03 a 0.75±0.03 a 0.77±0.03 a0.50 0.615
Table 4.2. Dive parameters of male Snares penguins (n = 6) equipped with GPS dive loggers and
TDRs performing long-term foraging trips during incubation in 2003. Comparison of data was only
possible for the fi rst three days at sea because of the GPS dive loggers’ limited battery life. All values
are given as daily means that derived from individual daily mean values for each bird. Asterisks
highlight signifi cant differences between days; superscript letters indicate relationship of differences,
i.e. values without common letter differ signifi cantly.
° Diving effi ciency= bottom time/(dive time+post-dive interval)
°° Diving effort=dive time/(dive time+post-dive interval)
4. Infl uence of Oceanography and Seasonality on Snares penguins 65
t-test
males females t3p
Trip length (days) 11.3±2.5 4.1±0 6.82 0.021
Daily dive activity ( dives*day-1) 259±75.5 499±4.24 -7.55 0.001
Hourly dive activity (dives*h-1) 17.2±4.6 34.3±3.0 -5.99 0.027
Descent rate (m/s) 0.8±0.2 0.9±0.2 -0.48 0.714
Max depth (m) 54.7±9.8 26.4±10.1 -3.11 0.090
Bottom time (s) 41.0±4.5 18.6±4.7 -5.31 0.034
Ascent rate (m/s) 0.9±0.1 0.8±0.1 -0.65 0.582
Dive time (s) 125.4±20.43 65.9±15.7 -8.68 <0.001
Post-dive interval (s) 31.6±5.4 20.1±3.8 -2.81 0.107
RMD dives° (% of all dives) 75.5±6.6 43.4±5.4 6.90 0.020
Diving effort°° 0.74±0.02 0.78±0.01 -2.32 0.259
Diving effi ciency°°° 0.23±0.01 0.28±0.01 -10.02 0.063
Table 4.3. Comparison of dive behaviour of male Snares penguins (n = 3) and female Snares
penguins (n = 2) equipped with TDRs on long-term foraging trips during incubation 2003. All
values are given as mean±sd that derived from individual means of each bird. Bold fi gures indicate
signifi cant difference.
° Repeated Maximum Depth dive = dives that return to the maximum depth±10% of the preceding dive
°° Diving effort = divetime/(divetime+post-dive interval)
°°° Diving effi ciency = bottom time/(divetime+post-dive interval)
4. Infl uence of Oceanography and Seasonality on Snares penguins 66
Stewart Island
The Snares
South Island
CTW SAW
T
a
s
m
a
n
C
u
r
r
e
n
t
S
o
u
t
h
l
a
n
d
C
u
r
r
e
n
t
166° 167° 168° 169° 170° 171°
47°
48°
49°
46°
EAST
SOUTH
New Zealand
S
U
B
T
R
O
P
I
C
A
L
F
R
O
N
T
Figure 4.1. Overview of the Snares, bathymetry and oceanography south of New Zealand’s South
Island. Course of the Subtropical Front and fl ow paths of and warm (red) surface currents in Central
Tasman Water (CTW) and cold (blue) surface currents in Subantarctic Water (SAW) currents adapted
from Carter et al. (1998). Dashed lines indicate 500, 1000 and 2000 m depth contours.
4. Infl uence of Oceanography and Seasonality on Snares penguins 67
Figure 4.2. Interannual synchrony in departure and arrival dates of adult Snares penguins during
incubation, and hatching dates in three consecutive breeding seasons, 2002-2004. Boxplot derives
from nest monitoring data collected each year between 10 October-15 November in one colony
(A3) on the Snares Islands. Only data from nests that successfully hatched at least one egg were
included. Medians are given as vertical lines, boxes enclose fi rst and third quartile of sample, whiskers
encompass 95% of sample, dots indicate outliers. Sample sizes in all years are given in top right
corner of fi gure.
no data
November
October
08 10 12 14 16 18 20 22 24 26 28 30 01 03 05 07 09 11 13
males leave
males return
females leave
females return
eggs hatch
15
2002
2003
2004
2002
2003
2004
2002
2003
2004
2002
2003
2004
2002
2003
2004
2002: 28 nests
2003: 126 nests
2004: 109 nests
4. Infl uence of Oceanography and Seasonality on Snares penguins 68
Figure 4.3. Partial foraging tracks of three male Snares penguins on long-term foraging trips during
incubation in 2003. Map shows the penguins’ at-sea movements in relation to weekly average
chlorophyll a (ChlA) concentration at the time (16-23 October 2003, from MODIS/Aqua data).
Isolines/shading represent ChlA concentration (mg/m³) at 0.05 mg/m³ intervals. No ChlA data were
available for white areas (cloud coverage). Dashed line indicates 200 m depth contour. Perpendicular
lines intersecting foraging tracks and adjacent numbers give position at midnight and according date
change. Line plots (A-D) summarise mean travelling speed and mean maximum dive depth (±standard
error, left column), and mean sea surface temperature and mean ChlA concentration (determined from
satellite data; ±standard deviation, right column) along the penguins’ tracks.
0.15
0.25
0.25
0.20
0.20
0.30
0.30
0.30
no data
no data
The Snares
16./17.
17./18.
15./16.
16./17.
15./16.
16./17. 17./18.
169°30’
169°00’
168°30’
168°00’
167°30’
167°00’
EAST
SOUTH
48°00’
48’30’
chlorophyll a concentration (weekly mean) - 16-23 October 2003
0100km
200m
Day 1 Day 2 Day 3
Horizontal speed (m/s)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
A
Sea Surface Temperature (°C)
8.0
8.5
9.0
9.5
10.0
10.5
11.0
11.5
Day 1 Day 2 Day 3
B
Chl A concentration (mg/m³)
0.10
0.15
0.20
0.25
0.30
0.35
0.40
Day 1 Day 2 Day 3
D
Day 1 Day 2 Day 3
Dive depth (m)
0
20
40
60
80
100
C
4. Infl uence of Oceanography and Seasonality on Snares penguins 69
Figure 4.4. Ambient water temperatures (between 5-10 m depth) recorded by Time-Depth Recorders
(TDR) on three male and two female Snares penguins undergoing long-term foraging trips during
incubation; and mean sea surface temperatures for October 2003 as determined from MODIS/Aqua
satellite data (ocean color). Temperature was sampled at 5 s (males) and 2 s (females) intervals,
hourly means were used to compile graphs. Grey bars in top graphs and grey area in temperature map
highlight sea surface temperatures between 9-10°C and represent the subtropical front (STF). Dashed
line indicates the 200 m depth contour. Note: monthly means used to compile map were skewed
towards cooler temperatures measured in the fi rst half of October; satellite data for the ‘warmer’
second half of the month (i.e. deployment period) was patchy.
The Snares
Stewart Island
sea surface temperature (monthly mean) - October 2003
8.0
8.5
9.0
9.5
10.0
10.5
11.0
11.5
11.0
12.0
169°30’
168°00’
167°30’
167°00’
EAST
SOUTH
47°30’
48°00’
48’30’
166°30’ 168°30’ 169°00’170°00’
0100km
females (TDR) - long term trips
STF
8.0
9.0
10.0
11.0
12.0
13.0
26.10. 27.10. 28.10. 29.10. 30.10. 31.10. 01.11.
ambient temperature (°C)
12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h
STF
8.0
9.0
10.0
11.0
12.0
13.0
14.10. 15.10. 16.10. 17.10. 18.10. 19.10. 20.10. 21.10. 22.10. 23.10. 24.10. 25.10. 26.10 27.10. 28.10.
ambient temperature (°C)
males (TDR) - long term trips
12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h 00h 12h
4. Infl uence of Oceanography and Seasonality on Snares penguins 70
Figure 4.5. Correlations of dive parameters determined for male Snares penguins (n = 3) with TDRs
on long foraging trips during the incubation stage in 2003. Graphs were compiled using daily means
of dive parameters and sea surface temperature of the respective foraging trips (durations: 10, 12 and
14 days). Signifi cance was tested using Pearson’s correlation.
sea surface temperature (°C)
Frequency of RMD dives (%)
20
30
40
50
60
70
80
90
9.0 10.0 11.0 12.0
Dive depth (m)
20
30
40
50
60
70
80
Diving effort (%)
68
70
72
74
76
78
80
82
84
86
88
Dive time (s)
80
100
120
140
160
180
sea surface temperature (°C)
Bottom time (s)
25
30
35
40
45
50
55
60
Post-dive interval (s)
15
20
25
30
35
40
45
50
r = -0.443
p = 0.008
r = -0.708
p < 0.001
r = -0.718
p < 0.001
r = -0.445
p = 0.022
r = -0.555
p = 0.001
r = -0.555
p = 0.001
9.0 10.0 11.0 12.0
9.0 10.0 11.0 12.09.0 10.0 11.0 12.0
9.0 10.0 11.0 12.09.0 10.0 11.0 12.0
ab
cd
ef
4. Infl uence of Oceanography and Seasonality on Snares penguins 71
Figure 4.6. Frequencies of dive depths in Snares penguins (3 males, 2 females) equipped with TDRs on long foraging trips during incubation 2003. Depth
frequencies were determined with regard of the watermass the birds foraged in: two males foraged at the Subtropical Front (STF, grey bars) and in Central
Tasman Waters (CTW, black bars); the third male and the two females remained in CTW for their entire duration of their foraging trips. Water masses were
distinguished via temperature readings (STF <10°; CTW >10°C) recorded by the dive loggers between 5 and 10 m depth (see Methods for details). Dotted
lines indicate the 50% mark of the cumulative depth frequencies.
0 2 4 6 8 1012141618
Frequency (%)
n = 5240 dives
5
20
35
50
65
80
95
110
125
024681012141618
Frequency (%)
n = 1270 dives
5
20
35
50
65
80
95
110
125
males (N = 2) male (N = 1) females (N = 2)
02468101214
246810
12
14
5
20
35
50
65
80
95
110
125
Frequency (%)
Dive depth (m)
STF CTW
n = 2851 dives n = 3730 dives
5. Foraging ranges and spatial distribution of dive activity 72
CHAPTER 5
FORAGING RANGES AND SPATIAL DISTRIBUTION OF
DIVE ACTIVIY IN FEMALE SNARES PENGUINS
EUDYPTES ROBUSTUS DURING THE
CHICK-GUARD STAGE
5. Foraging ranges and spatial distribution of dive activity 73
5.1 Introduction
During the breeding season, foraging ranges of seabirds are limited because the birds must
return to their nest sites regularly to attend to nesting duties (Gaston 2004). While this is
particularly true for the fl ightless penguins, there is still a considerable plasticity in foraging
strategies and ranges during the breeding season between different penguin species (e.g.
Wilson 1995, Davis & Renner 2003). Penguins have been broadly categorised as inshore and
offshore foraging species (Croxall & Davis 1999). During the chick guard stage of breeding
when chicks have to be fed frequently, inshore foragers such as Yellow-eyed penguins
Megadyptes antipodes or Little penguins Eudyptula minor tend to leave the nest in the
morning and return to feed the chicks in the afternoon of evening (e.g. Darby & Seddon 1990,
Numata et al. 2004). During the same stage, it is not unusual for offshore foragers like crested
penguins (Eudyptes spp.) to stay at sea for two days or more (Williams 1995). The main
determinant of whether a penguin species adopts an inshore or offshore foraging strategy is
food; or more specifi cally, the distribution and temporal availability of a species’ main prey
in the vicinity of the breeding location (Croxall & Davis 1999). This, in turn, is largely a
function of the prevailing environmental conditions at the breeding location (Davis & Renner
2003).
While the temporal prey availability for offshore foraging penguins is largely determined
by the seasonal nature of the environment (Williams 1995), the distribution of prey for
penguins, and seabirds in general, is often a result of physical oceanic processes (Hunt Jr.
1990). This is particularly so for penguin species breeding in isolated oceanic environment
like the subantarctic islands. Here, hydrographical features like fronts often represent areas
where nutrients and prey are accumulated by currents and, thus, provide enhanced foraging
conditions (Schneider 1990). It has been found that offshore foraging penguin species
target such areas to search for prey during the breeding season (e.g. Royal and Rockhopper
penguins, Hull et al. 1997, King penguins, Charrassin & Bost 2001). As a result, the foraging
ranges of the penguins vary according to the distance of their breeding site from areas of
enhanced prey availability.
5. Foraging ranges and spatial distribution of dive activity 74
The Snares penguin Eudyptes robustus is endemic to the small Snares island group some 200
km south of New Zealand’s South Island. Like other offshore foraging penguin species, the
foraging behaviour of Snares penguins is infl uenced by the presence of an oceanic front, the
Subtropical Front, which is located ca. 200 km east of the islands (Heath 1981; see Fig 4.1,
Chapter 4). During the incubation period, male Snares penguins perform long foraging trips
(>10 days) during which they target the productive waters of the front. However, as the males
need two days to reach the front, the time to travel there renders this destination unsuitable for
the Snares penguins after the chicks have hatched and must be fed regularly. This is especially
so for foraging females as for the fi rst three to four weeks following hatching they are the sole
providers of food while the males remain at their nests to guard the chicks (Warham 1974).
I studied the foraging behaviour of female Snares penguins during the chick-guard stage using
GPS data loggers and time depth recorders (TDR). The combination of geographical tracking
and simultaneous monitoring of dive behaviour and ambient water temperature allowed us
to analyse and interpret the penguins’ behaviour in the context of the marine environment in
which it occurred.
5.2 Methods
I studied the foraging ranges and dive activity of female Snares penguins during the chick
guard stage (November, Warham 1974) in the breeding seasons 2003 and 2004. The work was
carried out with female penguins breeding in one of the largest penguin colonies (A3, ~1200
pairs, Amey et al. 2001) on North-East Island (see Fig. 2.1, Chapter 2).
5.2.1 Data loggers
Foraging and diving behaviour was primarily studied with GPS data loggers (GPS-TDlog,
earth&OCEAN Technologies, Kiel, Germany; dimensions: L100xW48xH24 mm, mass:
~70 g). These devices record information of the penguins’ at-sea movements and dive
behaviour. The device comprises a GPS receiving unit that determines accurate geographical
5. Foraging ranges and spatial distribution of dive activity 75
position (position error <10 m) from signals from orbiting satellites of the Global Positioning
System (U.S. Department of Defence, USA). The device also contains high precision
environmental sensors that record dive depth (resolution: ~0.1 m) and ambient water
temperature (resolution: ~0.005°K). Data are stored with a timestamp in an internal non-
volatile fl ash memory and must be downloaded to a computer after the device is recovered
from the penguin. The logger’s sampling regime is freely programmable and was setup to
record a GPS position after each dive (“upon resurfacing”), and store depth and temperature
readings at 1 s intervals. As the acquisition of a GPS fi x takes between 25 and 30 seconds,
no position could be recorded if penguins stayed at the surface for shorter intervals between
dives (see Chapter 2).
In 2003, time depth recorders or TDRs (MK9, Wildlife Computers, Redmond, WA, USA;
dimensions: L67xW17xH17 mm, mass: ~30 g) were also deployed to test whether the larger
GPS loggers had a signifi cant impact on the females diving behaviour. The TDRs store dive
depth (resolution: 0.5 m) and temperature (resolution: 0.05°C). The devices were setup to
record dive depth and temperature at 1 s intervals similar to the GPS logger sampling regime.
The devices feature a wet/dry sensor which was activated so that sampling only occurred
when the logger was submerged.
Devices were attached with adhesive tape (Tesa-tape method, Wilson et al. 1997) to the
penguins’ lower backs to reduce drag (Bannasch et al. 1994). In 2003, loggers were deployed
at fi rst on birds in the colony. Females were captured at the nest while attending their
incubating mate, transferred out of the nesting area to be fi tted with a device and than released
again about 5 m from the nest. However, several of the females suffered from aggressive
responses of their mates directed at the attached device. This problem was circumvented by
deploying loggers at Station Cove (see Fig. 2.1, Chapter 2) just before the penguins entered
the sea. To do that, females from the observation area in colony A3 were marked at the nest
with a small dab of water soluble paint on their chest. The marked birds were then intercepted
at Station Cove when they emerged from the forest en route to the water. Before a logger was
fi tted, penguins were weighed with spring balances to the nearest 50 g. Only females heavier
than 2300 g (GPS logger mass <3% of body weight) were deemed fi t for deployment. During
5. Foraging ranges and spatial distribution of dive activity 76
the entire deployment procedure the penguins’ heads were covered with a cloth hood to
reduce stress. The handling time (i.e. capture, measurements, deployment and release) ranged
between 12 to 18 minutes.
In 2003, eight females were equipped with GPS loggers and an additional eight females were
fi tted with TDRs. In 2004, GPS loggers were deployed on a total 16 females. Due to the
limited battery life of the GPS loggers (see Chapter 2), the devices were generally recovered
after one short-term foraging trip (1-2 days). The birds fi tted with TDRs performed between
one and three foraging trips before recovery. After deployment, the penguins’ landing area
at Station Cove was kept under constant surveillance in order to intercept the logger birds
on their way back to the colony. If a bird was missed, it was recaptured in the colony. After
logger detachment the birds were again weighed and then released. The device recovery took
about 2-4 minutes.
5.2.2 General data analysis
I performed analysis of GPS and dive data was performed with custom-made software (T.
Mattern, unpublished data). The original GPS data were used to determine basic foraging
parameters for each recorded foraging trip, i.e. horizontal speed, travel distance and furthest
distance from island. Before further spatial analysis was performed, dive data were analysed
as follows.
Dive events were identifi ed when the loggers pressure transducers registered depth >1 m
(GPS loggers) or >3 m for the TDRs, which stemmed from the inaccuracy of the latter devices
at shallow depths. Dives generally consisted of three different phases, namely descent,
bottom and ascent phases. End of descent and start of ascent were defi ned as the moments
when a penguin’s rate of descent was <0.2 m/s and ascent was >0.2 m/s. Further parameters
determined for each dive were maximum dive depth, post-dive interval (i.e. the time spent
at the surface after the current dive), diving effort (dive time/[dive time+post-dive interval]),
diving effi ciency (bottom time/[dive time+post-dive interval]; Ydenberg & Clark 1989).
Furthermore, the current dive’s depth was examined with regard to the depth reached during
5. Foraging ranges and spatial distribution of dive activity 77
the preceding dive. If the current maximum depth fell within ±10% of the previous maximum
depth (cf Tremblay & Cherel 2000), the dive was defi ned to be a Repeated Maximum Depth
(RMD) dive. A high proportion of RMD dives during a foraging trip can be considered
an indicator for a penguin exploiting distinct prey patches at certain depths, whereas low
proportions of RMD dives suggest higher search effort at various depths.
5.2.3 Spatial analysis of dive parameters
I analysed the diving behaviour of penguins fi tted with GPS loggers with regard to the at-sea
movements of each bird. GPS data were interpolated linearly so that a geographic position
could be assigned to the temporal mid-point of each diving event. If more than one dive
occurred between two consecutive GPS fi xes, the dives were considered to have occurred
along the estimated linear path between both fi xes. This means that with increasing temporal
distance between consecutive fi xes, the estimated locations of dives between the fi xes became
increasingly inaccurate. Nevertheless, the linear connection between two fi xes was considered
an acceptable approximation when the time between fi xes was <6 h. As a result, data sets that
did not feature GPS positions during the day (see Chapter 2) were excluded from the spatial
analysis.
Spatial analysis of the dive data was conducted by superimposing an equal area grid (cell
size: 0.04°x0.04° longitude/latitude, ca. 4x2 km) over the general sea regions utilised by
all penguins in both years. Using the extrapolated dive data for each bird, mean values of
horizontal speed and dive parameters (number of dives, dive time, bottom time, dive depth,
proportion of RMD dives) were calculated for each grid cell a penguin passed through during
its foraging trip. For grid cells that were visited by more than one bird, the individual mean
values determined for these birds were averaged so that each grid cell comprised one average
value for each dive parameter. Using the geographic centre point of grid cells and their
corresponding mean value, contour graphs were generated via kriging interpolation (Fortin &
Dale 2005) to describe the spatial distribution of dive behaviour.
All statistical analyses were carried out in Minitab 14 (Minitab Inc, State College, PA,
5. Foraging ranges and spatial distribution of dive activity 78
USA). Details of statistical tests employed to compare data are given in the text or table and
fi gure captions. Averages are given as mean±standard deviation unless indicated otherwise.
Statistical signifi cance was accepted at the α<0.05 level.
5.3 Results
The eight deployments of GPS loggers on females during chick-guard 2003 yielded complete
sets of GPS and dive data for a total of fi ve foraging trips performed by fi ve different birds,
complete dive data but no GPS for two foraging trips (two birds), and one GPS logger
deployment resulted in only partial dive data being retrieved and was, therefore, excluded
from analysis. All of the TDR deployments were successful and dive data for a total of 18
foraging tips were recorded. During the deployment period, three females performed three
foraging trips each, four birds made two trips and one bird was recaptured after one trip. In
2004, 16 deployments of GPS loggers resulted in GPS and dive data for 14 trips made by 13
birds; the GPS data from one of these birds featured only night time fi xes and were excluded
from the spatial analysis. One deployment yielded only dive data. Two birds returned with
incomplete or no dive data and were subsequently excluded from analysis.
5.3.1 Infl uence of device size on foraging behaviour
In 2003, foraging behaviour was monitored with GPS loggers and TDRs that were ca. 30%
of the size of the GPS loggers. Despite the considerable size differences, no signifi cant
differences were apparent in dive parameters. Trip durations were similar for fi ve birds fi tted
with GPS loggers (mean duration: 28.9±6.5 h) compared to TDR birds (mean duration:
25.8±9.9 h; two-railed t-test: t10 = 0.73, p = 0.477). The same was true for other diving
parameters (GPS vs. TDR - dive time: 55.4±13.1 s vs. 57.9±8.5 s, t10 = 0.44, p = 0.669;
bottom time: 18.6±3.7 s vs. 17.6±3.2 s, t10 = 0.56, p = 0.585; post-dive interval: 27.5±12.9 vs.
31.9±12.3 s, t10 = 0.68, p = 0.511; diving effort: 0.68±0.07 vs. 0.65±0.08, t10 = 0.86, p = 0.405;
diving effi ciency: 0.24±0.06 vs. 0.20±0.02, t10 = 1.68, p = 0.136; dive depth: 16.0±4.6 m
vs. 19.7±3.0 m, t10 = 1.83, p = 0.097; proportion of RMD dives: 34.2±7.2% vs. 33.8±4.8%,
5. Foraging ranges and spatial distribution of dive activity 79
t10 = 0.12, p = 0.909). Therefore, dive parameters were considered independent from device
size and data from GPS loggers and TDR were pooled for further analysis.
5.3.2 At-sea movements
In 2003, four of the three birds that returned with GPS data foraged north to north-east of the
Snares (Fig. 5.1). One bird foraged north-west and spent the night and the fi rst fi ve hours of
the following day in very deep waters (>2000 m) beyond the shelf edge. In the following year
the foraging patterns of chick-guarding females were similar to that observed in 2003. Again,
the majority of the penguins foraged north to north-east of the Snares while only two birds
travelled north-west. However, neither of those birds reached waters >500 m (Fig. 5.1). The
similarity of the at-sea movements was also refl ected in the basic foraging parameters that did
not reveal any statistically signifi cant differences between the two years (Table 5.1).
5.3.3 Spatial distribution of diving performance
Only a small number of penguins returned with viable GPS data in 2003 and the overlap
of the individual penguins’ foraging tracks was low. Of the 119 distinct 0.04x0.04° grid
cells visited by the penguins, only 24 cells (~20%) featured more than one bird (Fig 5.2).
Consequently, the spatial analysis of diving behaviour in 2003 primarily refl ects individual
patterns. In the following year, penguin positions were recorded in 191 different grid cells,
106 of which (~56%) were visited by more than one bird (Fig. 5.2). The higher degree of
overlap of foraging trips by different birds provided a better spatial coverage of dive data so
that the resulting spatial interpolation provides a more general picture beyond the individual
patterns. Nevertheless, similar patterns in the spatial distribution of diving performance were
apparent in both years (Fig. 5.3a-d).
In both years, the horizontal speeds of penguins were highest within a radius of 25-50 km
from the island which indicates that the penguins exhibited predominantly travelling
behaviour closer to the island (Fig. 5.3a). This is underlined by the fact that the same sea areas
feature only short average dive times and relatively shallow (<20 m) average dive depths
5. Foraging ranges and spatial distribution of dive activity 80
(Fig. 5.3b&c). The proportion of RMD dives was also low closer to the island. Under the
assumption that travelling penguins dive to similar depths, the low RMD proportion shows
that travelling episodes were interspersed with deeper dives that presumably functioned as
prey search behaviour. The main prey searching and feeding activity, however, occurred at
distances >50 km. Beyond that distance, average dive depths were considerably deeper for
penguins foraging north-east of the Snares; the few birds that travelled north-west, performed
deeper dives between 25 and 50 km from the island. Greater dive depths generally involved
high proportions of RMD dives which further suggest that feeding is a major component of
the birds’ diving activity further away from the island. The areas of increased foraging activity
coincided with the presence of warmer surface waters (Fig. 5.4).
5.3.4 Differences in diving behaviour between the years
While the at-sea movements and the spatial distribution of the diving performance were
similar between 2003 and 2004, the comparison of diving parameters revealed considerable
differences. In 2003, the birds showed with an average 40.6±6.8 dives*h-1 against
28.5±10.2 dives*h-1 in 2004 a signifi cantly higher dive activity (Table 5.1). This difference is
a result of the signifi cantly greater mean dive depth observed in 2004 that ranged around 30 m
whereas in 2003 the mean dive depth was less than 20 m. Corresponding to the deeper dives,
dive times and diving effort also differed signifi cantly between both years, which also applied
to the bottom times that were longer 2004. Frequencies of dive depths reveal a bimodal
pattern in 2004 that was not apparent in 2003 (Fig. 5.5). The relationship between dive time
and bottom time was similar in both years, i.e. bottom time made up ca. 30% of the entire
dive time and ca. 20% of the entire duration of dive and post-dive interval in both years. As a
result the penguins’ diving effi ciency did not differ between 2003 and 2004 (Table 5.1). The
frequency of RMD dives in both years was comparable and suggests that the proportion of
travelling/searching behaviour and exploitation of prey patches at certain depths was similar
in both years.
5. Foraging ranges and spatial distribution of dive activity 81
5.4 Discussion
The foraging behaviour of female Snares penguins showed similarities but also some
differences between the two years. The foraging ranges and the spatial distribution of the
diving behaviour were comparable while the diving behaviour differed signifi cantly. In 2004,
the females showed to greater mean dive depths than in the previous year. It seems unlikely
that these differences were associated with the attachment of the devices.
5.4.1 Logger impact on foraging behaviour
Externally attached instruments alter the streamlined shape of penguins, cause additional drag
and therefore have the potential to negatively affect the diving performance and consequently
energy expenditure of penguins (Culik & Wilson 1991). However, depending upon the size
of the device – or more precisely the size if its frontal area – penguins are able to compensate
for the additional drag (Bannasch et al. 1994). The GPS loggers used during this study were
relatively large (frontal area: ca. 4.8% of a Snares penguin’s cross-sectional area) when
compared to the TDR deployed in 2003 (ca. 2.4%). However, the diving behaviour recorded
in 2003 with both device types did not reveal any obvious differences. Although samples sizes
were small this indicates that instrumented Snares penguins managed similarly to compensate
for drag caused by the devices, regardless of the size differences.
5.4.2 Foraging ranges of female Snares penguins
The female Snares penguins tended to forage primarily in an area between 50 and 100 km
north to north-east of the Snares. In this area the birds stayed over the continental shelf and
only rarely ventured for short periods into the deeper waters of the Solander Trough in the
west. The few birds that foraged west to north-west of the Snares, also remained in waters
<500 m for most of their time at sea; only in 2003 one female spent some time in very deep
water (Fig. 5.1). The shelf west of the Snares is relatively narrow and consequently the birds
performed shorter trips and had shorter foraging ranges than birds travelling north. However,
5. Foraging ranges and spatial distribution of dive activity 82
the predominance of the northerly courses taken by the birds suggests that the sea areas half-
way between the Snares and Stewart Island must have been considerably more attractive for
the birds despite the greater distance from the Snares.
5.4.3 Comparisons with other crested penguins
Several studies of crested penguins have revealed that there are considerable location-
dependent differences in foraging ranges and behaviour, which are generally believed to
refl ect local differences in the marine environment (e.g. Tremblay & Cherel 2003, Schiavini
& Raya Rey 2004). For example, chick-guarding female Rockhopper penguins Eudyptes
chrysocome from Macquarie Island forage on average longer (ca.7 days, Hull 1999) than do
conspecifi cs breeding on New Zealand’s Antipodes Island (ca. 0.4 days Sagar et al. 2005),
which also implies there will be differences in their foraging ranges. This shows that a direct
comparison of foraging patterns is of only limited value and it is more useful to consider the
crested penguins’ foraging behaviour in the context of their local marine environment.
The foraging movements of Rockhopper penguins from Macquarie Island are associated with
the presence of the Subantarctic Front (Hull 1999), where hydrographical processes are likely
to provide a predictable and enhanced food situation for seabirds in an oceanic environment
(Hunt Jr 1990). Although it is not mentioned by the authors, foraging tracks of Rockhopper
penguins from the Antipodes Islands presented in Sagar et al. (2005) suggest a strong
infl uence of the Subantarctic Front at that location too (cf tracks with the location of the front
given in Heath 1985). While male Snares penguins during incubation also forage in waters
of a productive oceanic front (the Subtropical Front, see Chapter 4), the location of the front
some 200 km east of the Snares lies beyond the foraging ranges determined for the female
penguins. Instead, the females remained in warm subtropical waters over the shelf north of the
Snares (Fig. 5.1).
5. Foraging ranges and spatial distribution of dive activity 83
5.4.4 Infl uence of the oceanic environment
Tremblay & Cherel (2003) found that Rockhopper penguins breeding on Amsterdam Island
in the Indian Ocean north of the Subtropical Front foraged harder and with less success than
conspecifi cs further south. Their higher foraging effort correlated with the low productivity of
the subtropical waters, which resulted in low zooplankton biomass and, hence, low abundance
of penguin prey. Oceanic primary production depends greatly on the availability of nutrients
which often represent limiting factors in open ocean areas (Falkwoski et al. 1998). Such a
situation seems to prevail in the subtropical waters around Amsterdam Island (e.g. Zentara &
Kamykowski 1981, Sieracki et al. 1993).
Although chick-rearing female Snares penguins also forage exclusively in subtropical waters,
the oceanographic situation they face is different. Around New Zealand, subtropical water
masses feature higher primary production than subantarctic waters (Chang & Gall 1998). The
reduced productivity in the subantarctic probably stems from a limitation of iron which is an
essential element for plankton growth (Boyd et al. 2000). In the waters around New Zealand,
iron limitation is not evident, probably due to nutrient infl ux from the landmasses (Boyd
et al. 1999). The sea area utilised by the female Snares penguins is strongly affected by a
strong surface current (Tasman current) that transports warmer, coastal waters from the South
Island’s west coast around the south of Stewart Island towards the east, where it then becomes
the Southland current (Heath 1981; see Fig. 4.1, Chapter 4). As a result, the sea areas around
the Snares feature considerably higher phytoplankton concentrations than the subantarctic
regions south of the Subtropical Front to the east and the south of the islands (Murphy et al.
2001). High phytoplankton biomass is benefi cial for the abundance of macrozooplankton
(Bradford-Grieve et al. 2003), which, in turn, comprises the bulk of the Snares penguins diet
(see Chapter 2).
The infl uence of the warm current refl ects clearly in the temperature data recorded with GPS
loggers on the penguins in 2004 (Fig. 5.4). Sea surface temperatures closer to the Snares were
cooler <11°C), whereas towards the north and beyond the 50 km radius, the water was up
to 2°C warmer. Hence, female Snares penguins profi t from the proximity of their breeding
5. Foraging ranges and spatial distribution of dive activity 84
colony to the New Zealand mainland and the transport of coastal waters towards the Snares.
The importance of the warm coastal waters on the foraging behaviour of the penguins is
underlined by the spatial distribution of the penguins’ dive activity which was location
dependent. Closer to the island and, thus, in cool waters, travelling dominated their behaviour,
while prey searching and feeding seemed to occur predominantly in the warmer water
between 50 and 100 km north-east of the island (Fig. 5.3a-d). The concentration of foraging
effort to this area could also be infl uenced by hydrographical processes where the Tasman
current meets the Snares shelf (see Fig. 4.1, Chapter 4). The steep bathymetrical gradient
north of the Snares is likely to affect the eastward fl ow of the current by causing turbulence
which, in turn, plays an important role in the vertical and horizontal distribution of plankton
(Jillett 2003) possibly contributing to a prey distribution benefi cial for the penguins.
5.4.5 Different diving behaviour between years
In 2004, the penguins dived signifi cantly deeper and, as a consequence, considerably longer
than the year before (Table 5.1). While this was also refl ected in an increased diving effort,
the diving effi ciency was similar in both years. Given the fact that the penguins exhibited
comparable frequencies of dive bouts during which they returned to specifi c depths
– presumably to exploit prey patches (RMD dives) – it seems reasonable to assume that the
penguins were equally successful in terms of fi nding prey in both years.
Penguins generally react to deteriorating prey situations in two ways: either by extending their
foraging range to increase their chances of encountering prey patches (e.g. Wilson & Wilson
1990) or by increasing their diving effort to search greater volumes of the water column
(e.g. Mattern 2001). Evidently, the foraging ranges of the female Snares penguins give no
indication for differences in prey distribution in both years. Furthermore, the frequencies of
dive depths in 2004 show a distinct bimodality (Fig. 5.5). Either the penguins performed dives
<30 m deep (like they did in 2003) or they reached depths between 60-80m. This contrasts
with, for instance, the foraging behaviour of Little penguins in the face of food shortage. In
this species, prey scarcity leads to increased vertical searching behaviour during which the
birds cover all depth classes evenly (Mattern 2001). The bimodality of the Snares penguins’
5. Foraging ranges and spatial distribution of dive activity 85
dive depths, therefore, suggests that deep dives were aimed at specifi c depths – possibly to
exploit prey patches associated with the thermocline (e.g. Gallager et al. 2004) – rather than
randomly searching for prey across the entire available depth spectrum. Hence, the differences
in dive behaviour between the two years seem to be more a result of differing vertical
distribution of prey than of disparate prey availability.
5. Foraging ranges and spatial distribution of dive activity 86
Table 5.1. Comparison of foraging and dive parameters of females Snares penguins during the chick-
guard stage in 2003 and 2004. Foraging parameters derived from data recorded with GPS loggers,
diving parameters were determined from data collected with GPS loggers and TDRs, the latter
deployed in 2003 only (see Methods for details). All values are given as mean±SD and were tested
for normality (Kolmogorov-Smirnov) before statistical comparison. Test results printed bold indicate
signifi cant differences.
two-tailed t-test
2003 2004 tp
Foraging parameters
Number of birds 5 14
Travel distance (km) 114.6±23.3 127.5±39.9 -0.81 0.434<