UC San Diego UC San Diego Electronic Theses and Dissertations

Title Over-winter behavior and annual survival of Pygoscelid penguins in the South Shetland Islands

Permalink https://escholarship.org/uc/item/8ff0947q

Author Hinke, Jefferson Thomas

Publication Date 2012

Peer reviewed|Thesis/dissertation

eScholarship.org Powered by the California Digital Library University of California UNIVERSITY OF CALIFORNIA, SAN DIEGO

Over-winter behavior and annual survival of Pygoscelid penguins in the South Shetland Islands

A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy

Marine Biology

Jefferson Thomas Hinke

Committee in charge:

Gerald Kooyman, Chair Ian Abramson Jay Barlow Jennifer Smith George Sugihara Wayne Trivelpiece

2012

The Dissertation of Jefferson Thomas Hinke is approved, and it is acceptable in quality and form for publication on microfilm and electronically:

______

Chair

University of California, San Diego

2012

iii

DEDICATION

This thesis is dedicated to my best friend and wife, Katrina.

TABLE OF CONTENTS

SIGNATURE PAGE ...... iii

DEDICATION ...... iv

TABLE OF CONTENTS ...... v

LIST OF FIGURES ...... vii

LIST OF TABLES ...... ix

ACKNOWLEDGEMENTS ...... x MATERIAL SUBMITTED FOR PUBLICATION IN THE DISSERTATION ...... xii

CURRICULUM VITAE ...... xiii

ABSTRACT OF THE DISSERTATION ...... xvi

GENERAL INTRODUCTION ...... 1

CHAPTER 1: Daily activity and minimum food requirements during winter for gentoo penguins (Pygoscelis papua) in the South Shetland Islands, Antarctica ...... 12 ABSTRACT ...... 13 INTRODUCTION ...... 14 MATERIALS AND METHODS...... 16 RESULTS ...... 24 DISCUSSION ...... 27 REFERENCES ...... 44

CHAPTER 2: Adélie penguin survival rates and their relationship to environmental indices in the South Shetland Islands ...... 49 ABSTRACT ...... 50 INTRODUCTION ...... 51 MATERIALS AND METHODS...... 55 RESULTS ...... 67 DISCUSSION ...... 72 APPENDIX 1 ...... 90 REFERENCES ...... 91

CHAPTER 3: Integrating long-term demographic data with a matrix population model ...... 98 ABSTRACT ...... 99 INTRODUCTION ...... 100 MATERIALS AND METHODS...... 103 RESULTS ...... 110 DISCUSSION ...... 113 REFERENCES ...... 126

CHAPTER 4: Rapid climate change and life history: how plastic is the Adélie penguin? ...... 131 ABSTRACT ...... 132 INTRODUCTION ...... 133 MATERIALS AND METHODS...... 139 RESULTS ...... 143 DISCUSSION ...... 146 REFERENCES ...... 163

LIST OF FIGURES

Fig. 0-1: Location of study sites in the South Shetland Islands, Antarctica ...... 7

Fig. 1-1: Map of the study area in the South Shetland Islands, Antarctica. Study sites were located at Admiralty Bay, King George Island and Cape Shirreff, Livingston Island...... 39

Fig. 1-2: Comparison of temperature records recorded by tags attached to two penguins, the ambient air temperature, and the water temperature during the test study ...... 40

Fig. 1-3: Summary of forging trip types conducted by gentoo penguins during the winter. ... 41

Fig. 1-4: Duration and timing of foraging trips by gentoo penguin from April through September...... 42

Fig. 1-5: Estimated average daily minimum food requirement necessary to meet daily energy expenditures throughout the winter...... 43

Fig. 2-1: Total number of breeding pairs in the Copacabana colony and the tagging rate relative to population size during the study period...... 83

Fig. 2-2: Proportion of banded individuals from each year of release that were eventually recaptured...... 84

Fig. 2-3: Comparison of return statistics for Adélie penguins with aluminum and stainless steel flipper bands...... 85

Fig. 2-4: Mean ages of individuals banded with aluminum and stainless steel flipper bands that were recaptured from each cohort released...... 86

Fig. 2-5: Estimates ± standard error of apparent survival rates from the best-fitting model for juvenile Adélie penguins (age 0-2)...... 87

Fig. 2-6: Estimates ± standard error of apparent survival rates from best-fitting model for adult Adélie penguins (age 2+)...... 88

Fig. 2-7: Estimates ± SE of recapture probability for Adélie penguins from the best fitting model...... 89

Fig. 3-1: Diagram of matrix model structure and transition matrix parameterization...... 122

Fig. 3-2: Model fits to number of nests and model predictions of chick production...... 123

vii

Fig. 3-4: Probability of local extirpation and mean population growth rates of Adélie penguins under future scenarios of environmental variability...... 125

Fig. 4-1: Core life history relationships and their correlations, denoted with a positive or negative signs, with adult survival rate...... 156

Fig. 4-2: Map of Antarctica with location of the three main study sites at Admiralty Bay, Cape Crozier, and Béchervaise Island ...... 157

Fig. 4-3: Comparison of age distributions for first-time male and female breeders at Cape Crozier and Admiralty Bay...... 158

Fig. 4-4: Average reproductive success of Adélie penguins across space and time at selected breeding colonies ...... 159

Fig. 4-5: Average mass of chicks at crèche and fledging across space and time at selected breeding colonies...... 160

Fig. 4-6: Adélie penguin census data at Admiralty Bay and the best fitting Leslie matrix model (solid line). Dashed lines are model predictions based on retention rates of 0.64 and 0.63...... 161

Fig. 4-7: Survivorship of Adélie penguins from Béchervaise Island (dashed line), Cape Crozier (thin line) and Admiralty Bay (thick line)...... 162

viii

LIST OF TABLES

Table 1-1: Summary of tag deployments and foraging trip data for gentoo penguins during winter ...... 36

Table 1-2: Parameters used to estimate the daily minimum food requirement ...... 37

Table 1-3: The top five models compared to a base model with only day length, ranked by AICc score. The null deviance for all models was 4828 and the lowest AIC value was 3906.5 ...... 38

Table 2-1: Annual estimate of population size and number of Adélie penguins banded with aluminum or stainless steel flipper bands and subsequently recaptured during the study period (1985-2009) ...... 79

Table 2-2: Model selection table for survival (φ ) and recapture (p) probabilities. Models are ranked according to ΔQAICc and assume a value of cˆ = 0.2 . Alternative rankings assuming lower values of cˆ and the resulting QAICc weights (in parentheses) are provided ...... 80

Table 2-3 Percent of temporal variation in survival (φ ) and recapture (p) models accounted for by environmental covariates ...... 81

Table 2-4 Correlation coefficients for the relationship between estimated survival and recapture parameters and select environmental variables ...... 82

Table 3-1: Mean reproductive success and breeding propensity of first time breeders and adults ...... 120

Table 3-2: Model estimates of the correction factors and final estimates of adult survival rates. For years of overlap, final estimates of juvenile and adult survival rates were calculated as a weighted average based on the proportion of birds with aluminum bands that were recaptured ...... 121

Table 4-1: Source of published data and parameters available from the original studies on Adélie penguin populations around the continent ...... 152

Table 4-2: Best fitting models for survival and recapture probabilities of Adélie penguins at Admiralty Bay, 1997-2008. The minimum QAICc was 2305.4. Quasi-AICc is calculated assuming over-dispersion in the data ( cˆ =1.62) ...... 153

Table 4-3: Life table analysis of Adélie penguins at Admiralty Bay ...... 154

Table 4-4: Population properties derived from life-table analyses...... 155

ACKNOWLEDGEMENTS

This dissertation was made possible by the support and guidance of many. First and foremost, I thank Dr. Wayne Trivelpiece for access to a rich data set from which most of this work is drawn. His life-time commitment to monitoring penguin populations in the name of basic and applied research in a remote field camp in the South Shetland Islands is inspiring and following his pioneering footsteps in Antarctic seabird research is humbling. Working with Wayne at his home away from home in the tiny field camp on King George Island has been enlightening, exciting, and, quite literally, life changing. Given the opportunity to visit the field camps in 2005, I abandoned prior plans to return to graduate school. Instead, I threw my lot in with the many others who, once having experienced work in Antarctica, find it difficult to envision working anywhere else. My first short excursion among penguins and seals in the Southern Ocean helped align my research interests in ecology, fisheries, behavior, and life history, which have culminated in this dissertation.

Among Antarctic research pioneers, I am equally lucky to have been advised by Dr.

Gerald Kooyman. Though we never had the opportunity to work together on the ice, his enthusiasm for work in Antarctica and genuine interest in the ecology and population changes of Antarctic marine life broadened my perspectives on research, and fostered my interests in

Antarctic conservation. His open mind, critical thinking, clear writing, and ingenuity provide a model for scientific endeavors that I can only hope to emulate in my career.

My colleagues at the Antarctic Ecosystem Research Division at the Southwest

Fisheries Science Center in La Jolla are also owed a great debt of gratitude. Chief among the conspirators who helped to set me on the road toward a doctoral degree is Dr. George Watters.

His mentoring, both before and during my tenure as a graduate student, have been invaluable.

From our work together on tuna fisheries in the eastern tropical Pacific, at-sea behaviors of

Pacific salmon, and food web modeling of krill fisheries in Antarctica, I’ve been given a breadth of research experience far beyond my expectations. I sincerely look forward to our future collaborations in Antarctic science and the conservation of Antarctic marine resources.

Many others at NOAA also provided critical sounding boards for ideas and helped keep my morale high over the past five years. Thanks to Dr. Christian Reiss, Stephanie Sexton, Dr.

Rennie Holt, Dr. Michael Goebel, Susan Trivelpiece, Anthony Cossio, Dr. Christopher Jones,

Dr. Doug Kinzey, Amy Van Cise, Jen Walsh, Jessica Lipsky, Raul Vasquez del Mercado,

Russell Haner, Douglas Krause, Aileen Miller, Ryan Driscoll, and the numerous field assistants that have worked in the camps with me and before me. I especially extend a big thanks to Elaine Leung, Stephen Agius, Michael Polito, Dave Loomis, Scott Rogers, and

Rachael Orben for help deploying and retrieving instruments that resulted in the data presented in the first Chapter.

The Scripps Institution of Oceanography at the University of California, San Diego is a wonderful place to have called school, work, and home. The breadth of material covered during the first year courses, the latitude to choose freely among the myriad academic offerings, a stunning setting, a healthy social and outreach scene all blended together to make life as a graduate student rich and full of opportunity. It is with great pride that I will be able to call UCSD my alma mater. Fellow students in the Kooyman and Ponganis lab (we’re a small bunch) helped immensely with questions about classes, proposals, qualifying and thesis writing. I extend a big thank you to Dr. Shannon Barber-Meyer, Dr. Jessica Meir, Dr.

Cassondra Williams, and Geoffrey Gearheart for setting great examples for how to succeed and have fun doing it.

None of this would have been possible without my family. My wife, Katrina, and I met for the first time shortly after I began my studies. She has only known me as a student, always engaged in reading, writing, too frequently lost in the vacant-eyed stares of thinking

about my research, or packing my bags for yet another months-long adventure in Antarctica.

Yet over the last 5 years we managed to find time for ourselves when we needed it most, to encourage each other when our mountains of tasks seemed insurmountable, and even stole away to get married while both of us were in the frantic midst of finishing our degrees. She helped make my time as a graduate student as enjoyable as any I remember. Katrina, here’s to the beginning of our long and happy post-graduate student life together. Finally, for my parents, who encouraged me to choose my path and unconditionally supported my decisions at every juncture, I extend a very special and heartfelt thanks.

MATERIAL PUBLISHED/SUBMITTED FOR PUBLICATION IN THE

DISSERTATION

Chapter 1, in full, was published in Polar Biology: Hinke JT, Trivelpiece WZ (2011).

Daily activity and minimum food requirements of gentoo penguins (Pygoscelis papua) in the

South Shetland Islands, Antarctica. Polar Biology 34:1579-1590. The dissertation author was the primary investigator and author of this paper.

Chapter 4, in full, is currently being prepared for publication as:

Hinke JT, Trivelpiece WZ. Rapid climate change and life history: how plastic is the Adélie penguin? The dissertation author was the primary investigator and author of this paper.

xii

CURRICULUM VITAE

EDUCATION

Ph.D. Scripps Institution of Oceanography, UCSD Marine Biology, 2012

M.S. University of Wisconsin - Madison Zoology, 2001 Advisor: James F Kitchell

B.S. University of Wisconsin - Madison Zoology, Biological Aspects of Conservation, 1999

HONORS AND AWARDS

NSF Travel Award for 6th International Penguin Conference 2007 Cheng An Lun Fellowship 2006 Doherty Fellowship 2006 Thomas Joseph Walsh Fellowship 2006 Phi Kappa Phi. 1999 Deans List. 1999 Research Experience for Undergraduates grant for research at the University of Notre Dame Environmental Research Center. 1998

PUBLICATIONS

Trivelpiece WZ, J.T. Hinke, A.K. Miller, C.S. Reiss, S.G. Trivelpiece, G.M. Watters. 2011. Variability in krill biomass links harvesting and climate warming to penguin population changes in Antarctica. Proceedings of the National Academy of Sciences. 108:7625-7628

Hinke J.T., W.Z. Trivelpiece. 2011. Daily activity and minimum food requirements during winter for gentoo penguins (Pygoscelis papua) in the South Shetland Islands, Antarctica. Polar Biology 34:1579-1590

Hill, S., K. Reid, S. Thorpe, J. Hinke, G. Watters. 2007 A compilation of parameters for ecosystem dynamics models in the Scotia Sea – Antarctic Peninsula region. CCAMLR Science 14:1-25

Hinke, J.T., K. Salwicka, S. G. Trivelpiece, G.M. Watters, W.Z. Trivelpiece. 2007. Divergent responses of Pygoscelis penguins reveal common environmental driver. Oecologia 153:845-855

Sass, G.G, C.M. Gille, J.T. Hinke, and J.F. Kitchell. 2006. Whole-lake influences of littoral structural complexity and prey body morphology on fish predator-prey interactions. Ecology of Freshwater Fish 15:301-308

xiii

Hinke, J.T., D.G. Foley, C. Wilson, G.M. Watters. 2005. Persistent habitat use by Chinook salmon (Oncorhynchus tshawytscha) in the coastal ocean. Marine Ecology Progress Series 304:207-220

Hinke, J.T., G.M. Watters, G. Boehlert, and P. Zedonis. 2005. Ocean habitat use in autumn by Chinook salmon in coastal waters of Oregon and California. Marine Ecology Progress Series 285:181-192

Hinke, J.T, I.C. Kaplan, K. Aydin, G.M. Watters, R.J. Olson, and J.F. Kitchell. 2004. Visualizing the food-web effects of fishing for tunas in the Pacific Ocean. Ecology and Society 9(1):10 [online] URL: http://www.ecologyandsociety.org/vol9/iss1/art10

Fisher, D.R., B.W. Hale, J.T. Hinke, and C.A. Overdevest. 2002. Social and ecological responses to climatic change: towards and integrative understanding. International Journal of Environment and Pollution 17:323-326

Greenfield, B.K., D.B. Lewis, and J.T. Hinke. 2002. Effect of injury in salt marsh periwinkles (Littoraria irrorata [Say, 1822]) on resistance to future attacks by blue crabs (Callinectes sapidus [Rathbun, 1896]). American Malacological Bulletin 17:141-146

RESEARCH EXPERIENCE

Seabird Biologist, Antarctic Ecosystem Research Division NOAA/SWSC. September 2005-September 2006.

Fishery Biologist. Joint Institute for Marine and Atmospheric Research, Pacific Fisheries Environmental Laboratory NOAA/SWFSC. September 2002- September 2005.

Research Intern. National Center for Ecological Analysis and Synthesis, Santa Barbara, CA. January - March 2002.

Volunteer. PISCO – Collaborative survey of intertidal organisms. Anacapa Island. 2001.

Volunteer. Grand Canyon Monitoring and Research Commission. Little Colorado River, 2001.

Trainee. Integrative Graduate Education and Research Training (IGERT) at UW-Madison, 1999-2001.

TEACHING EXPERIENCE

Teaching Assistant. Zoology 510/511. Ecology of Fishes Lecture/Laboratory. UW-Madison, 2001.

Undergraduate Teaching Assistant. Zoology 435. Comparative Vertebrate, Anatomy Laboratory. UW-Madison, 1998.

xiv

FIELDS OF STUDY

Marine and freshwater ecology Behavior and life history Archival and satellite tagging Demography Statistical modeling

ABSTRACT OF THE DISSERTATION

Over-winter behavior and annual survival of Pygoscelid penguins in the South Shetland

Islands

Jefferson Thomas Hinke

Doctor of Philosophy in Marine Biology

University of California, San Diego, 2012

Gerald Kooyman, Chair

Pygoscelid penguin populations throughout the Antarctic Peninsula region have changed rapidly in recent decades. Ongoing climate change is thought to underpin these changes through bottom-up effects on habitat suitability and prey availability, ultimately affecting penguin behavior, survival and reproduction. To quantify winter behavior and energetic requirements required to support winter activity and to estimate population-level consequences of variation in survival rates under conditions of rapid climate change, this dissertation investigates two projects, each using a different penguin species.

Daily activity and energetic demands during winter were estimated for gentoo penguins (Pygoscelis papua) using data from archival temperature tags. Foraging trip frequencies ranged from 0.85 to 1.0 trips day-1 and total trip durations were positively correlated with day length. Mean daily food requirements, based on a mixed diet of fish and krill (Euphausia superba) were estimated at 0.70 ± 0.12 kg day-1. Early winter foraging trips

xvi

matched day length better than late winter foraging trips, suggesting that individuals maximized foraging time during the early winter period to recover body mass following the breeding season and molt. The attenuated response of foraging trip durations to increasing day length in late winter may be related to differences in local resource availability or individual behaviors prior to the upcoming breeding season.

To investigate population-level consequences of variation in survival rates, data from long-term mark-recapture studies of Adélie penguins (Pygoscelis adeliae) were integrated in a stochastic population model to estimate the risk of local extirpation. No trends in survival rates were evident, and variability in survival rates was poorly explained by the selected suite of environmental covariates. Stochastic projections based on the extant variability of survival rates suggests that small increases in the frequency of years with poor survival result in a rapid increase in risk of near-term local extirpation. Compared to other populations of Adélie penguins around the Antarctic continent, survivorship and population growth rates were lowest in the northern Antarctic Peninsula region. Despite no simple correlations with environmental indices, it is readily apparent that Adélie penguins are vulnerable to the rapid environmental changes that are occurring in the Antarctic Peninsula region.

xvii

INTRODUCTION

The size and distribution of Pygoscelid penguin populations throughout the Antarctic

Peninsula region over the last few decades have changed rapidly (e.g., Forcada et al. 2006;

Hinke et al. 2007; Lynch et al. 2008; Schofield et al. 2010; Trivelpiece et al. 2011). Breeding populations of Adélie (Pygoscelis adeliae) and chinstrap (P. papua) penguins have declined by at least 50% in most locations, while the abundance of gentoo penguins has increased and their range has expanded southward (Lynch et al. 2008; Trivelpiece et al. 2011). Climate change is thought to underpin these changes through bottom-up effects on habitat suitability and prey availability, which ultimately affects survival and reproductive rates (Trathan et al.

1996; Wilson et al. 2001; Jenouvrier et al. 2006).

During this period of climate and population change, most penguin research has focused on foraging ecology, energetics, and reproductive biology during the short austral summer (e.g., Trivelpiece et al. 1987; Davis et al. 1989; Bevan et al. 2002; Lescroël et al.

2004; Miller et al. 2009; Kokubun et al. 2010; Wilson 2010). Inferences from studies conducted during the summer, however, suggest that environmental conditions during winter largely affect the annual reproductive, survival, and recruitment rates of penguins (Fraser et al.

1992; Vleck and Vleck 2002; Trathan et al. 1996; Ainley 2002; Hinke et al. 2007).

Recognition of the importance of winter conditions for penguin populations has led to a growing emphasis on behavioral and ecological studies conducted during winter (e.g.,

Williams 1991; Wilson et al. 1998; Trivelpiece et al. 2007, Ballard et al. 2010) and on estimating annual survival rates with respect to environmental conditions (e.g., Ballerini et al.

2009; Emmerson and Southwell 2011)

Compounding the importance of winter for penguins are on-going trends in environmental conditions that affect population-level processes. In the Antarctic Peninsula

region, such changes include reductions in sea ice extent and duration (Stammerjohn et al.

2008), increases in air and sea surface temperatures (Vaughan et al. 2003; Meredith and King

2005), potential declines in primary productivity (Montes-Hugo et al. 2009), and long-term declines in the density of key food resources in the Southern Ocean such as Antarctic krill

(Euphausia superba; Atkinson et al. 2004; Reiss et al. 2008). Such changes are likely to be important factors affecting the behavior, reproduction, abundance, and distribution of penguins around Antarctica (Wilson et al. 2001; Jenouvrier et al. 2006; Forcada et al. 2006;

Hinke et al. 2007; Emmerson and Southwell 2008).

Seabird life histories are generally described by low fecundity but stable, high adult survival rates which act to buffer populations from the vagaries of environmental variation.

However, numerous studies suggest that observed variation and directional change in climate and environmental conditions have affected the breeding phenology, ranges, and populations trends of numerous seabird populations (Viet et al. 1996; Grosbois and Thompson 2005;

Barbraud and Weimerskirch 2006; Sandvik and Erikstad 2008; Devney et al. 2009). Given the sensitivity of penguins to environmental change, information on the activity patterns, energetic requirements, and survival rates for Pygoscelid penguins during a period of large-scale environmental change are important for conservation and management of Antarctic wildlife.

Such information provides a foundation for understanding the ecological role of penguins in a changing environment and the likely causes of variation in their abundance and distribution.

Providing information on behavior and survival hinges on recent advances in biologging technologies and the fruition of long-term monitoring studies. For example, the miniaturization of archival tags enables direct observations of penguin behaviors in the wild that previously had been inaccessible, including full migratory pathways (e.g., Ballard et al.

2010) and foraging behaviors (e.g., Green et al. 2009) during winter. For long-lived animals, estimates of inter-annual variation in demographic rates require long-term monitoring

programs with uniquely marked individuals (Lebreton et al. 1992). The convergence of such studies has been made possible through field programs supported by the Antarctic Ecosystem

Research Division at the Southwest Fisheries Science Center in La Jolla, California. By leveraging data from a topical tagging study and from long-term demographic data sets collected since 1985, this dissertation seeks to quantify over-winter behavior and minimum energetic requirements required to support such activity, and to estimate survival rates under conditions of rapid climate change.

Objectives

This dissertation asks three main questions about over-winter behavior of penguins and the extent of variation in annual survival rates with respect to environmental change on population dynamics. First, what are the activity and energy budgets of penguins during the winter? Second, how have penguin survival rates responded to rapid climate change in the

Antarctic Peninsula and what are the consequences for population dynamics? Finally, how have different penguin populations responded to climate variation over space and time?

To answer these questions, I have undertaken two parallel projects, each using a different penguin species that breeds in the South Shetland Islands, an archipelago of north of the Antarctic Peninsula (Fig. 0-1). For investigating winter activity patterns and minimum food requirements required to support such activity, I used three years of data collected with archival temperature tags to quantify the timing and duration of winter foraging trips of gentoo penguins. To examine population dynamics with respect to environmental conditions, I used

25 consecutive years (1985-2009) of monitoring data on Adélie penguins banded as chicks and monitored for survival and reproductive success over the duration of their lives.

The Penguins

Gentoo penguins were chosen for the over-winter behavioral observations because they are a relatively large and non-migratory penguin (Croxall 1984; Tanton et al. 2004).

Their size helps to minimize the impact of the long-term tag attachments, and their short- distance dispersal and affinity to return to shore each day provides a repeatable behavior that can be monitored and quantified with simple tagging technologies. Moreover, gentoo penguins in the Antarctic Peninsula region reside near the southern extent of their current breeding range and are currently expanding their range southward into areas once dominated by Adélie and chinstrap (P. antarctica) penguins (Lynch et al. 2008; Schofield et al. 2010). As a growing population in the region, estimates of the energetic requirements of gentoo penguins during winter is important for understanding ecological interactions and informing the management of marine fisheries, which currently catch the same food resources (Antarctic krill) consumed by gentoo penguins.

For studying demographics in relation to environmental variability, I used mark- recapture data collected since 1985 on a small population of Adélie penguins. Adélie penguins are one of the true Antarctic penguins, whose life history is directly associated with sea ice, a substrate used for molting and staging winter foraging bouts. As ice-dependent species, the

Adélie penguin is considered vulnerable to recent climate change in the region (Croxall et al.

2002; Ainley et al. 2010) because of the potential threat of loss of sea-ice habitat and the energetic expense associated with migrating to and from distant pack-ice regions. Importantly, the demography of Adélie penguins has been studied extensively throughout its range, which spans over 20 degrees of latitude, from the cold, dry continental environment of the Ross Sea

(77ºS) to the relatively warmer and wetter maritime environment of the South Sandwich archipelago in the eastern Scotia Sea (56ºS). Long-term studies of Adélie penguin

demography conducted across this latitudinal gradient enable intra-specific comparisons of population responses to contrasting environmental conditions over time and space. Such comparisons are useful to understand how species have responded to different environmental conditions and may provide insight for how population may respond to continued environmental change in their natal habitats.

Fig. 0-1: Location of study sites in the South Shetland Islands, Antarctica. The gentoo tagging data derive from breeding colonies at Cape Shirreff and at the Copacabana colony in Admiralty Bay. The demographic data for Adélie penguins is collected at the Copacabana colony in Admiralty Bay. .

REFERENCES

Ainley DG (2002) The Adélie penguin: Bellwether of climate change. Columbia University Press, New York

Ainley D, Russell J, Jenouvrier S, Woehler E, Lyver PO, Fraser WR, Kooyman GL (2010) Antarctic penguin response to habitat change as Earth’s troposphere reaches 2ºC above preindustrial levels. Ecol App 80:49-66

Atkinson A, Siegel V, Pakhomov E, Rothery P (2004) Long-term decline in krill stock and increase in salps within the Southern Ocean. Nature 432:100-103

Ballard G, Toniolo V, Ainley DG, Parkinson CL, Arrigo KR, Trathan PN (2010) Responding to climate change: Adélie penguins confront astronomical and ocean boundaries. Ecology 91:2056-2069

Ballerini T, Tavechia G, Olmastroni S, Pezzo F, Focardi S (2009) Nonlinear effects of winter sea ice on the survival probabilities of Adélie penguins. Oecologia 161:253-265

Barbraud C, Weimerskirch H (2006) Antarctic birds breed later in response to climate change. Proc Natl Acad Sci USA 103:6248-6251

Bevan RM, Butler PJ, Woakes AJ, Boyd IL (2002) The energetics of gentoo penguins (Pygoscelis papua) during the breeding season. Funct Ecol 16:175-190

Croxall JP (1984) Seabirds. In: Laws RM (ed) Antarctic Ecology, Volume 2. Academic Press, London

Croxall JP, Trathan PN, Murphy EJ (2002) Environmental change and Antarctic seabird populations. Science 297:1510-1514

Davis RW, Croxall JP, O’Connell MJ (1989) The reproductive energetics of gentoo (Pygoscelis papua) and macaroni (Eudyptes chrysolophus) penguins at South Georgia. J Anim Ecol 58:59-74

Devney CA, Short M, Congdon BC (2009) Sensitivity of tropical seabirds to El Niño precursors. Ecology 90:1175-1183

Emmerson L, Southwell C (2008) Sea ice cover and its influence on Adélie penguin reproductive performance. Ecology 89:2096-2102

Emmerson L, Southwell C (2011) Adélie penguin survival: age structure, temporal variability and environmental influences. Oecologia doi:10.1007/s00442-011-2044-7

Forcada J, Trathan PN, Reid K, Murphy EJ, Croxall JP (2006) Contrasting population changes in sympatric penguin species n association with climate warming. Global Change Biol 12:411- 423

Fraser WR, Trivelpiece WZ, Ainley DG, Trivelpiece SG (1992) Increases in Antarctic penguin populations: reduced competition with whales or a loss of sea ice due to environmental warming? Polar Biol 11:525-531

Green JA, Boyd IL, Woakes AJ, Warren NL, Butler PJ (2009) Evaluating the prudence of parents: daily energy expenditure throughout the annual cycle of a free-ranging bird, the macaroni penguin Eudyptes chrysolophus. J Avian Biol 40:529-538

Grosbois V, Thompson PM (2005) North Atlantic climate variation influences survival in adult fulmars. Oikos 109:273-290

Hinke JT, Salwicka K, Trivelpiece SG, Watters GM, Trivelpiece WZ (2007) Divergent responses of Pygoscelis penguins reveal a common environmental driver. Oecologia 153:845- 855

Jenouvrier S, Barbraud C, Weimerskirch H (2006) Sea ice affects the population dynamics of Adélie penguins in Terre Adélie. Polar Biol 29:413-423

Kokubun N, Takahashi A, Mori Y, Watanabe S, Shin H-C (2010) Comparison of diving behavior and foraging habitat use between chinstrap and gentoo penguins breeding in the South Shetland Islands, Antarctica. Mar Biol 157:811-825

Lebreton JD, Burnham KP, Clobert J, Anderson DR (1992) Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecol Monogr 62: 67-118

Lescroël A, Ridoux V, Bost CA (2004) Spatial and temporal variation in the diet of the gentoo penguin (Pygoscelis papua) at Kerguelen Islands. Polar Biol 27:206-216

Lynch HJ, Naveen R, Fagan WF (2008) Censuses of penguin, blue-eyed shag Phalacrocorax atriceps and southern giant petrel Macronectes giganteus populations on the Antarctic Peninsula, 2001-2007. Mar Ornithol 36:83-97

Meredith MP, King JC (2005) Rapid climate change in the ocean west of the Antarctic Peninsula during the second half of the 20th century. Geophys Res Lett 32, L19604, doi:10.1029/2005GL024042

Miller AK, Karnovsky NJ, Trivelpiece WZ (2009) Flexible foraging strategies of gentoo penguins Pygoscelis papua over 5 years in the South Shetland Islands, Antarctica. Mar Biol 156:2527-2537

Montes-Hugo M, Dooney SC, Ducklow HW, Fraser W, Martinson D, Stammerjohn SE, Schofield O (2009) Recent changes in phytoplankton communities associated with rapid regional climate change along the western Antarctic Peninsula. Science 323:1470-1473

Reiss CS, Cossio AM, Loeb V, Demer DA (2008) Variation in biomass of Antarctic krill (Euphausia superba) around the South Shetland Islands, 1996-2006. ICES J Mar Sci 65:497- 508

Sandvik H, Erikstad KE (2008) Seabird life histories and climatic fluctuations: a phylogenetic- comparative time series analysis of North Atlantic seabirds. Ecography 31:73-83

Schofield O, Ducklow HW, Martinson DG, Meredith MP, Moline MA, Fraser WR (2010) How do polar marine ecosystems respond to rapid climate change? Science 328:1520-1523

Stammerjohn SE, Martinson DG, Smith RC, Iannuzzi RA (2008) Sea ice in the western Antarctic Peninsula region: Spatio-temporal variability from ecological and climate change perspectives. Deep-Sea Res Part II 55:2041-2058

Tanton JL, Reid K, Croxall JP, Trathan PN (2004) Winter distribution and behaviour of gentoo penguins (Pygoscelis papua) at South Georgia. Polar Biol 27:299-303

Trathan PN, Croxall JP, Murphy EJ (1996) Dynamics of Antarctic penguin populations in relation to inter-annual variability in sea ice distribution. Polar Biol 16:321–330

Trivelpiece WZ, Trivelpiece SG, Volkman NJ (1987) Ecological segregation of Adélie, gentoo, and chinstrap penguins at King George Island, Antarctica. Ecology 68:351-361

Trivelpiece WZ, Buckelew S, Reiss C, Trivelpiece SG (2007) The winter distribution of chinstrap penguins from two breeding sites in the South Shetland Islands of Antarctica. Polar Biol 30:1231-1237

Trivelpiece WZ, Hinke JT, Miller AK, Reiss CS, Trivelpiece SG, Watters GM. (2011) Variability in krill biomass links harvesting and climate warming to penguin population sin Antarctica. Proc Natl Acad Sci USA 108:7625-7628

Vaughan DG, Marshall GJ, Connolley WM, Parkinson CL, Mulvaney R, Hodgson DA, King JC, Pudsey CJ, Turner J (2003) Recent rapid regional climate warming on the Antarctic Peninsula. Climate Change 60:243–274

Veit RR, Pyle P, McGowan JA (1996) Ocean warming and long-term change in pelagic bird abundance with the California current system. Mar Ecol Prog Ser 139:11-18

Vleck CM, Vleck D (2002) Physiological condition and reproductive consequences in Adélie penguins. Integr Comp Biol 42:76-83

Williams TD (1991) Foraging ecology and diet of gentoo penguins Pygoscelis papua at South Georgia during winter and an assessment of their winter prey consumption. Ibis 133:3-13

Wilson PR, Ainley DG, Nur N, Jacobs SS, Barton KJ, Ballard G, Comiso JC (2001) Adélie penguin population change in the pacific sector of Antarctica: relation to sea ice extent and the Antarctic Circumpolar Current. Mar Ecol Prog Ser 213:301-309

Wilson RP, Alvarrez B, Latorre L, Adelung D, Culik B, R Bannasch (1998) The movements of gentoo penguins Pygoscelis papua from Ardley Island, Antarctica. Polar Biol 19:407-413

Wilson RP (2010) Resource partitioning and niche hyper-volume overlap in free-living Pygoscelid penguins. Funct Ecol 24:646-657

CHAPTER 1: Daily activity and minimum food requirements during winter for gentoo

penguins (Pygoscelis papua) in the South Shetland Islands, Antarctica

12 13

ABSTRACT

Estimates of daily activity and consequent demand for food during winter are scarce for many polar seabirds, yet essential for assessing constraints on foraging effort, demand for food, and potential competition with local fisheries. We affixed archival temperature tags to gentoo penguins (Pygoscelis papua) from two colonies in the South Shetland Islands to measure the frequency, timing, and duration of foraging trips and to estimate minimum food requirements during winter. Foraging trip frequencies ranged from 0.85 to 1.0 trips day-1 and were positively correlated with day length. Early winter foraging trips more closely matched day length than late winter foraging trips. The data suggest that individuals maximize foraging time during the early winter period, likely to recover body mass following the breeding season and molt. The more attenuated response of foraging trip durations to increasing day length in late winter may be related to differences in local resource availability or individual behaviors prior to the upcoming breeding season. Minimum food requirements also exhibited a seasonal cycle with a mid-winter minimum. On average, minimum food requirements were estimated at

0.70 ± 0.12 kg day-1. Extrapolated to the regional population of gentoo penguins, winter food requirements by gentoo penguins were equivalent to roughly 33% of annual krill catches by commercial fisheries in the South Shetland Island region over the past decade. Current expansion of the gentoo population and the krill fishery in the southern Scotia Sea warrants continued monitoring of gentoo penguins during winter.

INTRODUCTION

Throughout the Antarctic Peninsula region, Pygoscelid penguin populations are undergoing rapid changes in size and distribution (e.g., Forcada et al. 2006; Hinke et al. 2007;

Lynch et al. 2008; Schofield et al. 2010; Trivelpiece et al. 2011). Climate change is thought to underpin these changes in populations through bottom-up effects on habitat suitability and prey availability, ultimately affecting survival and reproductive rates (Trathan et al. 1996;

Wilson et al. 2001; Jenouvrier et al. 2006). During this period of rapid population change, research on penguins has focused on foraging ecology, energetics, and reproductive biology during the short austral summer (e.g., Trivelpiece et al. 1987; Davis et al. 1989; Bevan et al.

2002; Lescroël et al. 2004; Miller et al. 2009; Kokubun et al. 2010; Wilson 2010). Inferences from studies conducted during the summer, however, suggest that winter periods largely determine annual survival and recruitment rates (Fraser et al. 1992; Trathan et al. 1996; Hinke et al. 2007), and, via variation in foraging success or prey availability, may also limit the ability of penguins to successfully reproduce in subsequent breeding seasons (e.g., Vleck and

Vleck 2002). Yet there are relatively few behavioral and ecological studies conducted during winter (but see Williams 1991; Wilson et al. 1998; Trivelpiece et al. 2007). In particular, it is unknown how foraging effort and demand for food varies seasonally. Thus, measurements of daily activity during winter are needed for assessing basic constraints on foraging effort, demand for food, and the potential for competition with commercial krill fisheries that operate during winter months in the Antarctic Peninsula region.

In the Antarctic Peninsula region, the gentoo penguin (Pygoscelis papua) population is growing and its range is expanding southward (Lynch et al. 2008). As a predator of

Antarctic krill (Euphausia superba) and fish resources, gentoo penguins in this region are also

a focus of ecosystem- based fisheries-monitoring efforts (CCAMLR 2004). The krill fishery has changed in several important ways recently: catches have increased and fishing effort has shifted south as warmer temperatures and declines in winter sea-ice have allowed fishing during winter in the South Orkney and South Shetland Islands (CCAMLR 2009). Gentoo penguins remain in these areas during winter and as such are increasingly important for monitoring fishery affects; descriptions of the distributions, behaviors, and energetics of gentoo penguins during winter merit study.

Previous studies of gentoo penguins during winter have measured their spatial distributions and diet compositions. Data collected with electronic tagging technologies, including archival, satellite, and telemetric tags, have demonstrated that gentoo penguins range more widely from natal colonies during winter than during the summer breeding season

(Wilson et al. 1998; Clausen and Pütz 2003; Tanton et al. 2004) but are not migratory in winter like their sympatric congeners, the chinstrap (P. antarctica) and Adélie (P. adeliae) penguins (e.g., Trivelpiece et al. 2007; Ballard et al. 2010). Rather, gentoo penguins generally retain the use of near-shore foraging habitats year-round (Clausen and Pütz 2003; Tanton et al.

2004). Diet studies during the non-breeding season (Jablonski 1985; Berrow et al. 1999; Coria et al. 2000) indicate that diet composition is similar year-round but also suggest that the relative proportions of diet items can vary.

Daily activity patterns during winter and the minimum food requirements necessary to support that activity are less well studied. Previous studies of gentoo penguins during winter at

South Georgia reported foraging trips during July that were undertaken during daylight hours and lasted between 5 and 8 h day-1 (Williams 1991; Williams et al. 1992a). Tanton et al.

(2004), using data from satellite tags, corroborated the diurnal nature of daily foraging trips during winter. However, daily foraging trip data for the duration of winter (defined here from

April through September) for gentoo penguins have not been reported previously. Existing estimates of food requirements during winter have been extrapolated from summer measurements of diet and average metabolic rates (e.g., Croxall and Prince 1987; Davis et al.

1989; Croll and Tershy 1998), or from observed diet masses and compositions during winter

(Williams 1991). Information on daily activity patterns, coupled with activity-specific metabolic rates, however, may improve estimates of food requirements during winter

(Williams 1991).

Here, we use temperature-recording archival tags to measure the basic activity patterns of gentoo penguins from two breeding colonies in the South Shetland Islands during the 2005, 2006, and 2008 austral winters. We first ask whether foraging trip frequencies, timing, and durations during winter differ across space and time. We then examine the relationship of trip durations with day length to assess how rapid changes in day length during winter affect foraging behavior. Finally, from the observations of daily activity, we estimate the minimum food requirements that are necessary to meet daily energy expenditures associated with the observed daily activity.

MATERIALS AND METHODS

Tag deployment

Temperature-recording archival tags were deployed on gentoo penguins from breeding colonies at Cape Shirreff, Livingston Island (62º 28’S, 60º 46’W) and Admiralty

Bay, King George Island (62º 10’S, 58º 30’W), South Shetland Islands, Antarctica (Fig. 1-1).

The temperature tags (iBKrill, manufactured by Alpha Mach, Inc., QC, Canada) were 25 x 15

x 8 mm, weighed 3.2 g dry, and were 1 g negatively buoyant in water. The tags were attached to post-molt adult gentoo penguins in late February and early March. Post-molt adults were identified by a bright black plumage that characterizes newly molted feathers and the absence of full tail feathers; juveniles have fully developed tail feathers at this time.

For a penguin with a body mass around 5 kg, carrying an externally attached tag for up to 9 months, represents a large burden. We sought to minimize detrimental impacts of external tagging with a minimal attachment method. A single tag was attached to the back of each penguin with thin layer of cyanoacrylate glue and secured to four or five underlying feathers with a single self-locking plastic cable tie. Attaching the tag to the short feathers on the back was deemed preferable to attachment on the flipper or leg for hydrodynamic and mobility reasons (Bannasch et al. 1994). However, placement on the back allows preening around the tag and such behavior can result in tag shedding (cf. Green et al. 2009). We observed loose connections of tags to the birds at the times of recovery (S. Trivelpiece, personal communication), indicating that the attachments were compromised during the over- winter deployment. Nonetheless, the attachment method was retained for all years of study to standardize potential tagging effects. All tagging procedures were conducted in the field and typically required less than 5 minutes per tag for the glue to adhere to the feathers, at which time the bird was released. To further minimize impact on study birds, no other identifying marks were used.

In total, 200 temperature tags were deployed over the 3 years of this study. We successfully recovered twenty tags in the following springs, up to 9 months later. Of the twenty tags recovered, eleven failed to record data. Thus, nine tags contained useable data. Of the nine tags, four were recovered at Admiralty Bay in 2005, two each were recovered at

Admiralty Bay and Cape Shirreff in 2006, and one was recovered at Admiralty Bay in 2008.

The ambient temperature experienced by each bird was recorded at 90-min intervals for the duration of deployment. This sample interval was chosen as a trade-off, given the memory capacity of the tag, between the desire for high-frequency sampling to resolve daily movement between land and sea and the need for long-term data records that covered the full winter period. With a 90-min interval, the memory capacity of the tag was sufficient to record approximately 4 months of data. To collect data from the full winter period, tags were programmed in two batches to record data either during early winter (April through July) or late winter (July through September). Of the nine tags recovered with data, four recorded temperatures in early winter and five recorded temperatures in late winter (Table 1-1).

Criteria for the classification of over-winter data

The classification of temperature data between time spent at sea and time spent on land or ice is possible because of the near-constant temperature of surface waters in the polar ocean relative to the more variable ambient air temperatures. In the South Shetland Islands, coastal water temperatures in winter range from roughly -2 to +1ºC (Barnes et al. 2006), providing a baseline against which the movement of birds to and from the water can be determined. Specifically, we assume that temperatures recorded by the bird will match water temperatures while the bird is foraging, but show higher variability associated with ambient air temperatures otherwise.

We conducted a test study in February 2006 to verify this assumption and to identify specific characteristics of the temperature records from actively foraging penguins that could be used to classify the over-winter data. For the test study, two gentoo penguins were fitted with temperature tags using the method described previously, two tags were attached to

concrete anchors and submerged in the subtidal zone, and two tags were deployed on an exposed weather tower. All tags were retrieved after 3 days, and their temperature data were examined.

The temperature records from both birds in the test study showed U-shaped diel patterns (Fig. 1-2). The flat valleys of the U-shaped temperature records from the birds closely corresponded with water temperatures. We interpret these U-shaped patterns in temperature records as foraging trips undertaken in the relatively constant thermal environment of the coastal ocean. This assumption is supported by three main characteristics of the data. First, within the flat valley of a U-shaped profile, temperatures recorded by the birds matched water temperatures with a mean difference of 0.32 ± 0.13ºC (t9 = 4.99, P < 0.01). The positive bias of the bird temperature relative to ambient water temperature suggests that some thermal contamination from the bird is present. However, the bias can be ignored for present purposes due to the consistency of consecutive temperature readings within a foraging trip. Second, each foraging trip is bounded by abrupt changes in temperature immediately preceding and following the foraging trip. The changes in temperatures at the times of trip initiation and trip completion during the test study ranged from 9.3 to 1.0ºC. We interpret these abrupt changes to be indicative of movement from a variable ambient air temperature environment into a more constant water temperature environment. We use the minimum shift between consecutive measurements of 1.0ºC to identify start and end times of a foraging ship. Third, within the foraging trips, temperature measurements had a maximum range of 1.5ºC, and consecutive temperature measurements recorded by the birds had a mean difference of 0.30 ±

0.10ºC. Temperature records outside the foraging trip exhibited a larger maximum range

(6.1ºC) and had a larger mean difference between successive measurements (1.54 ± 0.46ºC; t37

= 5.23, P < 0.01). Thus, small changes between successive temperature measurements in the

range of coastal water temperature (-2 to +1ºC), bounded by abrupt changes in temperature greater than 1ºC, provide the key delimiters for identifying foraging trips in these data. The final classification of the winter time series was conducted visually on a point-by-point basis, guided by the above constraints.

From this classification of individual temperature observations, the timing of trip initiation, trip completion, and the duration of each foraging trip were measured. Each foraging trip was further classified as either a single trip that occurred within a single day

(single), one of multiple trips that occurred within a single day (multiple), or a trip that lasted past local midnight (overnight). We also recorded the number and duration of periods of time without a foraging trip to examine variation in the frequency of no-trip periods during winter.

Statistical analysis

The data on the timing and duration of foraging trips can be viewed as a hierarchy with four levels: individuals, periods of deployment (period), deployment sites (sites), and years. However, because of the high tag loss rates, the hierarchy is sparsely populated at all levels below year, thus limiting the estimation of potential hierarchical and interactive effects.

Our analysis therefore examines whether differences in the data are evident at each level. We test the null hypotheses of no difference in the proportion of trip types and trip frequencies within each level with Chi-square tests. We use ANOVA to test null hypotheses of no differences in mean trip duration within levels. Mean values and 95% confidence intervals are reported in the text unless otherwise noted. Statistical significance was judged at P ≤ 0.05.

To explain temporal variation in trip durations, we examine a set of additive models that consider each level (individuals, period, site, and year) as an independent grouping factor.

Since light availability limits foraging activity for visual predators like gentoo penguins, we include day length as the primary covariate in all models. Day length data for the South

Shetland Islands area were obtained online from the US Naval Observatory

(http://aa.usno.navy.mil/data/). We fit generalized linear models using R statistical software (R

2010), and models were ranked with Aikaike’s Information Criteria corrected for sample size

(AICc) to identify the most parsimonious model (Burnham and Anderson 2002). To further distinguish between competing models, we calculate the ratio of AIC weights of competing models with the most parsimonious model. The ratio of AIC weights provides an estimate of how much more support in the data there is for the best model relative to the alternatives

(Hobbs and Hilborn 2006).

Minimum food requirements

We generalized the approach used by Croll and Tershy (1998) to estimate minimum food requirements (MFR) from the daily activity patterns estimated from the temperature data described earlier. Daily MFR were calculated as the sum of daily energy expenditures incurred during time spent in the water and time spent out of water multiplied by the ratio of the proportion of a particular prey item in the diet to the energy density of that prey item:

 Ti * MRi *W  MFR αk * Eq. 1 d = ∑  ,ki  EDk δ 

At-sea or on-land activity is specified by i and prey type by k; T represents the total time spent each day on each activity i; MR is the average metabolic rate for activity i; W is the mass of an average penguin; δ is the metabolic efficiency, or the proportion of energy consumed that is

assimilated by the bird and not lost as waste; α is the proportion of prey item k in the diet, and

ED is the energy density of prey item k (Table 1-2).

Gentoo-specific parameter values for activity-specific metabolic rates and assimilation efficiency were collected from published studies. We used activity-specific metabolic rates for at-sea activity and resting activity as reported by Bevan et al. (2002). Metabolic efficiency was taken from a laboratory study of gentoo penguins by Davis et al. (1989). Diet compositions can vary in time, but studies of gentoo diets during winter unanimously report that krill and fish are the dominant prey items (Jablonski 1985; Williams 1991; Berrow et al. 1999; Coria et al. 2000). We assumed a mixed diet of 55% krill and 45% fish, which is similar to late winter diets estimated from stable isotope analyses of gentoo penguin egg shells (Polito et al. 2011).

This diet distribution falls within the range of diet compositions reported during winter at

King George Island (Jablonski 1985), but which has less krill than diets collected during the chick-rearing stage (Miller et al. 2009). To assess how different diet compositions affect the

MFR calculation, we conduct a sensitivity analysis. Energy densities for krill and fish were averaged from published reports (see references in Table 1-2).

Body mass strongly affects metabolic rate, and changes in mass over the winter period should affect daily energy expenditure and the daily MFR (Green et al. 2009). However, due to the one-month lapse between tagging and the initiation of recording by the tag, we cannot assess changes in body mass over the recorded winter period. We therefore assume a constant mass for the purpose of the MFR calculation, but examine the effect of mass on the MFR calculation with a sensitivity analysis. We used data on the mass of breeding adult gentoo penguins, measured with spring scales, collected during the chick-rearing phase at Admiralty

Bay from 1988 to 2010 (WZT, unpublished data) to estimate mean adult mass.

Assessing uncertainty in the MFR calculation

To account for uncertainty in the parameter estimates required in Eq. 1, including diet composition and adult mass, we conducted a parametric bootstrap to estimate 95% confidence intervals for the daily estimates of MFR. For metabolic rate, assimilation efficiency, and prey energy density parameters, we defined normal distributions with means taken from their respective published studies and standard deviations that were approximated to encompass the range of values reported in the literature (Table 1-2). Because there is no indication of a central tendency for diet proportions, αkrill was assumed to be uniformly distributed between

0.2 and 0.9 for krill prey and 1 – αkrill for fish prey, based on reported diet compositions (Table

1-2). For estimating the effect of body mass on MFR, we allowed body mass to vary according to the mean and standard deviation from weights collected at the study site, as defined previously. For overall estimates of 95% confidence intervals, all parameters were drawn independently from the distributions defined in Table 1-1, and the MFR was iteratively calculated 1 x 105 times. To assess the range of MFR estimates that could result from alternative diets, we calculated the MFR as earlier, but with a fixed diet of 90% krill or 20% krill. Likewise, we calculated the MFR with fixed adult masses at the 5th and 95th quantiles of the observed data.

RESULTS

Foraging trips

In each winter, the study birds exhibited foraging trips that occurred mainly during daylight hours. On average, one continuous foraging trip within a single day accounted for

91% of all trips observed (Fig. 1-3a). Days with multiple, distinct foraging trips (up to 3 per day) occurred 8% of the time. Roughly 1% of foraging trips lasted overnight (i.e., past local midnight).

Overall trip frequencies ranged from 0.85 to 1.0 trips day-1 (Table 1-1). Heterogeneity

2 among individuals was evident ( χ = 43.5, df = 8, P < 0.01), with the lowest trip frequencies arising from an individual that was deployed in late winter (Table 1-1). On average, however,

2 2 average trip frequency did not differ between years ( χ = 3.02, df = 2, P = 0.22), sites ( χ =

2 0.15, df = 1, P = 0.69), or periods of deployment ( χ = 2.25, df = 1, P = 0.13).

Days without foraging trips occurred at different frequencies among individuals

2 throughout the winter ( χ = 17.2, df = 8, P = 0.02), arising from fewer days with skipped

2 foraging trips in April and May relative to later months ( χ = 16.5, df = 5, P = 0.006; Fig. 1-

2 3b). The number of days without a foraging trip did not vary among years ( χ = 0.42, df = 2,

2 P = 0.81) or sites ( χ = 0, df = 1, P = 1), however. The longest consecutive period without a foraging trip was 4 days, which was observed during July for three different birds. In general, the frequency of occurrence of longer periods of time (3–4 days) without foraging trips was higher in June and July than other months (Fig. 1-3b).

Over the period of deployment for each individual, average trip durations ranged from

6 to 8 h (Table 1-1), with longer average foraging trip durations occurring at Cape Shirreff relative to Admiralty Bay (F1,888 = 46.8, P < 0.01). When the Cape Shirreff data were excluded to examine within-site differences at Admiralty Bay, there was no evidence for differences in trip duration that arose from the period of deployment (F1,727 = 1.84, P = 0.17) or the year of deployment (F2,726 = 1.13, P = 0.32).

On average, mean daily foraging trip durations were positively correlated with day length (r = 0.63, Fig. 1-4a). A comparison of models that included day length and additive combinations of each level indicates that day length, period, and site of deployment provided the most parsimonious explanation of the data; the best model had more than 2.5 times the support in the data over the next best model (Table 1-3). According to the best model, foraging trip durations increased (mean ± SE) 0.35 ± 0.03 h with each hourly increase in day length (t1 = 9.524, P < 0.01). Late winter trips, however, were shorter, on average, than early winter trips by 0.48 ± 0.18 h (t1 = -2.16, P = 0.01), but birds from Cape Shirreff had trips, on average, 1.17 ± 0.21 h (t1 = 5.35, P < 0.01) longer than birds from Admiralty Bay. A smoothing spline fit to the data suggested that early winter trips more closely matched day length than late winter foraging trips (Fig. 1-4a), suggesting that additive models (i.e., common slopes but different intercepts) may not best represent the data. However, because of missing data in the hierarchy, interactive effects of site and period of deployment are not estimable and we cannot test directly the null hypothesis of no difference in slopes between the two time periods. Indirectly, however, separate linear models of trip duration regressed only on day length for the two periods of deployment result in non-overlapping 95% confidence intervals for the slope coefficients (early winter: 0.25 ± 0.05, t1 = 9.5, P < 0.01;

late winter 0.08 ± 0.04, t1 = 4.3, P < 0.01), suggesting that the relationship between day length and late winter foraging trip durations is weaker.

Throughout the winter, the mean time of trip initiation (if multiple trips occurred on a given day, only the first trip was considered) was linearly related to sunrise (F1,180 = 163, P <

0.01; Fig. 1-4b). However, the time of trip completion (if multiple trips occurred on a given day, only the last trip is considered) did not vary with the time of sunset throughout the winter

(F1,180 = 1.86, P = 0.17). On average, birds routinely exited the water by 17:30 ± 0.17 h.

Minimum food requirement

As total daily time at sea diminished during mid-winter, so did the amount of food required to meet daily energy expenditures (Fig. 1-5). On average, based on a diet of 45% fish and 55% krill, the MFR was roughly 0.75 kg day-1 in April and dropped roughly 10% to 0.68 kg day-1 in late June. Over the full winter period, the average daily MFR was estimated at 0.70

± 0.12 kg day-1.

The estimates for MFR were largely insensitive to changing the diet composition. For example, a diet of 90% krill required 0.72 ± 0.13 kg day-1, while a diet of 20% krill required

0.69 ± 0.12 kg day-1. Adult body mass had a larger effect on the calculation of MFR. Using the

5th and 95th percentile of adult masses recorded at the study site, and assuming a mixed diet of fish (45%) and krill (55%), the average daily MFR for the full winter period ranged from

0.59 ± 0.06 kg day-1 for 4.5 kg adults up to 0.84 ± 0.08 kg day-1 for 6.3 kg adults.

DISCUSSION

Using archival temperature-recording tags, we estimated daily activity patterns and

MFR for gentoo penguins for a full winter period from two breeding colonies in the South

Shetland Islands. In general, foraging trips during winter tracked the cycle of day length and exhibited a regular diel period with diurnal foraging patterns. Temporal variation in trip durations was most strongly related to day length, but there is evidence that foraging trips differed by period and site of deployment. The estimated MFR exhibited a cyclic pattern with daily energy expenditures proportional to foraging trip durations. Together, these results expand the spatial and temporal scope of behavioral data on gentoo penguins during winter.

Important assumptions

One important assumption of this study is that a 90-min sampling interval sufficiently records foraging trip timing and duration. To address whether the chosen sampling rate biased the estimates of daily activity during winter, we compare our results with published data on foraging trip durations based on a higher-frequency sampling protocols reported by Williams

(1991) and Williams et al. (1992a). Data from continuous radio telemetry sampling (Williams

1991) and from time-depth recorders that sampled on 15-s intervals (Williams et al. 1992a) indicated that average foraging trip durations for gentoo penguins at South Georgia during

July ranged from 5.3 to 8.2 h, while trip frequencies ranged from 0.63 to 0.90 trips day-1. In the present study, trip durations during July averaged 6.3 h and trip frequency was 0.92 trips day-1. Given similar results from such disparate sampling intervals, we conclude that a 90-min interval sufficiently captures the relevant information on trip timing and duration.

A second assumption is that data from nine birds sufficiently represent the population of gentoos for which we seek to make inference. Moreover, we are compelled to comment on the high rate of tag shedding observed in this study and whether the data collected were in any way biased by the low return rate. First, the loss rate of tags was high when compared with other studies on penguins of similar duration (e.g., Green et al. 2005; Ballard et al. 2010). We attribute the observed loss rate to our tagging method, namely the position on the back and the use of a thin layer of glue were chosen to allow tag shedding while risking low recovery rate.

Secondly, the overall size of the temperature tags (25 x 15 mm, 3.2 g) is small compared with satellite (95 x 42 mm, 85 g; Tanton et al. 2004) and radio telemetry (80 x 18 mm, 35 g;

Williams 1991) previously used to measure winter foraging trips of gentoo penguins. The similarity of trip durations (discussed earlier), however, measured with differently sized technologies suggests that instrumentation biases on foraging trip data are at least similar across platforms. Finally, without comparable data from non-instrumented animals during the winter, true instrumentation effects on behavior necessarily remain unknown. Thus, while larger sample sizes can simplify generalizations to a population level, we conclude that the data from our small sample size are not unduly biased by low retention rates and are sufficient for the examination of daily behavior during winter.

Characteristics and constraints of winter foraging trips

Gentoo penguins are typically regarded as diurnal foragers dependent on light availability for successful foraging (Jablonski 1985; Williams et al. 1992b; Miller et al. 2009).

The seasonal cycle in foraging trip durations during winter, the predominance of single, diurnal foraging trips, and the general correspondence of trip duration with day length support

that characterization. However, different responses to day length in early versus later winter suggest that other processes also operate to constrain foraging trip durations during winter.

Across individuals from 3 years, the study birds exhibited a steeper response of daily foraging time to changing day length in early winter relative to later winter. This difference suggests that gentoo penguins may maximize the time spent foraging during early winter. Such a constraint may derive from two, non-exclusive reasons. First, the annual molt occurs from late

February through early March at our study sites. The molt, a full replacement of the plumage during which time gentoo penguins remain on land and do not feed, typically results in the highest rates of mass loss during the year (Williams 1995). It therefore seems plausible that the early winter period is used to replenish energy stores lost during the molt. Recovery of body mass is critical for determining the ability of penguins to survive the winter and initiate the next breeding cycle (e.g., Vleck and Vleck 2002). Consequently, post-molt individuals might be expected to spend as much time foraging as day length allowed during early winter to rapidly recover body mass and restock fat reserves. The observations that trip completion typically occurred after sunset also support this assertion. Since foraging efforts are limited by darkness, we infer that the study birds likely foraged up to sunset and returned to land after nightfall. This hypothesis could be tested directly in the future with detailed diving records from over-winter deployments of time-depth recorders.

Alternatively, maximizing foraging trip duration relative to day length may arise if local food availability is reduced. Reductions in local food resources can arise as a consequence of concentrated foraging efforts near natal colonies during the breeding season

(i.e., the halo effect; Ashmole 1963), resulting in density-dependent increases in foraging trip durations (Ballance et al. 2009). During the breeding season at the study colonies, local fish and krill resources are also consumed by Adélie and chinstrap penguins, and numerous

pinniped and cetacean species. After the breeding season, gentoo penguins typically remain near natal colonies, while the other penguin and mammal competitors migrate to alternative winter feeding grounds. Thus, maximizing foraging trip durations relative to available day length may reflect locally suboptimal foraging conditions that result from the heavy predation that occurred during the breeding season. Whether from the need to recover body mass and energy stores after the breeding season and molt, or because of decreased food availability, the diminishing time available for foraging as day length decreases in the early winter acts as a physical constraint on gentoo penguins foraging trip durations.

During the latter part of winter, however, foraging trip durations exhibited a more attenuated response to increasing day length. The decoupling of foraging trip durations from day length in late winter could arise from improved local foraging conditions in late winter or changes in behavior as the breeding season nears. We have no data to examine changing prey density in late winter, but we do expect over-winter mortality to reduce the local population size of gentoo penguins. Specifically, despite relatively high chick production rates at the study colonies (1.24 chicks crèched nest-1, Hinke et al. 2007), which increases the effective local population size by up to 62%, return rates of gentoo penguins after the first winter of life are typically low (Hinke et al. 2007), indicative of high mortality rates during winter.

Consequently, individuals surviving to the late winter may encounter fewer competitors and find sufficient prey to satisfy daily energy expenditures in less time than day length allows.

Decreased foraging trip durations in the late winter could also arise from increased on- land activity associated with courtship as the breeding season nears. Devoting more time to courtship, nest building, or nest-site defense necessarily limits the time available for foraging at sea. Thus, observations of longer late winter foraging trips at Cape Shirreff relative to

Admiralty Bay may be explained by the proximity to the breeding season at each colony;

gentoo penguins at Cape Shirreff initiate clutches up to 3 weeks later than at Admiralty Bay

(Lynch et al. 2009). Data from over-winter observations of macaroni penguins also suggest that the amount of time spent foraging in late winter may be affected by breeding behaviors.

Specifically, macaroni penguins at Bird Island, South Georgia, foraged longer during the late winter than at any other time of the year, maximizing the time spent foraging between sunrise and sunset (Green et al. 2005). The contrasting patterns of foraging time in late winter exhibited by macaroni and gentoo penguins may be related to the presence or absence of the breeding fast. Among macaroni penguins, the breeding fast and the consequent need to maximize energy stores prior to breeding may necessitate longer foraging times in late winter.

In contrast, the gentoo penguin typically exhibits continued foraging throughout the pre- breeding period and therefore would have less need to maximize foraging times in late winter.

However, to better understand the decoupling of foraging trip durations from day length during the late winter period among gentoo penguins, more detailed assessments of daily behaviors, including behavioral observations on land during winter months, will be necessary.

The timing and duration of winter foraging trips generally matches foraging patterns observed during the summer. Miller et al. (2009) reported that foraging trips during the breeding season at Cape Shirreff occurred daily with trip durations that varied interannually from 7.6 to 12 h (Miller et al. 2009). Kokubun et al. (2010) reported that gentoo foraging trips from Maxwell Bay, King George Island, typically lasted 9.2 ± 3.7 h, and all trips ended prior to or during twilight. However, an exception to such diurnal foraging behavior was recently reported from at New Island, Falkland Islands (Masello et al. 2010). There, gentoo penguins were engaged in frequent dives to depths <40 m between the hours of 20:00 and 03:00. Thus, it seems possible that gentoo penguins can forage under more light-limited conditions than reported previously at higher latitudes. The shorter duration of day length during winter would

suggest that night time dives may be relatively common if gentoo penguins can successfully forage in darkness. However, we find no such evidence in our data; the low frequency of nighttime trips (1%) and positive correlations of trip duration with day length (r = 0.63) suggest that gentoo foraging during winter is essentially diurnal. Nonetheless, our data indicate only that some individuals remained at sea overnight; we cannot examine whether such over-night excursions included foraging effort. Future over-winter deployments of depth recorders will be useful to test this assumption.

Comparison of MFR and krill fishery catches

Estimating the food requirements of penguins remains important for understanding their ecological role in their food web (Trivelpiece et al. 1987; Croxall et al. 1999; Boyd 2002;

Ainley et al. 2006) and the potential for competition with industrial fisheries (e.g., Croll and

Tershy 1998; Clausen and Pütz 2003). Here, the daily energy requirements of the st udy birds varied in proportion to the amount of time spent at sea over the winter, averagi ng 0.70 ± 0.12 kg day-1 over the full winter period. Jablonski (1985) reported winter diet masses at King

George Island between 0.61 and 1.04 kg day-1, which are as high as or higher than diet measured during the breeding season (Jablonski 1985; Volkman et al. 1980). Hill et al. (2007) estimated per capita winter consumption rate of gentoo penguins in the South She tland Islands area to range from 0.74 to 1.38 kg day-1. Thus, the MFR reported here overlaps the rang e of previously observed or estimated diets at King George Island. Moreover, in studies with sufficient temporal coverage during winter (Williams 1991; Berrow et al. 1999), there is evidence for a seasonal cycle in diet mass with a mid-winter minimum, which also agrees with estimates based on the activity-specific method used here.

We note that the estimates of winter consumption were most sensitive to our assumptions about penguin mass. Given that body mass following the molt is among the lowest during the annual cycle, food requirements during winter, especially during the post- molt period, are likely to be higher than the estimated MFR. However, in the absence of data on the rate and magnitude of changes in body mass during winter, calculations based on constant mass provide a first approximation of winter food requirements for gentoo penguins in the South Shetland Islands.

Estimates of individual food requirements, specifically the proportion of the diet that is krill, can be extrapolated to the population level and compared with recent commercial fishery catches of Antarctic krill in the South Shetland Island region. Such comparisons are useful starting points for understanding the increasing predation pressure of the growing gentoo penguin population in the Scotia Sea and the potential for fishery-induced effects on gentoo penguins. Given breeding populations of roughly 1,600 and 5,600 gentoo penguins at

Cape Shirreff and Admiralty Bay, respectively (Hinke et al. 2007), a diet of 55% krill, and assuming that breeding adults represent 64% of the total population (Williams 1995), gentoo penguins from these two small study sites (ca. 11,250 individuals) would consume roughly

800 tons of krill during the winter, equivalent to 2.5% of recent krill catches by the krill fishery around the South Shetland Islands (1999–2008 average catch in statistical area 48.1:

3.34 x 104 tons; CCAMLR 2009). In the wider Antarctic Peninsula region, recent censuses suggest there are an additional 84,000 gentoo penguins in this area (Lynch et al. 2008).

Assuming similar food requirements, total krill consumption by the gentoo population during winter may be 1.0 x 104 tons of krill, roughly 33% of recent annual krill catches in the South

Shetland Island region. At present, the krill fishery is poised to expand. In 2010, for example, the krill fishery reported catches around the South Shetland Islands in excess of 1.5 x 104 tons,

roughly five times the average catch during the preceding decade in this region (CCAMLR

2010). This catch triggered the first-ever precautionary closure of the fishery in this region in early October. The ecological and behavioral effects of such higher catches in the area where gentoo penguins forage year-round remains unknown. However, a growing population of gentoo penguins in this region will demand more food from the ecosystem. Given the observed constraints of day length on foraging time, increased catches by the fishery, particularly during winter months, are likely to increase competition for krill between the fishery and the penguins. Croll and Tershy (1998) noted the potential for competition with the fishery over a decade ago, when total catches and gentoo populations were lower. Clearly, continued monitoring of winter behavior and food consumption remains a research priority.

Chapter 1, in full, was published in Polar Biology: Hinke JT, Trivelpiece WZ (2011).

Daily activity and minimum food requirements of gentoo penguins (Pygoscelis papua) in the

South Shetland Islands, Antarctica. Polar Biology 34:1579-1590. The dissertation author was the primary investigator and author of this paper.

ACKNOWLEDGEMENTS

Many thanks to S. Trivelpiece, D. Loomis, M. Polito, S. Agius, S. Rogers, R. Orben,

A. Miller, and E. Leung for help with tag deployment and recovery. Solar data were provided by the US Naval Observatory. Comments by G. Kooyman, J. Graham, G. Watters, J. Barlow,

J. Greene, and 3 anonymous reviewers improved earlier versions of this manuscript. We gratefully acknowledge financial support from the US Antarctic Marine Living Resources program, the Lenfest Oceans Program at the Pew Charitable Trusts, and the National Science

Foundation (grant #1016936 to WZT). This research was conducted in accordance with national and international guidelines concerning the use of animals in research under permits from the National Science Foundation to WZT. Use of brand name does not imply endorsement by the National Marine Fisheries Service.

Table 1-1: Summary of tag deployments and foraging trip data for gentoo penguins during winter.

N days N Trip Mean trip Site Year Period Dates N days N trips without overnight frequency duration trips trips (trips day-1) (hr) Admiralty 2005 Early 3 Apr – 7 Aug 127 127 6 2 1.00 6.8±0.43 Bay 2005 Early 3 Apr – 7 Aug 127 119 13 2 0.94 6.42±0.56 2005 Late 11 Jul – 30 Sep 82 70 14 2 0.85 7.26±1.17 2005 Late 15 Jul – 30 Sep 78 77 7 0 0.99 6.91±0.65 2006 Early 2 Apr - 6 Aug 127 126 5 1 0.99 6.62±0.43 2006 Early 28 Apr – 1 Sep 127 114 15 0 0.90 5.98±0.35 2008 Late 16 Jun – 30 Sep 107 105 14 4 0.98 6.47±0.37 Cape 2006 Late 8 Jul – 30 Sep 85 84 5 1 0.99 7.59±0.53 Shirreff 2006 Late 5 Jul – 30 Sep 88 78 10 0 0.89 8.02±0.5

Table 1-2: Parameters used to estimate the daily minimum food requirement. A single dash (-) denotes that mean and standard deviation values were not used in the analysis; rather, we used uniform distributions bounded by the given ranges to generate these parameters. NA refers to dimensionless parameters.

Parameter Description Mean (SD) Range Units

Resting metabolic MR 4.14 (0.1)a 3.6-4.7 W/kg land rate on land Metabolic rate at MR 8.58 (0.25)a 7.6-9.7 W/kg sea sea

W Adult mass 5.33 (0.56)b 3.7-7.0 kg

Metabolic 0.74 (0.01)c 0.70-0.77 NA δ efficiency Proportion of krill - 0.2-0.9d,e,f,g NA α krill in the diet Proportion of fish - 1- NA α fish in the diet α krill Energy density of ED 4.57 (0.23)c,h,i,j 3.7-5.5 kJ/g krill krill (wet mass) Energy density of ED 5.0 (0.38)k,l,m 3.5-7 kJ/g fish fish (wet mass) a Bevan et al. 2002, b Trivelpiece, unpublished data, c Davis et al. (1989), d Jablonski (1985), e Croll and Tershey (1998), f Volkman et al.(1980), g Williams (1991), h Costa et al. (1989), i Nagy and Obst (1992), j Janes and Chappell (1995), k Boyd (2002), l Barrera- Oro (2002), m Cherel and Ridoux (1992).

Table 1-3: The top five models of foraging trip durations compared to a base model with only day length, ranked by AICc score. The null deviance for all models was 4828 and the lowest AIC value was 3906.5.

Residual AIC Model specification DF ΔAIC AIC w deviance c w Ratio Day length+Period+Site 4152.1 4 0 0.59 Day length+ID 4104.9 10 2.005 0.22 2.73 Day length+Period+Site+Year 4147.4 6 3.042 0.13 4.58 Day length+Site 4184.2 3 4.829 0.05 11.18 Day lenth+Year+Site 4179.1 5 7.789 0.01 49.13 Day length 4286.4 2 24.29 0.00 -

Fig. 1-1: Map of the study area in the South Shetland Islands, Antarctica. Study sites were located at Admiralty Bay, King George Island and Cape Shirreff, Livingston Island. The dashed box in the insert identifies the local study region.

Fig. 1-2: Comparison of temperature records recorded by tags attached to two penguins (solid lines in panels a and b), the ambient air temperature (dotted line), and the water temperature (dashed line) during the test study. In each panel, temperature points identified as belonging to a foraging trip are marked with an open square (฀).

A 1

0.75

0.5

overnight 0.25 multiple Proportion of trips of Proportion single 0 1 2 3 4 5 6 7 8 9 Bird B 0.20 4-day 3-day

0.15 19 2-day 1-day 30

0.10 17 12

6 0.05

Proportion of days of Proportion 1 0.00 Apr May Jun Jul Aug Sep Month

Fig. 1-3: Summary of trip types during the winter. A) Proportion of all trips owing to single trips within a single day (single), multiple trips within a single day (multiple) or overnight trips (overnight). B) Cumulative proportion of days without foraging trips owing to no-trip durations of differing lengths. Proportions are calculated relative to the total number of observations made in each month. The total number of days with no foraging trip is printed above each bar.

14 A 12 10 8 6 4 Foraging trip trip Foraging (hrs) duration 2 0 B 21:00 18:00 15:00 12:00 9:00 Time of day of Time 6:00 3:00 0:00 Apr May Jun Jul Aug Sep Oct

Fig. 1-4: Duration and timing of foraging trips by gentoo penguin from April through September. Each point is the average from all trips undertaken on a given day across all years. A) Average foraging trip duration (○) relative to local day length (solid line). A smoothing spline (dashed line) indicates the trend in the data. B) Time of trip initiation (○) and trip completion (●) relative to sunrise and sunset (solid lines), respectively. Smoothing splines for time of trip initiation and trip completion (dashed lines) indicate the trend in the data.

1.0

0.9

0.8

MFR (kg) 0.7

0.6

0.5

Apr May Jun Jul Aug Sep Oct Month

Fig. 1-5: Estimated average daily minimum food requirement (MFR) necessary to meet daily energy expenditures throughout the winter. Smoothing splines identify the 50th (solid line), 5th, and 95th quantiles (dashed lines) from the bootstrap procedure.

REFERENCES

Ainley DG, Ballard G, Dugger KM (2006) Competition among penguins and cetaceans reveals trophic cascades in the western Ross Sea, Antarctica. Ecology 87:2080-2093

Ashmole NP (1963) The regulation of numbers of tropical oceanic birds. Ibis 103:458- 473

Ballance LT, Ainley DG, Ballard G, Barton K (2009) An energetic correlate between colony size and foraging effort in seabirds, an example of the Adélie penguin Pygoscelis adeliae. J Avian Biol 40:279-288

Ballard G, Toniolo V, Ainley DG, Parkinson CL, Arrigo KR, Trathan PN (2010) Responding to climate change: Adélie penguins confront astronomical and ocean boundaries. Ecology 91:2056-2069

Bannasch R, Wilson RP, Culik B (1994) Hydrodynamic aspects of design and attachment of a back-mounted device in penguins. J Exp Biol 194:83-96

Barnes DKA, Fuentes V, Clarke A, Schloss IR, Wallace MI (2006) Spatial and temporal variation in shallow seawater temperatures around Antarctica. Deep-Sea Res Pt II 53:853-865

Barrera-Oro E (2002) The role of fish in the Antarctic marine food web: differences between inshore and offshore waters in the Southern Scotia Arc and west Antarctic Peninsula. Antarct Sci 14:293-309

Berrow SD, Taylor RI, Murray AWA (1999) Influence of sampling protocol on diet determination of gentoo penguins Pygoscelis papua and Antarctic fur seals Arctocephalus gazella. Polar Biol 22:156-163

Bevan RM, Butler PJ, Woakes AJ, Boyd IL (2002) The energetics of gentoo penguins (Pygoscelis papua) during the breeding season. Funct Ecol 16:175-190

Boyd IL (2002) Estimating food consumption of marine predators: Antarctic fur seals and macaroni penguins. J Appl Ecol 39:103-119

Burnham KP, Anderson DR (2002) Model selection and multimodel inference: A practical Information-Theoretic approach. Springer-Verlag, New York

CCAMLR (2004) CCAMLR Ecosystem Monitoring Program Standard Methods. CCAMLR, Hobart

CCAMLR (2009) Statistical Bulletin, Volume 21. CCAMLR, Hobart

CCAMLR (2010) Report of the Twenty-Ninth Meeting of the Scientific Committee. 25- 29 October 2010. CCAMLR, Hobart

Cherel Y, Ridoux V (1992) Prey species and nutritive value of food fed during summer to king penguin Aptenodytes patagonica chicks at Possession Island, Crozet Archipelago. Ibis 134:118-127

Clausen A, Pütz K (2003) Winter diet and foraging range of gentoo penguins (Pygoscelis papua) from Kidney Cove, Falkland Islands. Polar Biol 26:32-40

Coria N, Libertelli M, Casaux R, Darrieu C (2000) Inter-annual variation in the autumn diet of the gentoo penguin at Laurie Island, Antarctica. Waterbirds 23:511-517

Costa DP, Croxall JP, Duck CD (1989) Foraging energetics of Antarctic fur seals in relation to changes in prey availability. Ecology 70:596-606

Croll DA, Tershy BR (1998) Penguins, fur seals, and fishing: prey requirements and potential competition in the South Shetland Islands, Antarctica. Polar Biol 19:365-374

Croxall JP, Prince PA (1987) Seabirds as predators on marine resources, especially krill, at South Georgia. In: Croxall JP (ed) Seabirds: Feeding ecology and role in the marine environment. Cambridge University Press, Cambridge

Croxall JP, Reid K, Prince PA (1999) Diet, provisioning and productivity responses of marine predators to differences in availability of Antarctic krill. Mar Ecol Prog Ser 177:115-131

Davies N, Lundberg A (1985) The influence of food on time budgets and timing of breeding in the Dunnock Prunella modularis. Ibis 127:100–110

Davis RW, Croxall JP, O’Connell MJ (1989) The reproductive energetics of gentoo (Pygoscelis papua) and macaroni (Eudyptes chrysolophus) penguins at South Georgia. J Anim Ecol 58:59-74

Forcada J, Trathan PN, Reid K, Murphy EJ, Croxall JP (2006) Contrasting population changes in sympatric penguin species in association with climate warming. Glob Change Biol 12:411-423

Fraser WR, Hofmann EE (2003) A predator’s perspective on causal links between climate change, physical forcing and ecosystem response. Mar Ecol Prog Ser 265:1-15

Green JA, Boyd IL, Woakes AJ, Green CJ, Butler PJ (2005) Do seasonal changes in metabolic rate facilitate changes in diving behavior? J Exp Biol 208:2581-2593

Grémillet D, Kuntz G, Woakes AJ, Gilbert C, Robin JP, La Maho Y, Butler PJ (2005) Year-round recordings of behavioral and physiological parameters reveal the survival strategy of a poorly insulated diving endotherm during the Arctic winter. J Exp Biol 208:4231-4241

Harding AMA, Piatt JF, Schmutz JA, Shultz MT, Van Pelt TI, Kettle AB, Speckman SG (2007) Prey density and the behavioral flexibility of a marine predator: the common murre (Uria aalge). Ecology 88:2024-2033

Hill SL, Reid K, Thorpe SE, Hinke J, Watters, GM (2007) A compilation of parameters for ecosystem dynamics models of the Scotia Sea – Antarctic Peninsula Region. CCAMLR Sci 14:1-25

Hinke JT, Salwicka K, Trivelpiece SG, Watters GM, Trivelpiece WZ (2007) Divergent responses of Pygoscelis penguins reveal common a environmental driver. Oecologia 153:845-855

Hobbs NT, Hilborn R (2006) Alternative to statistical hypothesis testing in ecology: a guide to self teaching. Ecol App 16:5-19

Jablonski B (1985) The diet of penguins on King George Island, South Shetland Islands. Acta Zool Crocov 29:117-186

Janes DN, Chappell MA (1995) The effect of ration size and body size on specific dynamic action in Adélie penguin chicks, Pygoscelis adeliae. Physiol Zool 68:1029- 1044

Jenouvrier S, Barbraud C, Weimerskirch H (2006) Sea ice affects the population dynamics of Adélie penguins in Terre Adélie. Polar Biol 29:413-423

Lescroël A, Ridoux V, Bost CA (2004) Spatial and temporal variation in the diet of the gentoo penguin (Pygoscelis papua) at Kerguelen Islands. Polar Biol 27:206-216

Lynch HJ, Fagan WF, Naveen R, Trivelpiece SG, Trivelpiece WZ (2009) Timing of clutch initiation in Pygoscelis penguins on the Antarctic Peninsula: towards an improved understanding of off-peak correction factors. CCAMLR Sci 16:149-165

Lynch HJ, Naveen R, Fagan WF (2008) Censuses of penguins, blue-eyed Phalacrocorax atriceps and southern giant petrel Macronectes giganteus populations on the Antarctic Peninsula, 2001-2007. Mar Ornithol 36:83-97

Masello JF, Mundry R, Poisbleau M, Demongin L, Voigt CC, Wikelski M, Quillfeldt P (2010) Diving seabirds share foraging space and time within and among species. Ecosphere 1(6):art19. doi:10.1890/ES10-00103.1

Miller AK, Karnovsky NJ, Trivelpiece WZ (2009) Flexible foraging strategies of gentoo penguins Pygoscelis papua over 5 years in the South Shetland Islands, Antarctica. Mar Biol 156:2527-2537

Nagy KA, Obst BS (1992) Food and energy requirements of Adélie penguins (Pygoscelis adeliae) on the Antarctic Peninsula. Physiol Zool 65:1271-1284

Polito MJ, Lynch HJ, Naveen R, Emslie SD (2011) Stable isotopes reveal regional heterogeneity in the pre-breeding distribution and diets of sympatrically breeding Pygoscelis spp. penguins. Mar Ecol Prog Ser 421:265-277

R Development Core Team (2010) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051- 07-0, URL http://www.R-project.org.

Ricklefs RE (1983) Some considerations on the reproductive energetics of pelagic seabirds. Stud Avian Biol 8:84-94

Schofield O, Ducklow HW, Martinson DG, Meredith MP, Moline MA, Fraser WR (2010) How do polar marine ecosystems respond to rapid climate change? Science 328:1520-1523

Tanton JL, Reid K, Croxall JP, Trathan PN (2004) Winter distribution and behaviour of gentoo penguins (Pygoscelis papua) at South Georgia. Polar Biol 27:299-303

Trathan PN, Croxall JP, Murphy EJ (1996) Dynamics of Antarctic penguin populations in relation to inter-annual variability in sea ice distribution. Polar Biol 16:321–330

Trivelpiece WZ, Trivelpiece SG, Volkman NJ (1987) Ecological segregation of Adélie, gentoo, and chinstrap penguins at King George Island, Antarctica. Ecology 68:351-361

Trivelpiece WZ, Buckelew S, Reiss C, Trivelpiece SG (2007) The winter distribution of chinstrap penguins from two breeding sites in the South Shetland Islands of Antarctica. Polar Biol 30:1231-1237

Trivelpiece WZ, Hinke JT, Miller AK, Reiss CS, Trivelpiece SG and Watters GM (2011) Variability in krill biomass links harvesting and climate warming to penguin population changes in Antarctica. Proc Natl Acad Sci USA 108:7625-7628

Vleck CM, Vleck D (2002) Physiological condition and reproductive consequences in Adélie penguins. Integr Comp Biol 42:76-83

Volkman NJ, Presler P, Trivelpiece W (1980) Diets of Pygoscelid penguins at King George Island, Antarctica. Condor 82:373-378

Williams TD (1991) Foraging ecology and diet of gentoo penguins Pygoscelis papua at South Georgia during winter and an assessment of their winter prey consumption. Ibis 133:3-13

Williams TD (1995) The Penguins. Oxford University Press, Oxford

Williams TD, Kato A, Croxall JP, Naito Y, Briggs DR, Rodwell S, Barton TR (1992a) Diving pattern and performance in nonbreeding gentoo penguins (Pygoscelis papua) during winter. Auk 109:223-234

Williams TD, Briggs DR, Croxall JP, Naito Y, Kato A (1992b) Diving pattern and performance in relation to foraging ecology in the gentoo penguin, Pygoscelis papua. J Zool Lond 227:211-230

Wilson PR, Ainley DG, Nur N, Jacobs SS, Barton KJ, Ballard G, Comiso JC (2001) Adélie penguin population change in the pacific sector of Antarctica: relation to sea-ice extent and the Antarctic Circumpolar Current. Mar Ecol Prog Ser 213:301–309

Wilson RP, Alvarrez B, Latorre L, Adelung D, Culik B, R Bannasch (1998) The movements of gentoo penguins Pygoscelis papua from Ardley Island, Antarctica. Polar Biol 19:407-413

Wilson RP (2010) Resource partitioning and niche hyper-volume overlap in free-living Pygoscelid penguins. Funct Ecol 24:646-657

CHAPTER 2: Adélie penguin survival rates and their relationship to environmental

indices in the Antarctic Peninsula region

ABSTRACT

The life history of seabirds is typified by low fecundity, high adult survival rates, and relatively long lives. Such traits can act as buffers to enable persistence of populations under variable environmental conditions. Numerous studies, however, have suggested that seabirds can be highly sensitive to environmental variability. In the rapidly warming Antarctic

Peninsula region, Adélie penguin (Pygoscelis adeliae) populations have declined rapidly over the last 2 – 3 decades, attributed largely to changes in temperatures, sea ice conditions, and food availability. Relatively few studies, however, have attempted to estimate survival rates of

Adélie penguins with respect to environmental variation. Here, I used 25 years of mark- recapture data from known-age individuals at a colony on King George Island, Antarctica and capture-mark-recapture modeling techniques to estimate apparent survival and recapture rates.

I asked whether trends or increased variability in survival rates existed that could account for the observed changes in population. I also examined a suite of global, regional, and local environmental conditions thought to be important for Adélie penguin survival to identify environmental covariates that best explain variability in survival rates. Overall, no long-term trends in survival were evident, but adult survival was highly variable, with frequent reductions in survival probabilities below 0.30. In particular, two consecutive years of low adult survival corresponded to the near 50% decline in breeding population between 1989 and

1991. The variation in survival, however, was not predicted well by the candidate environmental covariates. A model of survival based the Southern Oscillation Index performed best, but only explained 13% of the variation relative to a generic, time-dependent model. Limited predictive capacity of environmental conditions suggests difficulty in predicting future responses from single environmental indices.

INTRODUCTION

The general life history of seabirds is typified by low fecundity, high adult survival rates, and relatively long lives. Such traits can act as buffers that enable persistence of populations under variable environmental and climatic conditions (Stearns 1992). Despite such buffering capacity, however, observed variation and directional change in climate and environmental conditions have affected the breeding phenology, ranges, and populations trends of numerous seabird populations (Viet et al. 1996; Grosbois and Thompson 2005;

Barbraud and Weimerskirch 2006; Sandvik and Erikstad 2008; Devney et al. 2009). In particular, long term data on Antarctic seabirds dependent on sea ice habitats, including emperor (Aptenodytes forsteri) and Adélie penguins (Pygoscelis adeliae), and snow petrels

(Pagodroma nivea) suggest that variation in the seasonal duration and spatial extent of winter sea ice can be important drivers of population change (Croxall et al. 2002; Jenouvrier et al.

2005; Emmerson and Southwell 2011). Demographic models that link vital rates to environmental drivers (e.g., Jenouvrier et al. 2009) and conceptual habitat-association models

(e.g., Ainley et al. 2010) have suggested that continued warming and loss of sea ice in

Antarctica may hasten declines in the abundance of ice-obligate species.

During the last few decades, marked changes in the Antarctic Peninsula ecosystem have occurred, garnering recognition that the region is one of the most rapidly changing ecosystems on the planet. The Western Antarctic Peninsula (WAP) region is characterized by increases in air and sea temperatures (Turner et al. 2005a; Meredith and King 2005), a shrinking duration of winter sea ice coverage (Stammerjohn et al. 2008a; Turner et al. 2009), an increased frequency of storms with major precipitation events (Turner et al. 2005b) and greater accumulation rates of snow (Thomas et al. 2008). Satellite-based observations of ocean

color have revealed potential declines in marine primary production (Montes-Hugo et al.

2009), while long-term net tow survey data suggests a possibly 80% decline in the abundance of Antarctic krill (Euphausia superba) (Atkinson et al. 2004), the main food item for most seabirds, marine mammals, and fishes in the Antarctic Peninsula region. As ice-obligate, krill- dependent predators that depend on snow-free ground for successful breeding, Adélie penguin populations would be expected to decline under such changing environmental conditions

(Fraser et al. 1992; Smith et al. 1999; Trivelpiece et al. 2011). Indeed, Adélie penguin populations throughout the Antarctic Peninsula region have declined since the 1970s (Forcada et al. 2006; Hinke et al. 2007; Lynch et al. 2008). In particular, reductions in population size and recruitment to the breeding population have been correlated to variation in winter sea ice extent and krill demography throughout their circumpolar range (Trathan et al. 1996; Wilson et al. 2001; Fraser and Hofmann 2003; Jenouvrier et al. 2006; Hinke et al. 2007). The weight of evidence suggests that Adélie penguins are highly sensitive to environmental variability

(Ainley 2002).

A more thorough understanding of how changing environmental conditions have influenced Adélie penguin populations is facilitated by long-term field studies of uniquely marked individuals. Such longitudinal demographic studies enable estimation of the patterns of survival among individuals over time with respect to changing environmental conditions.

For example, in east Antarctica, complex relationships between age-structured survival rates of Adélie penguins, annual sea ice dynamics, and large scale atmospheric indices like the

Southern Oscillation Index (SOI) have been clearly linked to key demographic features like survival rates (Jenouvrier et al. 2006; Emmerson and Southwell 2011). Reports from Adélie penguin colonies in the Ross Sea offer differing views on the effects of such environmental covariates on survival rates that include either non-linear effects of sea ice on adult survival

rates (Ballerini et al. 2009), potential lagged effects of ice extent on juvenile survival rates

(Wilson et al. 2001), or no clear evidence of such environmental effects (Lescroël et al 2009).

Rather, adult survival rates appear to depend on prior reproductive activity and breeding success (Lescroël et al. 2009). Thus, throughout the southern range of Adélie penguins, there are contrasting reports of demographic effects of environmental conditions on Adélie penguin survival rates.

A prior study of Adélie penguin demography in the South Shetland Islands, a more northerly colony than the populations studied in east Antarctica or the Ross Sea, reported that recruitment, defined as the first return to the natal colony by a banded individual, had declined by 90% during the period 1981 to 2004 (Hinke et al. 2007). The authors suggested that the decline in recruitment was due to negative trends in the survival rates of young animals.

Furthermore, the trends in recruitment of Adélie penguins were correlated with an index of

Antarctic krill recruitment, supporting hypotheses about the important effects of changing environmental conditions and general bottom-up forcing on penguin populations. However, the recruitment rates reported by Hinke et al. (2007) did not provide information on survival rates, because survival and recapture probabilities are confounded when a proportional return rate is calculated. Further, the recruitment index was based on the first return of any bird, irrespective of age. Thus, information on trends and variation of survival rates for particular age-classes or life stages of Adélie penguins, and their relationship with specific environmental variables in the South Shetland Islands region, remains unreported. This study aims to fill that void to foster a better understanding of factors affecting Adélie penguin survival rates throughout their circumpolar range.

A robust approach for estimating survival rates with respect to environmental drivers rests on capture-mark-recapture (CMR) methodology (Lebreton et al. 1992). Current CMR

methods derive largely from a time-dependent model that was developed independently for recapture data by Cormack (1964), Jolly (1965), and Seber (1965), collectively referred to as the CJS model. Advances in CMR methods and generalization of the basic CJS model over time (Robson 1969; Pollack 1981; Pollack et al. 1984; Clobert and Lebreton 1985; Clobert et al. 1985; Lebreton et al. 1992) have led to a flexible suite of methods capable of examining competing hypotheses about survival rates, where alternative constraints on survival and recapture probabilities, including environmental covariates and age-, time-, or cohort- dependent rates, can be viewed as competing hypotheses for explaining observed patterns in recapture data (Lebreton et al. 1992). Here, I apply a modified CJS model to long-term CMR data on Adélie penguins to estimate survival rates during a period of rapid environmental change in the Antarctic Peninsula region. By building models with different constraints on survival and recapture probability structures, I ask whether the CMR data provide support for declines in juvenile survival rates, declines in adult survival rates, or increased variability in survival rates. I also examine the ability of environmental indices to explain variation in estimated survival rates through an integrated CMR modeling approach (Lebreton et al. 1992;

Grosbois et al. 2008).

A general life history of Adélie penguins

The life history and breeding biology of the Adélie penguin has been well characterized (Sladen 1953; Ainley et al. 1983; Trivelpiece and Trivelpiece 1990). As relevant to this study, the Adélie penguin is a pagophilic seabird that uses the marginal pack ice zone

(Ainley et al 1994) during winter for molting and as a stable platform for foraging on small fishes, krill, and other invertebrates (Polito et al. 2011). In the austral spring, adults return to

their natal colony to initiate breeding activities on the ice-free beaches of Antarctica and the

Antarctic islands. The breeding season in the northern Antarctic Peninsula region generally begins in mid October, when males arrive and build nests, followed shortly by courtship with arriving females, pair bonding, copulation, and egg laying. The breeding season typically ends by late January when adults leave the colony and migrate back to pack ice habitats prior to beginning their annual full-plumage molt (Ballard et al. 2010; Dunn et al. 2011). Chicks remain at the colony for up to 2 longer while their plumage matures, and then enter the sea to forage independently for the first time. Most chicks will not return to their natal colony for at least 2 years. By mid February, the Adélie penguin colonies are empty. For the purpose of estimating survival rates, the winter period, which includes long-distance migrations to and from the distant edge of the pack ice zone (Ballard et al. 2010; Dunn et al. 2011), an energetically expensive molt (Williams 1995), and a shifting spatial distribution of pack ice

(foraging) habitat as sea ice advances and retreats during winter, is considered an important bottleneck (Hinke et al 2007).

MATERIALS AND METHODS

Study Site

Long-term CMR data from a banded population of known-age Adélie penguins were collected at the Copacabana colony in Admiralty Bay, King George Island, South Shetland

Islands (62° 10’S, 58° 30’W). I used 25 years of data from birds that were banded from the

1984/85 through 2008/09 austral summer field seasons. The Copacabana colony is located near the northern limit of the Adélie breeding range (Trivelpiece et al. 1987), and within the

Scotia Sea and northern Antarctic Peninsula region of rapid Adélie penguin population decline

(Lynch et al. 2008; Trivelpiece et al. 2011). In addition to Adélie penguins, the Copacabana colony also supports breeding populations of two other Pygoscelid penguin species, the chinstrap (P. antarctica) and gentoo (P. papua) penguins. Nesting locations in the study colony are contained within an area less than 0.1 km2 on the flat, rocky beaches adjacent to

Admiralty Bay, and on the northwest facing slopes above the beaches. During the study period, the population of Adélie penguins has declined from a high of roughly 10,000 breeding pairs to nearly 2000 breeding pairs (Table 2-1, Fig. 1-1).

Initial release and subsequent recapture of banded individuals

In each year of the study between 250 and 2000 chicks were banded on the left flipper with either an aluminum band or, in later years, a stainless steel flipper band (Table 2-1). Each flipper band was marked with a unique, field-readable identifying number. Chicks were banded at an age of approximately 5 to 6 weeks, when small groups of chicks from throughout the colony were captured with either hand nets or corralled with a modified beach seine. All birds were immediately released following banding. At the time of banding, the sex of individuals was not determined.

During subsequent breeding seasons any observation of a banded known-age penguin in the colony was recorded. Attendance and visitation patterns of mature and immature Adélie penguins during the breeding season produces a predictable cycle in the number of birds present (Sladen 1953). The greatest numbers of birds are typically encountered at the peak of egg laying, then decline as breeding females depart to sea to feed. Any immature, unpaired, and failed breeders also depart the colony at this time for most of the remaining breeding

season. Toward the end of the breeding period a reoccupation period occurs where the numbers of animals present in the colony increases. Many younger birds return to their natal colony for the first time during this reoccupation period. To maximize recaptures of known- age individuals given the underlying variation in colony attendance, recapture efforts in the colony continued throughout the breeding season. Recapture effort included daily visits to the colony to scan peripheral and central nesting areas for banded individuals. During the reoccupation period, the daily searches were expanded to include systematic searches along landing beaches and areas adjacent to the colonies where penguins often congregated when ashore. Thus, recaptures of an individual for a given year could have occurred anytime between October and February. In general, all recaptures are visual in nature, made with binoculars or by eye, rather than physical recaptures of the bird with nets or by hand. Search effort for banded individuals at the colony is assumed to have remained constant over time. In total, the mark-recapture data set used here consists of 36,852 recaptures of 3035 individual birds banded as chicks and recaptured between the 1984/85 and 2010/11 field seasons.

Band retention study

Estimating survival probabilities with CJS models requires 4 main assumptions:

1) All marked individuals alive at time t have the same probability of

recapture

2) All marked individuals alive at time t have the same probability of survival

to time t+1.

3) Marks are not lost

4) Recapture events occur instantaneously relative to the intervals between t

and t+1.

While it is often assumed that assumptions 3 and 4 are met in field studies, the loss of marks from the population can bias estimates of survival (Nelson et al. 1980). Under most circumstances, however, the bias caused by band loss is small relative to the standard error of the estimated survival rate unless the animal under study is long-lived and band loss is unusually severe (Nelson et al. 1980). Adélie penguins are considered long-lived birds; the oldest Adélie penguin in this study was 17 years old and can exceed 20 years in the wild

(Ainley et al. 2002). Because the type of flipper band changed from aluminum to stainless steel during the 1997/98 and 1998/99 field seasons (Table 2-1) it is prudent to establish whether both band types had similar retention rates, thus similar bias on survival in order to assess potential differences in estimated parameters from different band types. Therefore, I used data from a banding study conducted over one winter period from 2000/01 to 2001/02.

The methods, data, and results from this study are presented in Appendix 2-1. Because of an apparent difference in retention rates (97% for stainless steel versus 60% for aluminum bands), it appears necessary to attempt to account for potential band effects on survival rates. I therefore investigated the effect of band type on survival and recapture probability with the

CMR models (below).

Estimate of survival and recapture probabilities

To estimate annual survival and recapture probabilities, the recapture data were summarized as an encounter history, represented as a sequence of 1s and 0s, for each individual. The initial 1 denotes the original capture and banding event, which occurs at age 0

for all birds in this study. For subsequent years, a recapture is denoted with a 1 (i.e., the bird survived and was recaptured alive) while a 0 represents a year without a recapture event (i.e., possibly survived but not recaptured or did not survive and was not recaptured). Note that recaptures were not physical recaptures of the bird, but rather visual observations a banded bird and confirmation of the band number, often made with binoculars from distances up to

10m.

CMR modeling derives from recognizing that an individual encounter history is the result of a multinomial process of survival, φ , and recapture, p, conditional on the animal having survived. Each recapture event can then be modeled as the product of the survival and the recapture probabilities. Survival and recapture probabilities for discrete encounter histories can be modeled in numerous ways, including as constant across time and groups, time- dependent, age, or stage-dependent, or as a function of environmental covariate; choosing an appropriate model structure depends on the life history of the animal under study and on the aims of the study. Here, I am interested in understanding how juvenile and adult survival rates vary, and whether environmental indices provide useful predictors of survival rates; I therefore focus model development on alternative age-class structures for use with environmental covariates. I used Program MARK (White and Burnham 1999), interfaced with R (version

2.12.0, R Core Development Team 2010) via the RMark package (Laake 2011). Program

MARK is flexible, purpose-built software that readily enables analysis of CMR data via construction of alternative models of survival and recapture processes to estimate survival and recapture probabilities from encounter history data via maximum likelihood methods.

The first step toward estimating survival and recapture probabilities with CMR data is to generate a global model that provides an adequate fit to the encounter history data to which more parsimonious models can be compared. Traditionally in mark-recapture modeling with

live recapture data, the global model assumes strict time-dependence in survival (φt ) and recapture rates (pt) (Cormack-Jolly-Seber (CJS) models, Lebreton et al. 1992); i.e. all individuals have equal probability of survival from t to t+1 and an equal probability of recapture at t. For the full data set used here, however, initial goodness of fit testing for the

CJS model with the program RELEASE (Burnham et al. 1987), suggested that the time- dependent model did not fit the data well (χ2 = 6949, P < 0.001). The lack of fit suggests that the CJS assumption of an equal probability of survival of each tagged animal is violated and/or the assumption of equal probability of recapturing each tagged animal is violated. One known violation of equal probability of recapture is that age 1 Adélie penguins are rarely recaptured. Furthermore, survival of juvenile animals may not be similar to adult survival.

Such violations of equal recapture and survival probability suggested that a modification of the global model to include age structure in the population was required. Therefore, I considered a global model that included time- and age-dependent effects on survival (φ *ta ) and recapture probability ( p *ta ) for all ages, a. This model allowed different survival and recapture probabilities through time for each age-class represented in the data. Additionally, I included an additive effect of band type, denoted with subscript band, on survival and recapture probability in the global model to account for potential differences in survival and recapture rates arising from different band types. Ideally, sex would be included as a predictor variable in the global CMR models; however, it could not be used for these data because sex is not known for animals at the time of release and sex is only documented accurately for individuals that attempted to breed. Therefore, I do not consider sex effects on survival or recapture probabilities.

In CMR studies, some individuals may be more (or less) likely to be recaptured than others based on their prior recapture history. Such behavior, termed trap dependence, can

violate the assumption of equal recapture probability of all banded individuals. Trap dependence often arises if the act of a recapture (i.e. in a trap or net) leads to behaviors that affect future recapture events. While such physical recaptures were not conducted in this study and it is unlikely that visual observations of a banded individual in one year affected our ability to detect that animal in subsequent years, differences in recapture probability among individuals can still arise. A method to account for such heterogeneity in recapture probability is to include an individual’s prior encounter history as a covariate for recapture probability

(Sandland and Kirkwood 1981; Lebreton et al. 1992). This enables autocorrelation in the encounter history data, such that individuals that have been encountered alive previously may be more (or less) likely to be recaptured again than individuals that had not been seen previously. Thus, I included trap-dependence effects on recapture probabilities, denoted with subscript td, in the model selection process (below). However, individual-level covariates like trap-dependence cannot be included in available goodness of fit tests for the global model (J

Laake, pers comm). Therefore, I did not include trap dependence in the global model for goodness of fit testing.

Goodness of fit for the chosen global model, (φ *ta +band ⋅ p *ta +band ) was assessed using a bootstrapping procedure implemented in Program MARK (White and Burnham 1999).

Briefly, the procedure simulates encounter history data based on the fitted global model and then re-fits the model to the simulated data to estimate the model deviance. The mean deviance from the bootstrapped data was then compared to the deviance of the model fit to the observed data. The ratio of observed model deviance to the mean bootstrapped deviance is an estimator of the variance inflation factor, cˆ . The variance inflation factor can be used to account for lack of fit; a perfect fit of model and data would theoretically produce a cˆ = 1 and estimates of cˆ < 3 are considered acceptable levels of over-dispersion (Lebreton et al. 1992).

If the global model provides an adequate fit to the data, further model simplification and model comparison is warranted. Because trap-dependence was likely to be important in the data (i.e. improving model fit), but unable to be assessed directly with the bootstrapping procedure, I calculated post-hoc model comparison tables using arbitrarily lower estimates of cˆ than that estimated by the bootstrap procedure to investigate the sensitivity of the model selection process to the variance inflation factor.

Simplifying the CMR model structure

The global model of full age structure was simplified by considering discrete age-

classes, denoted with subscript ac, for both survival (φ *tac ) and recapture probabilities ( p *tac ).

For survival rates, I considered a set of models with alternative combinations of juvenile, pre- breeder and adult stages to determine the best age-class structure. After fledging, Adélie penguins typically return to natal colonies during the second year of life. Thus, survival rates over the first two years of life are best modeled as a single juvenile stage. There is evidence from other demographic studies of Adélie penguins that survival following the first breeding attempt may be reduced relative to non-breeders (Ainley et al. 1980; Lescroël et al. 2009), thus a pre-breeding age-class may exhibit higher survival than breeding-age adults. Adélie penguins do not mature until age 3 and most individuals did not initiate breeding until age 4, so that the survival from age 2 to 3 or from age 2 to 4 plausibly represent pre-breeding stages.

For adults in this study, recaptures of individuals older than 8 years was rare. Since Adélie penguins can live up to 20 years in the wild (Ainley 2002), effects of senescence cannot be assessed with the available data. For this reason, I defined a single adult age-class, with a variable minimum of either age 3 (physically mature) or 4 (most frequent first breeding age)

selected to correspond to the maximum age of the pre-breeder stage. Finally, I consider a model that included only juvenile (0-2) and adult (2+) age-classes for survival. Thus, I examined 3 alternative life-stage models for survival probability, denoted ac2 for a 2 age-class model (juveniles 0-2, adults 2+), and a three age-class model, ac3, for juveniles (0-2), pre- breeders (2-3), and adults (3+) and an alternative 3 age-class model, ac3', for juveniles (0-2), pre-breeders (2-4) and adults (4+).

For recapture probabilities, I considered similar age-classes as those for survival. I included a juvenile stage for the recapture of age 1 animals. For older birds, I considered separate recapture rates for age 2 (immature) and all birds aged 3+ (adult), or separate recapture rates for birds of age 2, 3 and all birds 4+ to represent different probability of recapture for first time returns, first time to breed returns, and adult returns. Thus, the candidate recapture models included a 2-stage model with juveniles (age 1) and adults (ages

2+), denoted ac2, a three stage model, ac3, for juveniles (age 1), pre-breeders (age 2), and adults (ages 3+), and an 4 stage model, ac4, for recapture of juveniles (age 1), separate pre- breeder stages for recapture of age 2 and age 3 birds, and adults (ages 4+). For survival and recapture models, exploratory analyses with more resolution for older age-classes suggested that limiting the models of survival and recapture probability to a generic adult stage, as indicated above, were sufficient.

I included band type (stainless steel or aluminum) as an additive effect in both survival and recapture rates. With only two years of overlapping releases (Table 2-1), there was insufficient data to estimate full interaction terms of time or age with band type. Models with and without the banding effect were compared for assessing improvement of model performance. I included trap-dependence as an additive effect for recapture probability.

Models with and without trap dependence were compared for assessing improvement of

model performance. For the model selection process, I modeled time effects as interactive or additive for survival and recapture probability. All combinations of base models were run and compared using AIC.

Correlates of survival and recapture probability

Environmental and biological factors are thought to affect Adélie penguin populations

(Ainley et al. 2010; Trivelpiece et al. 2011). To assess the amount of variation in survival and recapture probability that could be explained by environmental indices, I examined the role of six candidate correlates of penguin survival and recapture, along with a combined index constructed with a principal components analysis. The chosen variables covered a time span of

1987 through 2009 and included indices of large-scale atmospheric forcing (Southern

Oscillation Index (SOI) and the Southern Annular Mode (SAM), regional indices of physical habitat (sea ice extent) and availability of prey (krill density), and local measures of environmental conditions (spring temperatures and precipitations). The variables have been implicated in variation in Adélie population dynamics and survival throughout much of its breeding range (Forcada et al. 2009; Jenouvrier et al. 2009; Trivelpiece et al. 2011). Note that the time span considered for the environmental covariates is two years shorter than that available for the full CMR analysis described above because of a lack of data for krill density prior to 1987. Thus, these models were run using CMR data for 1987-2009, but based on the model developed above for the full study period (1985-2009).

In the Southern Ocean, the SOI and the SAM are important drivers of environmental variability whose effects on sea ice dynamics (Kwok and Comiso 2002; Yuan 2004) and temperatures (Turner 2004) have been correlated with variation in penguin populations

throughout Antarctica (Forcada et al. 2009). The SOI index, which measures sea-level pressure differences between Darwin, Australia and Tahiti, has been linked variation in

Antarctic sea ice (Kwok and Comiso 2002) and to variation in Antarctic krill populations

(Loeb et al. 2009), both key components in the Adélie penguin life history. I used an annual average of monthly values of the SOI index available at the Australian Bureau of Meteorology

(http://www.bom.gov.au/climate/current/soihtm1.shtml, accessed 9/6/2011). Likewise, recent trends in the SAM index, which characterizes the latitudinal pressure gradient between

Antarctic and middle latitudes, have been implicated in the accelerated rate of warming in the western Antarctic Peninsula region (Kwok and Comiso 2002) resulting in a loss of sea ice, hence foraging area during winter for Adélie penguins. I used the SAM index values, available from the British Antarctic Survey (http://www.nerc-bas.ac.uk/icd/gjma/sam.html, accessed

9/6/2011), for the winter period (June, July and August). The winter SAM index has exhibited the strongest annual trend, as compared to the other seasons, and is correlated with the overall increase in the SAM, during the study period.

The regional physical and biological indice.s were taken from on-going programs conducted by the US Antarctic Ecosystem Research Division (AERD). Sea ice is a necessary platform for Adélie penguin foraging during winter. The US AERD provided annual estimates, based on the analysis by Hewitt (1997), of annual sea ice coverage in the study region. Much of the foodweb in the Antarctic Peninsula is supported energetically by

Euphausia superba, the Antarctic krill (Laws 1985) and Adélie penguin diets throughout the region are dominated by krill (Volkman et al. 1979; Jablonski 1985; Polito et al. 2011). I used estimates of krill density, obtained from systematic net-based surveys of krill conducted by the

US AERD and reported by Reiss et al. (2008). To extend the time series backwards in time to

best match the banding data from Adélie penguins, I used only estimates of krill density in the vicinity of Elephant Island, located some 200km northeast of the Copacabana colony.

Local conditions during the breeding season can influence the decision of whether animals return to breed or not. In particular, temperature and precipitation determine the amount of snow and ice that is present in the breeding colonies, which can prevent access to the stones necessary for nest construction (Patterson et al. 1997; Boersma 2008). Thus, local conditions may be related to annual survival and recapture rates. I used average spring

(September – November) temperatures and total spring precipitation amounts that were observed systematically at Bellingshausen Station, a Russian research station located approximately 22 km southwest of the study colony. Data were acquired online at http://www.aari.aq, accessed 9/6/2011.

Once the best model with time and age-class effects had been determined, the covariates were used in place of the generic time dependence in the best-fitting models of survival and recapture probability. This procedure allowed an analysis of deviance to compare the fits of the generic time-varying model ( Ψtime ) and the environmental covariate model (

Ψenv ) with respect to a constant model ( Ψcst ) to estimate the proportion of deviance explained (a deviance R2; Grosbois et al. 2008). Specifically, I calculated the relative proportion of deviance explained by an environmental covariate relative to the generic time- varying model as:

2 Dev(Ψ ) − Dev(Ψ ) Deviance R = cst env (Grosbois et al. 2008). Dev(Ψcst ) − Dev(Ψtime )

RESULTS

Summary of banding data

In total, 16343 aluminum bands were released between 1985 and 1999, while 5503 stainless steel bands were released between 1998 and 2009 (Table 2-1). The size of banded cohorts was reduced over time to maintain tagging rates as the population size declined.

Relative to population size, tagging rates averaged 15.7% of population size (SD=0.04), and ranged from 9.2 to 23.3% (Fig. 2-1).

For both band types, the majority of banded individuals were never recaptured. Total return rates from each cohort ranged from 3% to 30% (Fig. 2-2). On average, 13% of all releases from each band type were recaptured, with 2272 individuals with aluminum bands and 763 individuals with stainless steel bands having returned to their natal colony at least once through March 2011 (Table 2-1). There was a negative trend the proportion of each banded cohort that was recaptured from 1985 through 1999, followed by an apparent increase in return rates through the early 2000s. Despite a limited number of years available to judge total return rates for the cohorts released in 2008 and 2009, the proportional return rates for those cohorts appeared to be the lowest on record.

The recapture data were dominated by young animals. For both band types, most individuals returned to the natal colony for the first time at ages 2 or 3 (Fig. 2-3a).

Furthermore, most individuals were seen in only one year. The cumulative proportion of the number of years in which an individual was recaptured was similar (Fig. 2-3b); 95% of aluminum and stainless steel banded birds were seen in three or fewer years after initial release. The age distributions of recaptured penguins from each band type were also similar

(Fig. 2-4), with a mean age of recaptured birds with aluminum bands at 4.04 ± 1.52 (SD) and a mean age of 3.85 ± 1.24 (SD) for all individuals tagged with stainless steel bands. Returns of 1 year olds was rare (N=36), so the mean age of recaptured penguins is dominated by those individuals that returned once or twice between the ages of 2 and 4.

Despite similar characteristics in age distribution of returning birds and the frequency of recapturing tagged individuals, there is an indication that proportional return rates differed by band type. When both band types were released at the same time in 1998 and 1999, the proportions of total recaptures of birds with stainless steel bands was higher that the return rate than birds with aluminum bands (Fig. 2-2). A one-year tag retention study, conducted during the 2000/01 and 2001/02 field season (Appendix 2-1) suggested that stainless steel bands had a higher retention rate. The retention rate of stainless steel bands was estimated at 97%, while aluminum bands were estimated to have a retention rate of 60%. Estimates of survival and recapture probability may therefore be biased by tag loss. For this reason, estimation of survival and recapture rates from the mark-recapture data accounts for potential differences due to band type.

Capture-Mark-Recapture Models

Goodness of fit for candidate global model

Bootstrapped estimates of the deviance in the global model ()φ *ta +band ∗ p *ta +band suggested a lack of ft between the global model and data ( cˆ = 0.2 ). This level of overdispersion is within the range of acceptable levels for model comparison (Lebreton et al.

1992), so model selection therefore used quasi-likelihoods and small-sample corrected quasi-

AIC criteria (QAICc, White and Burnham 1999).

Model selection

Of the models considered, the best model selected included interactive effects of time and the 2 age-class model for survival (ac2) with an additive effect of band type ()φac *2 t+band and 3 age-classes with additive effects of time and trap dependence for recapture probability

()pac3+time+td . The best model collected 67% of the QAICc weight (Table 2-2). The same best model was also selected first if cˆ = 75.1 , garnering 58% of the QAICc weight and was within 2 QAICc units of the best fitting model when cˆ = 5.1 , with 14% of the model weight

(Table 2-2). Across a broad range of acceptable levels of overdispersion the choice of the best model in Table 2-2 for further analysis appears justified.

Estimates of survival and recapture probabilities

The best model of survival rates included a juvenile and an adult age-class only. The model further indicated an important effect of band type, with higher apparent survival of stainless steel bands relative to aluminum bands (Figs. 2-5 and 2-6). This is consistent with the observation of higher rates of band loss among aluminum bands in the PIT tag study. A number of survival rates for the juvenile age-class were poorly estimated, however, as indicated by large standard errors for juvenile survival in 1985, 1987, 2001, 2008, and 2009

(Fig. 5). When only estimates for which standard errors were able to be calculated were considered, there were no significant trends in juvenile survival for birds banded with

aluminum bands (F1,12 = 1.83, P = 0.2) or stainless steel band (F1,6 = 0.17, P = 0.69). On average, juvenile survival rates were estimated to be 0.45±0.15. Apparent survival rates for adults were better estimated (Fig. 2-6) and there were no apparent long-term trends for aluminum (F1,14 = 0.8, P = 0.38) or stainless steel bands (F1,7 = 0.36, P = 0.56). On average, adult survival rates were more variable that juvenile survival rates, but had equivalent mean values (mean survival = 0.45±0.18). Adult and juvenile survival rates were not highly correlated for aluminum bands (r = 0.34, t11 = 1.2, P = 0.25) or stainless steel bands (r = -0.2, t4=-0.41, P = 0.7). The main feature among the adult survival rates were the frequent reductions in apparent survival from relatively high rates of 75% to values that would be considered extremely low for long-lived species (< 25%). This suggests that periodic boom and bust conditions confront this population of Adélie penguins. Notably, the two-year span of very low apparent survival rates among adults in 1989 and 1990 corresponded to the near 50% decline in the breeding population that was documented between 1989 and 1991 (Fig. 2-1).

Adult survival rates since that population decline have continued to fluctuate strongly; this feature was common to both band types.

Recapture probability of age 1 animals was estimated to be near 0, (Fig. 2-7) consistent with very low proportions of animals ever seen as age 1 (Fig. 2-3a), and recapture probability increased in older age-classes. An additive effect of time was also found to be important, such that recapture probabilities were time-varying, but that recapture of all age- classes was subject to the same time effect, which suggests a common local effect on return rates of animals to the study colony. Furthermore, trap-dependence was included in the best fitting model, and suggested individuals recaptured previously were more likely to be recaptured again. Different band types did not appear to affect recapture rates. Similar to the survival rates, there were no apparent trends in recapture probability for the age-classes,

although recapture probability may have declined in the late 1990s and then recovered to values similar to those in the late 1980s and early 1990s. Overall, the relative constancy of recapture probability over time is consistent with the assumption of equal recapture effort expended during the breeding season searching for banded individuals.

Environmental correlates of survival and recapture probability

The variability evident in apparent survival and recapture probabilities was not well explained by any single index of environmental conditions, or by the principal components of a multivariate index created from the single variables considered here. Using the mark- recapture models, the potential percent of null deviance explained by each covariate was less than 13% (Table 2-3), relative to the best time-varying generic model identified above. For both recapture and survival probability, the SOI explained roughly 10% of the variation, while ice and krill explained only 3-7% of the variation. A post-hoc correlation analysis of the de- trended indices agreed, suggesting only non-significant (P > 0.05 in all comparison) relationships with apparent survival (Table 2-4). Krill density and ice extent exhibited the strongest correlations with apparent survival of juvenile Adélie penguins, generally suggesting higher survival with greater krill density, but reduced sea ice extent during winter. Among adults, temperature and ice extent exhibited the strongest correlations with survival rates, suggesting higher survival rates with colder, icier conditions. In general, the large-scale atmospheric drivers and the first two principal components of the combined indices were weakly related to observed survival rates. For recapture probability, none of the potential correlates provided a significant relationship.

DISCUSSION

Life history theory predicts that population growth rates in long-lived animals like many seabirds ought to be most sensitive to variation in adult survival rates (Lebreton and

Clobert 1991). Consequently, long-lived organisms are typically characterized as exhibiting little variability in adult survival rates (Morris and Doak 2004; Sæther and Bakke 2000).

During a period of rapid population decline among Adélie penguins in the South Shetland

Islands (Lynch et al. 2008; Trivelpiece et al. 2011), survival rates of banded Adélie penguins demonstrated no long term trends, but varied strongly over the course of a 25 year study, ranging from less than 30% to greater than 75%. The high variability resulted in average adult survival rates that were equivalent to average juvenile survival rates. Such low adult survival rates, coupled with high interannual variability, can be deleterious for long-lived organisms

(e.g., Lewontin and Cohen 1969; Frederiksen et al. 2008) and help explain the rapid decline of the breeding population in the study colony. For example, two consecutive years of poor adult survival corresponded to the 50% decline in breeding population abundance that occurred from 1989 to 1991. The early 2000s were also marked by relatively low adult survival rates that corresponded to another period of continual decline in the breeding population. Thus, it seems evident that the current negative trend in the breeding population of Adélie penguins at the study colony is driven by the large degree of variability in adult survival rates.

The first winter of independence is widely considered an important bottleneck for survival among juvenile penguins due to their limited foraging experience in a marine environment (Ainley et al. 2002). Consistent with that assertion, the estimated juvenile survival rates remained relatively low throughout the study period and did not exhibit temporal trends. This suggests that large-scale environmental changes considered important

for penguins survival, including trends for reduced krill abundance (Atkinson et al. 2004) and duration of winter sea ice coverage throughout the area (Stammerjohn et al. 2008a) may have a muted effect on juvenile survival rates. Emmerson and Southwell (2011) also noted that large-scale environmental conditions poorly explained the variation in survival rates of juvenile penguins, despite matching spatially resolved indices of environmental conditions with probable over-winter locations of juvenile birds. Conditions contributing to variation in juvenile survival rates may occur on smaller spatio-temporal scales. For example, variation in local food availability or changes in predation risk from leopard seals (Hydruga leptonyx) and other marine mammals (e.g., Ainley et al. 2005; Ciaputa and Siciński 2006; Pitman 2010) in the vicinity of breeding colonies may lower survival rates of juveniles immediately following the fledging period, when the novice birds are exposed to marine conditions for the first time.

Alternatively, juvenile survival rates may respond in a more complex manner to broad-scale environmental conditions, such that a linear approach for assessing variation in survival rates inadequately address the relationship between survival and environmental conditions. In either case, opportunity remains for further research to understand the factors that contribute to variation in juvenile survival rates of Adélie penguins.

A lack of a declining trend in juvenile survival contrasts with an earlier report of demographic effects on the population decline among Adélie penguins at the study colony.

Based on proportional return rates of banded individuals, Hinke et al. (2007) suggested that a decline in survival rates of juvenile penguins could explain the observed decline in the breeding population of Adélie penguins at the study colony. However, juvenile survival rates have exhibited no long term decline and variability in adult survival rate appears relatively more important for the decline in population size. The apparent contradiction likely stems from two different reasons. First, the definitions of “juvenile” used for each study differed.

The recruitment index of Hinke et al. (2007) was not exclusive to birds surviving from age 0 to age 2 (the definition of juvenile used here), but included all first-time returns to the natal colony regardless of age. Second, the proportional return rate used by Hinke et al. (2007) combined the survival and recapture probabilities, so that a declining return rate could result from just reduced recapture probabilities. Indeed, while recapture probabilities showed no trend, there was an apparent dip in recapture probabilities around the year 2000 that coincided with relatively low survival rates. Thus, the reported declines in recruitment (Hinke et al.

2007) are consistent with the estimates of survival and recapture probability reported here, but the decline in recruitment is partially driven by reduced return rates of mature (age 3+) birds, rather than reduced survival among true juvenile (immature) birds only.

Complex environmental effects on Adélie penguin survival?

Large-scale atmospheric indices and associated regional changes in marine conditions have been implicated in trends in numerous seabird populations around the world (Veit et al.

1996; Croxall et al. 2002; Sandvik et al. 2005; Devney et al. 2009). In Antarctica, the responses of penguin populations to environmental variability have included changes in abundance (e.g., Trathan et al. 1996; Wilson et al. 2001; Le Bohec 2008), but also included shifts in range (Emslie et al. 1998), changes in breeding phenology (Barbraud and

Weimerskirch 2006; Lynch et al. 2009; Hinke et al. in review), and changes in foraging behavior (Fraser and Hofmann 2003; Lescoël and Bost 2005). However, despite major changes in environmental conditions in the Antarctic Peninsula region (Turner et al. 2005a;

Stammerjohn et al. 2008a; Meredith and King 2008), the patterns evident in survival rates of

Adélie penguins were not well explained by the selected indices of environmental conditions.

Among the selected indices, the SOI and krill density were the best predictors of survival rates, but each explained less than 15% of the variation.

The relatively weak effects of environmental conditions on survival rates of Adélie penguins in the northern Antarctic Peninsula region contrasts with demographic studies of

Adélie penguins from more southern locales. There, stronger effects of winter sea ice and large-scale atmospheric drivers, like the SOI, on survival rates (Jenouvrier et al. 2006;

Ballerini et al. 2009; Emmerson and Southwell 2011) have been reported. To understand such differences, it is useful to consider how the two systems may differ. One important difference between the southern and northern regions is the prevalence and seasonal variation in sea ice.

Since the mid 1970s, the winter sea ice season has shrunk by nearly one month in the western

Antarctic Peninsula, while sea ice extent in the Ross Sea and along East Antarctica has grown or remained relatively stable (Parkinson 2004; Stammerjohn et al. 2008b). Shifting environmental conditions (and any synergistic interactions among them) can make predicting the response of species difficult (e.g. Folke et al. 2004; Brook et al. 2008; Tylianakis et al.

2008), and the WAP is among the most rapidly changing regions on earth. One possible result of the decline in critical habitat in the northern part of the Adélie range is an elevated importance of alternative drivers that affect variation in survival rates. For example, food web changes in the southern Scotia Sea that are associated with commercial krill fisheries, increased competition for prey due to the recovery of once-depleted marine mammal populations, and overall declines in preferred prey (Trivelpiece et al. 2011) may interact with

ENSO events, reductions in sea ice extent, or local weather conditions to affect survival rates in non-linear, complex ways. Such interactions could dampen the signal of a single index of environmental conditions, but are difficult to identify with standard CMR models.

Nonetheless, the relatively weak correlations between estimated survival rates from the

generic time-varying model and candidate environmental conditions support the conclusion for a lack of strong, direct effects on Adélie penguin survival rates. While expectations of future climate change and its effects on the physical (Kwok and Comiso 2002; Yuan 2004) and biological systems (Ducklow et al. 2007; Loeb et al. 2009) remain important considerations for understanding changes in penguin populations (Forcada and Trathan 2009;

Jenouvrier et al. 2009; Ainley et al 2010), confronting data with models of the interactions among environmental conditions and a fuller consideration of food web effects on penguins

(e.g., Ainley et al. 2007) will be useful research directions.

Band retention and potential negative bias on survival estimates

The use of flipper bands on penguins for individual identification and demographic studies has a long history. However, effects of flipper bands include increased energy expenditure, reduced reproductive success, and reduced survival rates (Culik et al. 1993;

Froget et al. 1998; Jackson and Wilson 2002; Gauthier-Clerc et al. 2004). Other studies report little to no effect of properly designed and attached bands (Hindell et al. 1996; Boersma and

Rebstock 2009). With respect to the estimation of survival rates, increased mortality rates that arise from the presence of the band or the loss of flipper bands from the marked population are an important source of potential bias. The double banding study (Appendix 2-1) conducted during two field seasons indicated that band retention rates of aluminum bands were much lower than for stainless steel bands. Furthermore, the modeling results presented here indicated that birds with aluminum bands exhibited lower survival rates relative to animals with stainless steel bands. Together, the results suggest that the estimates of survival from the banded population of long-lived Adélie penguins were biased low, though the magnitude of

that bias is not known. The possibility that such band effects also mediate the influence of environmental drivers may also exist. For example, band effects may be minor or absent under normal conditions, but quickly become deleterious under adverse conditions. Such an affect was suggested among Adélie penguins in the Ross Sea, where band-induced reductions in adult survival occurred with the presence of grounded icebergs that severely restricted access to preferred foraging grounds (Dugger et al. 2006). At worst, such sensitivity could undermine the accuracy of results that relate survival rates to environmental indices, thus explaining the poor predictive power of several environmental indices examined here. However, the estimated survival and recapture rates were consistent with the overall trends in the population and with the recruitment indices of Hinke et al (2007). Thus, despite potential bias, the overall results from a long-term banding program are largely unchanged; namely that a high degree of variation in adult survival rates, rather than long-term trends, appear responsible for the rapid decline in the population of Adélie penguins.

Conclusion

Recent studies in seabirds have demonstrated how increased variance in survival rates due to environmental variability can drive population declines (Frederiksen et al. 2008;

Jenouvrier 2009). While no definitive links to environmental conditions were established, a continuation of similar levels of variation in survival rates would lead to the expectation for continued negative population growth. A hallmark of future climate change includes continued directional change, but also increases in climate variation and extreme events

(Solomon et al. 2007). If such events cause the frequency of years with poor survival rates to increase, accelerated population decline may occur. This prediction can be tested with a

demographic model of population dynamics (e.g., Jenouvrier et al. 2009), and is the focus of the next chapter.

ACKNOWLEDGEMENTS

Many thanks to W. and S. Trivelpiece, and the numerous field biologists whose commitment to long-term studies in the Antarctic made this study possible. The U.S. Antarctic

Ecosystem Research Division at the Southwest Fisheries Science Center, National Oceanic and Atmospheric Administration and grants from the National Science Foundation to W.Z.

Trivelpiece provided logistical and financial support. Additional support was generously provided by the Lenfest Oceans Program at the Pew Charitable Trusts.

Table 2-1: Annual estimate of population size and number of Adélie penguins banded with aluminum or stainless steel flipper bands and subsequently recaptured during the study period (1985-2009).

Aluminum bands Stainless Steel bands YEAR Population Released Recaptured Released Recaptured 1985 8071 1200 294 1986 6870 1600 309 1987 8559 2000 465 1988 8592 1048 133 1989 9663 1000 137 1990 7589 1500 166 1991 5541 1000 171 1992 5721 1000 86 1993 6874 1000 72 1994 6219 1000 67 1995 6000 998 101 1996 6769 1000 132 1997 5644 1000 99 1998 5612 497 22 494 50 1999 7093 500 18 500 29 2000 5432 501 88 2001 5154 500 36 2002 4533 500 26 2003 3877 506 77 2004 3066 500 88 2005 2679 500 71 2006 2393 500 152 2007 2322 500 123 2008 2142 250 15 2009 2577 252 8 Total 16343 2272 5503 763

Table 2-2: Model selection table for survival ( φ ) and recapture ( p ) probabilities. Models are ranked according to ΔQAICc and assume a value of cˆ = .0.2 Alternative model rankings assuming lower values of cˆ and resulting QAICc model weights (in parentheses) are provided. In the table, ac2 refers to a 2 age-class model, ac3 refers to a 3 age-class model, band refers to a factor for different band types, and td refers to trap dependence.

Alternative rank Model N (QAICc weight) Survival (φ ) Recapture (p) pars -2LogLike QAICc ΔQAICc weight cˆ = 75.1 cˆ = 5.1 ac2:time + band ac3 + time + td 74 27292.69 13794.77 0.00 0.668 1 (0.58) 3 (0.14) ac2:time + band ac3 + time + band + td 75 27292.59 13796.74 1.96 0.250 2 (0.22) 6 (0.06) ac2:time ac3 + time + band + td 74 27302.21 13799.54 4.76 0.062 5 (0.04) 9 (0.01) ac3':time + band ac3 + time + td 94 27231.16 13804.27 9.50 0.006 4 (0.05) 2 (0.21) ac3:time + band ac3 + time + td 95 27227.39 13804.40 9.63 0.005 3 (0.05) 1 (0.27) ac3':time + band ac3 + time + band + td 95 27230.72 13806.07 11.3 0.002 7 (0.02) 5 (0.09) ac3:time + band ac3 + time + band + td 96 27226.89 13806.17 11.4 0.002 6 (0.02) 4 (0.11) ac3':time ac3 + time + band + td 94 27235.14 13806.26 11.5 0.002 8 (0.01) 6 (0.06) ac3:time ac3 + time + band + td 95 27232.34 13806.88 12.1 0.002 9 (0.01) 8 (0.05)

Table 2-3: Percent of temporal variation in survival (φ ) and recapture (p) models accounted for by environmental covariates. For the models with environmental covariates of survival, the recapture model used was the generic time-varying model identified in Table 2. For the models of recapture probability, p, the survival model is the generic time-varying model identified in Table 2. The constant survival model used for reference included 2 age-classes and an additive effect band type. The constant recapture probability model used for reference included 3 age-classes and an additive effect of trap dependence. The best explanatory variables are highlighted in bold.

Covariate Survival (φ ) Recapture ( p ) Ice 3.7 3.3 Krill density 7.3 2.6 PCA1 9.9 8.8 PCA2 6.1 0.5 Precipitation 6.3 0.1 SAM (winter) 0.6 0.4 SOI 12.9 10.3 Temperature 6.6 11.3

Table 2-4: Correlation coefficients for the relationship between estimated survival (φ ) and recapture ( p ) probabilities and select environmental variables. The strongest correlations are in bold type. SOI = Southern Oscillation Index, SAM =mean winter (June-August) values for the Southern Annular Mode, PC1 and PC2 = first and second, respectively, principal component for the standardized environmental variables.

Krill Age-class SOI SAM (JJA) Ice extent density Precipitation Temperature PC1 PC2 φ juvenile -0.01 0.14 -0.17 0.19 -0.07 -0.02 -0.14 -0.01 adult -0.29 -0.02 0.31 -0.16 0.29 -0.42 0.30 -0.14 p 1 -0.06 -0.08 -0.01 0.07 0.27 -0.23 0.06 0.32 2 -0.08 -0.06 0.02 0.03 0.21 -0.22 0.08 0.25 3+ -0.10 -0.04 0.06 -0.04 0.11 -0.21 0.12 0.16

12000 0.25

10000 0.2

8000 0.15 6000 0.1

4000 Tagging rate N breeding pairs breeding N 0.05 2000 Population Tagging rate 0 0 1985 1990 1995 2000 2005 Year

Fig. 2-1: Total number of breeding pairs in the Copacabana colony and the tagging rate relative to population size during the study period.

0.35 Aluminum 0.3 Stainless steel 0.25

0.2

0.15

0.1 Proportion Proportion recaptured 0.05

0 1980 1985 1990 1995 2000 2005 2010 Year

Fig. 2-2: Proportion of banded individuals from each year of release that we eventually recaptured. Note that recaptures through 2011 are included, but only releases through 2009.

0.6

0.5 Aluminum 0.4 Stainless steel

0.3

0.2 Proportion Proportion seen

0.1

0 1 2 3 4 5 6 Age at first resight

0.9

0.8

0.7 Aluminum Stainless steel Cumulative Cumulative proportion 0.6

0.5 0 2 4 6 8 10 12 14 Number of years

Fig. 2-3: Comparison of return statistics for Adélie penguins with aluminum and stainless steel flipper bands. A) Proportion of first-time recaptures of banded birds with respect to age. B) Cumulative proportion of the total number of times an individual was recaptured during the study period.

14 Aluminum 12 Stainless steel 10

Age 6

0 1980 1985 1990 1995 2000 2005 2010 2015 Year

Fig. 2-4: Mean ages of individuals banded with aluminum and stainless steel flipper bands that were recaptured from each cohort released. Minimum and maximum ages are also plotted for aluminum (solid line) and stainless steel (dotted line) bands.

1.00

0.75

0.50

Survival Survival probability 0.25 Aluminum

Stainless steel 0.00 1980 1985 1990 1995 2000 2005 2010 Year

Fig. 2-5: Estimates ± SE of apparent survival rates from the best-fitting model for juvenile Adélie penguins (age 0-2). Years for which standard errors could not be estimated are denoted with an asterisk. Reference lines for apparent trends in survival for the years with full estimates of SE are provided for aluminum bands (short-dash line) and stainless steel bands (long-dash lines).

1.00 Aluminun Stainless steel

0.75

0.50

Survival Survival probability 0.25

0.00 1985 1990 1995 2000 2005 2010 Year

Fig. 2-6: Estimates ± SE of apparent survival rates from best-fitting model for adult Adélie penguins (age 2+). Reference lines for apparent trends in survival are provided for aluminum bands (short-dash line) and stainless steel bands (long-dash lines).

1.000

Age 1 Age 2 0.750 Ages 3+

0.500

0.250 Recapture probabilty Recapture

0.000 1980 1985 1990 1995 2000 2005 2010 Year

Fig. 2-7: Estimates ± SE of recapture probability for Adélie penguins from the best fitting model. Note that there was not banding effect included in the best model of recapture rates.

Appendix 2-1: Methods, data and results of a double banding study to estimate the retention rates of aluminum and stainless steel bands

In the austral spring of 2000, 200 nesting male Adélie penguins were implanted with a passive integrated transponder (PIT tags). The PIT tags were placed under the skin of the lower neck/upper back with a sterile, single-use hypodermic applicator. Of the 200 penguins with PIT tags, 100 also received an aluminum flipper band and another 100 received a stainless steel flipper band. Male Adélie penguins were chosen for the study because they exhibit high nest site fidelity, returning to the same nest location with >90% probability in subsequent years (Trivelpiece and Trivelpiece 1990). Such high nest fidelity helps ensure recapture of as many individuals as possible. The following spring, the study colonies were visually searched for banded birds and a hand-held receiver was used to scan all individuals in the study colony, both banded and unbanded, for PIT tags. The resulting recapture data are presented in Table 1.

Band retention rates were estimated following the method of Kendall et al. (2009). The probabilities of retaining the flipper band (θf) or the pit tag (θp) are estimated by iteratively solving the following equations for each band type (I use the data from the aluminum bands to illustrate the method):

Θ f Θ p Proportion of birds retaining both bands /13 24 ≈ []1 − 1( − Θ f )(1 − Θ p )

Θ f 1( − Θ p ) Proportion of birds retaining the flipper band only: 24/2 ≈ []1− 1( − Θ f )(1− Θ p )

1( − Θ f )Θ p Proportion of birds retaining the PIT tag only: 24/9 ≈ []1− 1( − Θ f )(1− Θ p )

Flipper Flipper PIT band PIT N. N. Both band tag retention Retention Band Type released Detected Bands only only rate (θf) rate (θp) Alum + Pit 100 24 13 2 9 0.6 0.87 SS + Pit 100 33 29 3 1 0.97 0.9

The retention rate of aluminum flipper band were estimated to be 0.6, while stainless steel retention rates were much higher at 0.97. The difference between bands is highly 2 significant ( χ = 11.4, df =1, p = 0.0007).

REFERENCES

Ainley DG (2002) The Adélie penguin: Bellwether of climate change. Columbia University Press, New York

Ainley DG, Ballard G Ackley S, Blight LK, Eastman JK, Emslie SD Lescroël A, Olmastoni S, Townsend SE, Tynan CE, Wilson P, Woehler E (2007) Paradigm lost, or is top-down forcing no longer significant in the Antarctic marine ecosystem? Antarct Sci 19:283-290

Ainley DG, LeResche RE, Sladen WJL (1983) Breeding biology of the Adélie penguin. University of California Press, Berkeley

Ainley DG, Ribic CA, Fraser WF (1994) Ecological structure among migrant and resident seabirds of the Scotia-Weddell confluence region. J Anim Ecol 63:347-364

Atkinson A, Siegel V, Pakhomov E, Rothery P (2004) Long-term decline in krill stock and increase in salps within the Southern Ocean. Nature 432:100-103

Ballard G, Toniolo V Ainley DG, Parkinson CL, Arrigo KR, Trathan PN (2010) Responding to climate change: Adélie penguins confront astronomical and ocean boundaries. Ecology 91:2056-2069

Ballerini T, Tavechia G, Olmastroni S, Pezzo F, Focardi S (2009) Nonlinear effects of winter sea ice on the survival probabilities of Adélie penguins. Oecologia 161:253-265

Barbraud C, Weimerskirch H (2006) Antarctic birds breed later in response to climate change. Proc Natl Acad Sci USA 103:6248-6251

Boersma PD (2008) Penguins as marine sentinels. BioScience 58:597-607

Boersma PD, Rebstock GA (2009) Flipper bands do not affect foraging-trip durations of Magellanic penguins. J Field Ornithol 80:408-418

Brook BW, Sodhi NS, Bradshaw CJA (2008) Synergies among extinction drivers under global change. Trends Ecol Evol 23:453-460

Burnham KP, Anderson DR, White GC, Brownie C, Pollock KH (1987) Design and analysis methods for fish survival experiments based on release-recapture. Am Fish Soc Monogr 5

Ciaputa P, Siciński J (2006) Seasonal and annual changes in Antarctic fur seal (Arctocephalus gazella) diet in the area of Admiralty Bay, King George Island, South Shetland Islands. Polish Polar Res 27:171-184

Clobert J, Lebreton JD (1985) Dependance de facteurs de milieu dans les estimations de taux de survie par capture-recapture. Biometrics 41:1031-1037

Clobert J, Lebreton JD, Clobert-Gillet M, Coquillart H (1985) The estimation of survival in bird population by recaptures or resighting of marked individuals. In: Morgan BJT, North PM, (eds) Statistics in ornithology. Springer-Verlag, New York

Cormack RM (1964) Estimates of survival from the sighting of marked animals. Biometrika 51:429-438

Croxall JP, Trathan PN, Murphy EJ (2002) Environmental change and Antarctic seabird populations. Science 297:1510-1514

Culik B, Wilson RP, Bannasch R (1993) Flipper-bands on penguins: what is the cost of a life-long commitment? Mar Ecol Prog Ser 98:209-214

Devney CA, Short M, Congdon BC (2009) Sensitivity of tropical seabirds to El Niño precursors. Ecology 90:1175-1183

Ducklow HW, Baker K, Martinson DG, Quetin LB, Ross RM, Smith RC, Stammerjohn SE, Vernet M, Fraser W (2007) Marine pelagic ecosystems: the west Antarctic Peninsula. Philos Trans R Soc Lond B 362:67-94

Dugger KM, Ballard G, Ainley DG, Barton KJ (1996) Effects of flipper bands on foraging behavior and survival of Adélie penguins (Pygoscelis adeliae). Auk 123:858-869

Dunn MJ, Silk JRD, Trathan PN (2011) Post-breeding dispersal of Adélie penguins (Pygoscelis adeliae) nesting at Signy Island, South Orkney Islands. Polar Biol 34:205-214

Emmerson L, Southwell C (in press) Adélie penguin survival: age structure, temporal variability and environmental influences. Oecologia doi:10.1007/s00442-011-2044-7

Emslie SD, Fraser W, Smith RC, Walker W (1998) Abandoned penguin colonies and environmental change in the Palmer Station area, Anvers Island, Antarctic Peninsula. Antarct Sci 10:257-268

Folke C, Carpenter S, Walker B, Scheffer M, Elmqvist T, Gunderson L, Holling CS (2004) Regime shifts, resilience, and biodiversity in ecosystem management. Ann Rev Ecol Syst 35:557-581

Forcada J, Trathan PN (2009) Penguin responses to climate change in the Southern Ocean. Global Change Biol 15:1618-1630

Forcada J, Trathan PN, Reid K, Murphy EJ, Croxall JP (2006) Contrasting population changes in sympatric penguin species n association with climate warming. Global Change Biol 12:411-423

Fraser WR, Hofmann EE (2003) A predator’s perspective on causal links between climate change, physical forcing and ecosystem response. Mar Ecol Prog Ser 265:1-15

Frederiksen M, Daunt F, Harris MP, Wanless S (2008) The demographic impact of extreme events: stochastic weather drives survival and population dynamics in a long-lived seabird. J Anim Ecol 77:1020-1029

Froget G, Gautier-Clerc M, Le Maho Y, Handrich Y (1998) Is penguin banding harmless? Polar Biol 20:409-413

Gauthier-Clerc M, Gendner JP, Ribic CA, Fraser WR, Woehler EJ, Descamps S, Gilly C, Le Bohec C, Le Maho Y (2004) Long-term effects of flipper bands on penguins. Proc R Soc Lond B 271:S431-S436

Grosbois V, Gimenez O, Gaillard J-M, Pradel R, Barbraud C, Clobert J, Møller AP, Weimerskirch H (2008) Assessing the impact of climate variation on survival in vertebrate populations. Biol Rev Camb Philos Soc 83:357-39

Grosbois V, Thompson PM (2005) North Atlantic climate variation influences survival in adult fulmars. Oikos 109:273-290

Hewitt RP (1997) Areal and seasonal extent of sea-ice coverage off the northwestern side of the Antarctic Peninsula: 1979 to 1996. CCAMLR Sci 4:65-73

Hindell MA, Lea MA, Hull CL (1996) The effects of flipper bands on adult survival rate and reproduction in the Royal Penguin, Eudyptes schlegeli. Ibis 138:557-560

Hinke JT, Polito MJ, Reiss CS, Trivelpiece SG, Trivelpiece WZ (In review) Flexible reproductive timing can buffer reproductive success of Pygoscelis penguins in the Antarctic Peninsula region. Mar Ecol Prog Ser

Hinke JT, Salwicka K, Trivelpiece SG, Watters GM, Trivelpiece WZ (2007) Divergent responses of Pygoscelis penguins reveal a common environmental driver. Oecologia 153:845-855.

Jablonksi B (1985) The diets of penguins on King George Island, South Shetland Islands. Acta Zool Cracov 29:117-186

Jackson S, Wilson RP (2002) The potential costs of flipper bands to penguins. Funct Ecol 16:141-148

Jenouvrier S, Barbraud C, Weimerskirch H (2005) Long-term contrasted responses to climate of two Antarctic seabird species. Ecology 86:2889-2903

Jenouvrier S, Barbraud C, Weimerskirch H (2006) Sea ice affects the population dynamics of Adélie penguins in Terre Adélie. Polar Biol 29:413-423

Jenouvrier, S Caswell H, Barbraud C, Holland M, Strœve, Weimerskirch H (2009) Demographic models and IPCC projections predict the decline of an emperor penguin population. Proc Natl Acad Science USA 106:1844-1847

Jolly GM (1965) Explicit estimates from capture-recapture data with both death and immigration stochastic model. Biometrika 52:225-247

Kendall WL, Converse SJ, Doherty PF Jr., Naughton MG, Anders A, Hines JE, Flint E (2009) Sampling design considerations for demographic studies: a case of colonial seabirds. Ecol App 19:55-68

Kwok R, Comiso JC (2002) Spatial patterns of variability in Antarctic surface temperature: connections to the southern hemisphere annular mode and the Southern Oscillation. Geophys Res Lett 29:1-4 doi:10.1029/2002GL015415

Laake J (2011) RMark: R code for MARK analysis. R Package version 2.0.1

Laws RM (1985) The ecology of the Southern Ocean. Am Sci 73:26-40

Le Bohec C, Durant JM, Gauthier-Clerc M, Stenseth NC, Park YH, Pradel R, Grémillet D, Gendner JP, Le Maho Y (2008) King penguin population threatened by Southern Ocean warming. Proc Natl Acad Sci USA 105:2493-2497

Lebreton JD, Burnham KP, Clobert J, Anderson DR (1992) Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecol Monogr 62: 67-118

Lebreton JD, Clobert J (1991) Bird population dynamics, management, and conservation: the role of mathematical modeling. In: Perrins CM, Lebreton JD, Hirons GJM (eds) Bird Population Studies: Relevance to Conservation and Management. Oxford University Press, Oxford

Lescroël A, Bost C (2005) Foraging under contrasting oceanographic conditions: the gentoo penguin at Kerguelen Archipelago. Mar Ecol Prog Ser 302:245-261

Lescroël A, Dugger KM, Ballard G, Ainley DG (2009) Effects of individual quality, reproductive success and environmental variability on survival of a long-lived seabird. J Anim Ecol 78:798-806

Lewontin RC, Cohen D (1969) On population growth in a randomly varying environment. Proc Natl Acad Sci USA 62:1056-1060

Loeb VJ, Hofmann EE, Klinck JM, Holm-Hansen O, White WB (2009) ENSO and the variability of the Antarctic Peninsula pelagic marine ecosystem. Antarct Sci 21:135-148

Lynch HJ, Fagan WF, Naveen R, Trivelpiece SG, Trivelpiece WZ (2009) Timing of clutch initiation in Pygoscelis penguins on the Antarctic Peninsula: Towards an improved understanding of off-peak census correction factors. CCAMLR Sci 16:149-165

Lynch JH, Naveen R, Fagan WF (2008) Censuses of penguin, blue-eyed shag Phalacrocorax atriceps and southern giant petrel Macronectes giganteus populations on the Antarctic Peninsula, 2001-2007. Mar Ornithol 36:83-97

Meredith MP, King JC (2005) Rapid climate change in the ocean west of the Antarctic Peninsula during the second half of the 20th century. Geophys Res Lett 32, L19604, doi:10.1029/2005GL024042

Morris WF, Doak DF (2004) Buffering of life histories against environmental stochasticity: accounting for a spurious correlation between variabilities of vital rates and their contribution to fitness. Am Nat 163:579-590.

Nelson LJ, Anderson DR, Burnham KP (1980) The effect of band loss on estimates of survival. J Field Ornithol 51:30-38

Parkinson C (2004) Southern Ocean sea ice and its wider linkages: insights revealed from models and observations. Antarct Sci 16:387-400

Patterson DL, Easter-Pilcher AL, Fraser WR (2003) The effects of human activity and environmental variability on long-term changes in Adélie penguin populations at Palmer Station, Antarctica. In: Huiskes AHL, Gieskes WWC, Rozema J, Schorno RML, van der Vies SM, Wolf W (eds) Antarctic Biology in a Global Context, Backhuys Publishers, Leiden

Pitman R, Durban JW (2010) Killer whale predation on penguins in Antarctica. Polar Biol 33:158-1594

Pollock KH (1981) Capture-recapture models allowing for age-dependent survival and capture rates. Biometrics 37:521-529

Pollock KH, Hines JE, Nichols JD (1984) The use of auxiliary variables in capture- recapture and removal experiments. Biometrics 40:329-340

R Development Core Team (2010) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07- 0, URL http://www.R-project.org/

Reiss CS, Cossio AM, Loeb V, Demer DA (2008) Variation in biomass of Antarctic krill (Euphasia superba) around the South Shetland Islands, 1996-2006. ICES J Mar Sci 65:497-508

Robson DS (1969) Mark-recapture methods of population estimation. In: Johnson NL, Smith H Jr. (eds) New developments in survey sampling. John Wiley and Sons, New York

Sæther B-E, Bakke Ø (2000) Avian life history variation and contribution of demographic traits to the population growth rate. Ecology 81:642-653

Sandland RL, Kirkwood GP (1981) Estimation of survival in marked populations with possibly dependent sighting probabilities. Biometrika 68:531-541

Sandvick H, Erikstad KE (2008) Seabird life histories and climatic fluctuations: a phylogenetic-comparative time series analysis of North Atlantic seabirds. Ecography 31:73-83

Sandvick H, Erikstad KE, Barreett RT, Yoccoz NG (2005) The effects of climate on adult survival in five species of North Atlantic seabirds. J Anim Ecol 74:817-831

Seber GAF (1965) A note on multiple recapture census. Biometrika 52:249-259

Sladen WJL (1953) The Adélie penguin. Nature 171:952-955

Smith RC, Ainley D, Baker K, Domack E, Emslie S, Fraser B, Kennett J, Leventer A Mosley-Thompson E, Stammerjohn S, Vernet M (1999) Marine ecosystem sensitivity to climate change. BioScience 49:393-404

Solomon S, Qin D, Manning M, Marquis M, Averyt K, Tignor, MMB, Miller HL Jr, Chen Z (eds) (2007) Climate Change 2007. The physical science basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge

Stammerjohn SE, Martinson DG, Smith RC, Iannuzzi RA (2008a) Sea ice in the western Antarctic Peninsula region: Spatio-temporal variability from ecological and climate change perspectives. Deep-Sea Res Part II 55:2041-2058

Stammerjohn SE, Martinson DG, Smith RC, Yuan X, Rind D (2008b) Trends in Antarctic sea ice retreat and advance and their relation to El Niño – Southern Oscillation and Southern Annular Mode variability. J Geophys Res 113, C03S90, doi:10.1029/2007JC004269

Stearns SC (1992) The evolution of life histories. Oxford University Press, Oxford

Thomas ER, Marshall GJ, McConnell R (2008) A doubling of snow accumulation in the western Antarctic Peninsula since 1850. Geophys Res Lett 35: L01706, doi:10.1029/2007GL032529

Trathan PN, Croxall JP, Murphy EJ (1996) Dynamics of Antarctic penguin population in relation to inter-annual variability in sea ice distribution. Polar Biol 16:321-330

Trivelpiece WZ, Trivelpiece SG (1990) Courtship period of Adélie, gentoo, and chinstrap penguins. In: Davis JS, Darby JT, (eds), Penguin Biology. Academic Press, Inc, San Diego

Trivelpiece WZ, Trivelpiece SG, Volkman NJ (1987) Ecological segregation of Adélie, gentoo, and chinstrap penguins at King George Island, Antarctica. Ecology 68:351-361

Turner J (2004) The El Niño-Southern Oscillation and Antarctica. Int J Climatol 24: 1-31

Turner J, Colwell SR, Marshall GJ, Lachlan-Cope TA, Carleton AM, Jones PD, Lagun V, Reid PA, Iagovkina S (2005a) Antarctic climate change during the last 50 years. Int J Climatol. 25:279-294

Turner J, Comiso JC, Marshall GJ, Lachlan-Cope TA, Bracegirdle T, Maksym T, Meredith MP, Wang Z, Orr A (2009) Non-annular atmospheric circulation change induced by stratospheric ozone depletion and its role in the recent increase of Antarctic sea ice extent. Geophys Res Lett 36:doi:10.1029/GL07524

Turner J, Lachlan-Cope T, Colwell S, Marshall GJ (2005b) A positive trend in western Antarctic Peninsula precipitation over the last 50 years reflecting regional and Antarctic- wide atmospheric changes. Annals Glaciol 41:85-91

Tylianakis JM, Didham RK, Bascompte J, Wardle JA (2008) Global change and species interactions in terrestrial ecosystems. Ecol Lett 11:1351-1363

White GC, Burnham KP (1999) Program MARK: Survival estimation from populations of marked animals. Bird Study 46(Suppl.):120-139

Williams TD (1995) The penguins. Oxford University Press, Oxford

Veit RR, Pyle P, McGowan JA (1996) Ocean warming and long-term change in pelagic bird abundance with the California current system. Mar Ecol Prog Ser 139:11-18

Volkman NJ, Presler P, Trivelpiece W (1980) Diets of Pygoscelid penguins at King George Island, Antarctica. Condor 82:373-378

Yuan X (2004) ENSO-related impacts on Antarctic sea ice: a synthesis of phenomenon and mechanisms. Antarct Sci 16:415-425

CHAPTER 3: Integrating long-term demographic data with a matrix population model

ABSTRACT

Predicting population responses in changing environments is an important task for conservation biology. In the Antarctic Peninsula region, climate change has wrought large- scale changes throughout the marine ecosystem, including declines in abundance of Adélie

(Pygoscelis adeliae) penguins. Given expectations of continued warming, loss of sea ice, and more frequent anomalous events in the Antarctic environment, further reductions in Adélie penguin populations appear likely. To quantitatively estimate future population changes, I developed a stochastic matrix population model parameterized with 25 years (1985-2009) of data from the Copacabana colony on King George Island, Antarctica. For validation, the model was fit to census data by estimating correction factors for previously estimated survival rates. The fitted model provided a baseline for future projections. Recognizing the high degree of variability in survival rates, but no clear statistical relationships with a suite of environmental variables, survival rates during the projection period were randomly sampled from their historical distribution based on an environmental proxy modeled as a two-state

Markov chain. Monte Carlo simulation was used to estimate population trajectories across a range of progressively worse environmental conditions. Model projections suggested that the

Adélie penguin population will continue to decline if the frequency of years with poor adult survival remains at, or increases above, its current state. Under status quo conditions, the risk of local extirpation within 30 years was 1% but rose rapidly as the frequency of years with poor adult survival increased. Given large-scale physical changes in the environment, conservation actions may be insufficient to slow or reverse the rate of decline among Adélie penguins in the Antarctic Peninsula region.

100

INTRODUCTION

As environmental conditions change, predicting population responses is an important task for ecologists. Soberingly, estimating the risk of extinction due to climate change has become a common task in conservation biology (Thomas et al. 2004; Maclean and Wilson

2011) and an important component of managing endangered resources (Beissinger and

Westphal 1998). In polar marine ecosystems, extinction risks may be particularly acute given the rapid pace of warming and consequent loss of physical habitat in the form of sea ice. Sea ice is a necessary habitat for numerous marine mammals and seabirds, used as a substrate for, inter alia, reproduction, foraging, molting, and resting. Animals with life histories that have evolved a direct dependence on sea ice are of particular conservation concern (Croxall et al.

2002; Laidre et al. 2008), and recent studies on Pacific walrus (Odobenus rosmarus divergens;

Jay et al 2011), polar bears (Ursus maritimus; Armstrup et al. 2007) and emperor penguins

(Aptenodytes forsteri; Jenouvrier et al. 2009) suggested varying, but non-zero probability of extinction due to climate change-induced reductions in sea ice within 100 years.

In the Antarctic Peninsula region, home to numerous ice-dependent seabird and marine mammal species, the pace of climate change is particularly rapid (Vaughan et al. 2003) and large-scale changes throughout the marine ecosystem are evident. Positive trends in air and sea surface temperatures (Turner et al. 2005a; Meredith and King 2005) along with reductions in the extent and duration of winter sea ice (Stammerjohn et al. 2008) have accompanied potential declines in primary production (Ducklow et al. 2007) and up to 80% reductions (Atkinson et al. 2004) in the density of Antarctic krill (Euphausia superba), one of the main energetic resources for higher trophic-level predators in the region (Laws 1985;

Smetacek and Nicol 2005). Among such predators are the Adélie penguins (Pygoscelis adeliae), an ice-obligate seabird whose populations in the northern and western Antarctic

101

Peninsula region have declined by more than 50% since the mid 1970s (Forcada et al 2006;

Hinke et al. 2007; Lynch et al. 2008; Schofield et al. 2010). Those declines have been attributed largely to loss of sea-ice habitats during winter and reductions in krill availability

(Fraser and Hofmann 2003; Ducklow et al. 2007; Trivelpiece et al. 2011). Inter-governmental

Panel on Climate Change (IPCC) projections suggested that future warming, loss of sea ice, and an increased frequency of anomalous environmental events in the Antarctic Peninsula region is likely (Solomon et al. 2007). For Adélie penguins at the northern extent of their circumpolar range, further environmental change may hasten their rapid regional population declines.

For projecting population response to such future environmental conditions, quantifying the relationships between environmental indices and demographic rates is useful for parameterizing realistic dynamics in stochastic models. However, the statistical relationships between demographic rates and environmental drivers are not always clear, potentially because of poor quality demographic data (Beissinger and Westphal 1998) or synergistic effects of multiple climate and biological drivers that are difficult to identify a priori (Brook et al. 2008). This difficulty appears to arise for some populations of Adélie penguins. For example, the variation in survival rates among juvenile Adélie penguins in east

Antarctica (Emmerson and Southwell 2011) and among all juvenile and adult stages in South

Shetland Islands (previous chapter) were not well explained by observed variation in sea ice, air temperature, or a suite of other environmental indices thought to affect penguin populations. Regardless of the cause of apparent non-relationships of demographic rates and environmental variability, the ability to use projected climate indices, such as those from the

Intergovernmental Panel on Climate Change (IPCC), for modeling future population trends is necessarily restricted in such cases. Thus, an alternative method is needed to forecast likely future scenarios for such populations.

102

Here, I developed a stochastic model to facilitate quantitative predictions of population responses to changes in the frequency of years with poor survival. The model is based on the methods of Jenouvrier et al. (2009), who integrated demographic data and IPCC projections with a matrix population model (Caswell 2001) to estimate extinction risk for emperor penguins based on relationships between temperatures and adult survival rates

(Jenouvrier et al. 2005). The method developed here does not use climate projections, however. Rather, variation in adult and juvenile survival rates depends on future environmental states generated from a Markov chain that switches between states of good and poor survival. The method is motivated by the analyses of mark-recapture data (Chapter 2) that revealed typically high survival rates punctuated by years of low survival. In particular, episodic occurrences of low adult survival rates have disproportionately large effects on population dynamics among long-lived species (Sæther and Bakke 2000; Weimerskirch 2002) and can cause rapid declines in abundance (Frederiksen et al. 2008). Given climate projections with continued warming, loss of sea ice in the Antarctic Peninsula region, and expectation of more frequent anomalous events (Solomon et al. 2007), changes in the frequency of years with poor survival among adult and juvenile penguins may be likely.

To assess future Adélie penguin population growth in the Antarctic Peninsula region under conditions with an increasing frequency of years with poor survival, I integrated long- term demographic data, including 25 years of mark-recapture data (previous chapter) and age- specific data on chick production, with a stochastic matrix population model. Summarizing data from long-term studies within a population model provides a robust framework for evaluating how historical changes in survival and reproductive indices influenced observed population size at the study site and enables forecasting of short-term population trends across a range of plausible environmental regimes. My goals are twofold. First, I ask whether the model, parameterized with the survival rates estimated from mark-recapture data and

103 reproductive rates from monitoring studies, adequately characterized the observed changes in population over the last 25 years. Secondly, I estimate the risk of local extirpation as the probability of the population declining to less than 10% of current abundance over a 30 year time frame, across a range of future environmental conditions. A 90% decline in the abundance of the breeding population over 30 years is equivalent to the greatest rate of reductions observed in breeding populations in the Antarctic Peninsula region over the last 3 decades (Schofield et al. 2010) and corresponds to the time frame used by the Commission for the Conservation of Antarctic Marine Living Resources (CCAMLR) to manage risks to the ecosystem that may arise from commercial fishing activities (CCAMLR 1980).

MATERIALS AND METHODS

Matrix model structure

I developed a post-breeding matrix population model (Caswell 2001) to estimate future population sizes under differing expectations of environmental conditions. The model is generalized to allow for constant, age-specific, time-specific, or state-specific variation in key components of Adélie life cycle, including recruitment, breeding success, survival, and the ability to move between breeding and non-breeding states. The model incorporates 6 age- classes (ages 0 through 5), a breeding adult stage and a non-breeding adult stage (Fig. 3-1a).

Ages 0 through 5 represent birds that have not yet bred, but birds in ages 3 through 5 can breed and recruit into the adult breeding stage. All individuals in the populations are assumed to recruit to the adult breeding stage after age 5. Once recruited to the adult breeding stage, birds can skip breeding periodically by entering a non-breeding adult stage.

104

Transitions between age-classes are described by the stage-specific (a) and time- specific (t) survival rates (Sa,t) estimated with the mark-recapture models presented in Chapter

2. Transitions between the pre-breeding age-classes are described by juvenile survival rates only. Recruitment to the adult breeding state is modeled as Sa,t*Pbi, where Pbi is the probability of breeding for the first time at age i. Birds that are older than the minimum age of first breeding can remain non-breeding birds with probability Sa,t*( 1-Pbi). After recruiting to the adult breeding stage, the probability of reproducing in subsequent years is Sa,t*Pbα, where

Pbα is the probability of breeding once adult status has been achieved. Adults in the breeding stage can skip breeding with probability (1-Pbα). First-time breeding birds contribute offspring to the population with a fecundity defined as Sa,t*Pbi*BSi, where BSi is the average reproductive success of first time breeders of age i. Adult birds with prior breeding experience have a fecundity defined as Sa,t*Pbα*BSa,t, where BSa ,t is the average reproductive success of all adult birds at time t.

The resulting model forms a generalized matrix population model

nt+1=At*nt Eq. 1

where At is a time-dependent population projection matrix (Fig. 3-1b) and n is a vector that contains the abundance of each stage represented in the model at time t. The time varying At is based on the annual estimates of survival rates for adult and juveniles and reproductive success of adult birds with prior breeding experience, whose values are updated in the transition matrix each time step. Parameters for reproductive success of first-time breeders, the probability of breeding for the first time, and the probability of breeding once adult status has been achieved are held constant. The resulting sequence of transition matrices were used to model historical population trends before projection of the model under alternative

105 environmental scenarios. The model was coded and run in the R environment (R Core

Development Team 2010).

Parameterization of the model

Survival rates Si,t were derived from the estimates from the best-fitting mark-recapture model described in detail in the previous chapter. Therefore, estimates of juvenile survival are used to model the transitions from age 0 to age 1 and from age 1 to age 2. Adult survival rates were used to model the transitions between all older stages.

Data on reproductive success and breeding probability derive from two long-term studies. Annual estimates of breeding success of known-age individuals were monitored in the

“known-age” study. Known-age individuals were banded as chicks as described in the previous chapter and monitored for reproductive success throughout their lives. The other study monitored between 20 and 40 sets of 5 adjacent nests with individuals of unknown-age

(hereafter the “reproductive” study). The sample sizes in the reproductive study were reduced over time to accommodate a declining Adélie population. The nests in the reproductive study were distributed throughout the colony and each mate was banded for identification prior to egg laying. Birds that were banded as part of the reproductive study were observed for reproductive success for one year only, but the bands were left on the birds to ensure that those individuals were not monitored in subsequent reproductive studies. The reproductive study is therefore assumed to represent a population-level reproductive success study for adult birds in the breeding population. For both studies, breeding attempts were monitored from egg laying until successful crèche or nest failure, whichever occurred first. A successful crèche was defined at the first observation of an unattended chick in the nest, which typically occurs when chicks are between three and four weeks old. At this time chicks are typically large

106 enough to deter avian predators. Reproductive failure was assumed following the death or disappearance of an egg or chick from the nest prior to crèche.

For estimates of reproductive success of first-time breeders (BSi), I used data from the known-age study. Due to relatively small sample sizes (median = 5) of first-time, known-age breeders in any given year, an average age-specific reproductive success for first-time breeders was calculated as the total number of chicks crèched per total number of individuals that attempted to breed for the first time in each . Reproductive success of adults with prior experience (BSa,t) was estimated for each year of the reproductive study as the total number of chicks crèched per number of nests monitored.

The probability of breeding for the first time at age i (Pbi) was estimated from the mark-recapture data of known-age individuals, described in Chapter 2, as the proportion of birds that bred for the first time at age i relative to the total number of birds of age i known to be alive. The number of animals known to be alive, but not yet breeding, was determined from the total number of individuals encountered from each cohort during the study period. The probability of breeding as an adult (PBa) was estimated from the life-long breeding histories of known-age individuals. The breeding history for all birds that bred multiple times (N=296) was converted to an encounter history with a value of 1 indicating alive and breeding and a value of 0 indicating either alive and not breeding or possibly not alive. With this format, the each observed breeding attempt is equivalent to a live recapture in standard mark-recapture models. Therefore, I used Program MARK (White and Burnham 1999) to estimate a constant breeding probability for adult birds.

The initial vector of population abundance for all stages (nt=1) was based on the the census of adult birds in 1985 and the stable age distribution of the first transition matrix, A t=1, estimated as the right eigenvector.

107

Adjusting survival rates for potential band loss in the marked population

If band loss occurred during the mark-recapture study or if banding induced additional mortality, estimates of survival rates from mark-recapture modeling may be biased low (Conn et al. 2004, Dugger et al. 2006). Comparisons of the survival rates estimated in Chapter 2 with published estimates of Adélie penguin survival suggested that the estimated juvenile survival rates are comparable to other populations, but that adult survival rates were biased low (e.g.

Emmerson and Southwell 2011). Therefore, to adjust adult survival for this apparent negative bias, I used a generic correction factor. The correction factor was estimated by embedding the population model in an optimization routine and fitting the population model to adult census data via maximum likelihood. Following Arnason and Mills (1981), the correction factor ( δ )

ˆ for each band type ( f ) was used to derive a corrected survival rate, S ,tf , as:

ˆ S ,tf S ,if = . Eq. 2 δ f

The corrected survival estimates were constrained to remain in the interval [0,1] with a logit transformation. The logit transformation also usefully allows a single estimate of δf in logit space to result in a time-varying correction factor for survival rates of each band type in the transformed space. Such a time-varying correction factor may be preferential to a fixed correction factor, since banding effects on survival have been suggested to vary with respect to the environmental conditions that are experienced by a banded bird (e.g. Dugger et al. 2006).

To compare a fixed correction factor versus a variable correction factor, as estimated above, I ran the model with the fixed retention rates that were estimated from a double banding study

108

(Chapter 2, Appendix 2-1). The corrected estimates of adult survival from the fitted model were used to project future population sizes.

Proxy environmental indices for variation in future adult survival

Environmental proxies for juvenile survival and adult survival were modeled stochastically as a two-state Markov chains. In each chain, the environmental state at time t was determined by two parameters, g and p. The values of g and p determine two aspects of the future environmental regime. First, the autocorrelation between time steps, quantified as r

= 1-g-p (Caswell 2001), determines the rate of switching between environmental states. Larger positive correlations lead to long runs of a single state, while larger negative correlations lead to more frequent switching between states. Second, g and p determines the expected frequency, ω, of states as ω=g/(g+p); g > p yields a higher proportion of good states, while g

< p results in more frequent poor states. To estimate the risk of local extirpation, I ran the model across a range of expected proportions of years with poor survival states, from p = 0 to p = 1, with corresponding values of g set to achieve the level of autocorrelation observed in the historical survival rates. The sensitivity of the simulations to the auto-correlation structure from the historical survival rates was tested by setting r = ± 0.2 and r = ± 0.5 and re-running the model across the range of expected poor survival states.

Projecting future population sizes

For each time step in the projection period, adult and juvenile survival rates were randomly sampled from the distribution of survival rates in the fitted model based on a random quantile, q, generated a uniform distribution. If the state in the environmental index at

109 time t was good, q was drawn from the upper 50th percentile of the uniform distribution. If the environmental state was poor, q was drawn from the lower 50th percentile. A survival rate for the projection time step t was then extracted from the qth quantile of survival rates from the fitted model. The randomly sampled survival rates of juvenile and adult stages at time t were incorporated into the projection matrix At. All other parameters in At were held at their average historical values.

In total, three sets of projection scenarios were simulated. First, the environmental proxy for juvenile survival was parameterized to vary according to historical levels (r = -0.2,

ω = 0.5, g = 0.6, p = 0.6), while the expected proportion of years with poor survival rates of adults varied stochastically. Second, the environmental proxy for adult survival rates was fixed at historical levels of variability (r = 0, ω = 0.66 ,g = 0.66, p = 0.34) while the expected proportion of years with poor survival rates of juveniles was varied. Finally, the expected proportion of years with poor survival rates of juveniles and adults were varied simultaneously. The scenarios were used to estimate the sensitivity of extirpation risk in the population to the plausible future scenarios of variability in adult and juvenile survival rates.

For all simulations of future population size, I used a projection period of 30 years and conducted 1000 Monte Carlo trials for each scenario of future environmental states.

From the collection of model projections, I calculated two quantities to summarize population status. First, the stochastic generation of expected future population trajectories allows an assessment of the risk of extirpation at the local breeding colony. A metric for assessing the risk of extirpation of the breeding population is the proportion of trajectories that decline to 0 over the projected period (Hildenbrandt et al. 2005). Here, I used a more conservative level of 90% reductions in breeder abundance relative to the estimated 2009 population size (2577) to estimate the risk of local extirpation (e.g. Jenouvreir et al. 2009).

Detection of a high probability of risk of local extirpation over a relatively short time frame

110 is applicable to the management protocols of Antarctic living marine resources, which seeks to minimize the risk “of changes in the marine ecosystem which are not potentially reversible over two or three decades” (CCAMLR 1980). Second, the mean annual rate of population change can be calculated as

/1 T  N   T  Eq. 3 λ =    N 0  where N0 is the size of the adult breeding population at the start of the projection period and

NT is the adult breeding population size at the end of the projection period T. Whereas the risk of local extirpation on a 30 years horizon may be small, consistently negative population growth rates would indicate that further population decline is likely.

RESULTS

Reproductive success and breeding probability

Reproductive success and the probability of breeding increased with age (Table 3-1).

The average reproductive success of novice 3-year olds was 0.28 ± 0.04 chicks per nest and increased to 0.52 ± 0.07 chicks per nest for birds that advanced to the adult stage after age 5.

Average reproductive success of adults was 0.91 ± 0.31 chicks per nest (range = 0.31 ± 0.19 to

1.49 ± 0.25) and exhibited no linear trend (F1,21 = 0.32, P = 0.58) during the study period.

Similarly, conditioned on being alive with no prior breeding experience, the probability of breeding for the first time at age 3 was 0.06 and increased to 0.46 for 5-year olds (Table 3-1).

Adults with prior breeding experience returned to breed in their natal colony in consecutive years with a higher probability (Pba = 0.94). For model simulations, the average reproductive

111 success of adult birds from all years was used for the annual estimate of reproductive success in 1985.

Model fitting

Simulations with the estimates of reproductive success, probability of breeding, and survival rates of juvenile (aged 0-2) and adult (aged 3+) penguins from long-term mark- recapture data suggested that estimation of correction factors for adult survival rates was required to reproduce the population trajectory. Without any correction factors, the model ran to extinction within 10 years (Fig. 3-2a). Applying the fixed retention rates that were estimated for adult birds from the double banding study for aluminum (δ = 0.6) and stainless steel bands (δ = 0.97; Chapter 2, Appendix 2-1) also failed to fit to the data (Fig. 3-2b).

Moreover, when the estimated band retention rate of 0.6 was used for all years with aluminum bands, some survival rates exceeded 1.0, which is not biologically feasible. The model provided a better fit to the data when the correction factors were estimated by the model.

Consistent with the double banding study, the estimated correction factor for stainless steel bands was smaller than for aluminum bands (Table 3-2), but the estimated correction factors were generally larger than the respective estimated retention rates from the double banding study, particularly for the survival rates estimated from birds with stainless steel bands.

With the estimated correction factors, the resulting model fit captured the steep decline in breeder abundance from 1989-1991 and tracked the continual decline of the population in the late 2000s. The model also reproduced the high degree of variation in chick production and the general decline in total chick production observed within the colony (Fig.

3-2b). However, the model did not capture the largest reproductive events that occurred prior to 2000. Given the model fit, average adult survival probabilities over the last 25 years were

112

0.84 ± 0.09, compared to mean juvenile survival rates of 0.49 ± 0.13. The corrected adult survival rates retained the pattern of relatively high rates punctuated by several years of relatively lows survival. In particular, survival rates in 8 of 24 years were less than the overall average. However, 5 of the 8 years with low adult survival occurred during the last 10 years of the study, a period of persistent population decline. For juveniles, survival rates were more normally distributed, with 50% of years being characterized as having survival less than the mean.

Future projections and risk of local extirpation

Example results from two scenarios of Monte Carlo simulations of variation in the environment affecting adult Adélie penguin survival rates are illustrated in Fig. 3-3.

Simulations assuming that future environmental conditions remain equivalent to the historical conditions (1/3 of years with poor survival) yielded a median population trajectory with a negative trend and 1% risk of the population falling below the local extirpation threshold within 30 years (Fig. 3-3a). If the frequency of years with poor survival among adults was increased to 50%, consistent with the frequency observed for survival rates among adult

Adélie penguins since 1999, the risk of local extirpation within three decades increased to 8%

(Fig. 3-3b).

The population projections were less sensitive to the level of auto-correlation used to define the proxy environmental index than to the expected proportion of years with poor survival. Whether positive or negative with small or large magnitude autocorrelation in the projected environmental state, all simulations tracked the risk of extirpation estimated from an autocorrelation of r = 0. Examples of this variability are plotted for simulations with a variable environment that affects adult survival rates only (Fig. 3-4a).

113

Across the set of scenarios, increasing the frequency of years with poor survival among juvenile Adélie penguins resulted in the lowest increase in risk of local extirpation.

Extirpation risk was more sensitive to an increase in the frequency of years with poor adult survival, and the risk rose steadily with an increase in variability just beyond current conditions. When survival rates of both juveniles and adults were subject to an increasing frequency of poor survival, the risk of local extirpation within 30 years increased quickly from the relatively low risk under historical condition to near 80% under a doubling of the frequency of years with poor survival. Despite relatively low risk of local extirpation within

30 years under scenarios with fewer year of poor survival than exhibited historically, most projections were characterized by negative population growth rates ( λ < 1; Fig. 3-4b). The increasingly negative population growth rates under more variable survival conditions indicate that long-term declines in the population are near certain. The relatively rapid increase in extirpation risk at levels beyond current historical average also matches the observation of persistent population declines since 1999, where the frequency of years with poor survival has increased over the long-term average.

DISCUSSION

Model projections suggested that the population of Adélie penguins will continue to decline if the frequency of years with poor adult survival remains at, or increases above, its current state. Under status quo conditions, the risk of >90% declines within 30 years was estimated at only 1%, but increased rapidly with further increases in the frequency of years with poor adult survival. Thus, current variability in adult survival rates among Adélie penguins is sufficient to drive continued declines in the breeding population. Given expectations of an increasing frequency of anomalous events that can trigger acute mortality

114 events and more chronic stressors that include warming trends and gradual loss of sea ice in the Antarctic Peninsula region (Solomon et al. 2007), it is highly likely that population growth rates of Adélie penguins in the rapidly warming Antarctic Peninsula region will remain negative.

A key uncertainty concerning the rate of population decline in this study arises from the lack of correlation between indices of environmental conditions and estimates of penguin survival rates from the available mark-recapture data. Without such a statistical underpinning, predictions of population trends based on IPCC-type climate projections are difficult, despite widespread recognition of the importance of environmental factors on Adélie penguins (e.g.,

Ainley 2002; Croxall et al. 2002; Forcada and Trathan 2009; Trivelpiece et al. 2011). A post- hoc correlation analysis of model-fitted survival rates with the suite of environmental conditions identified in Chapter 2 suggested that adult survival rates may be more strongly influenced positively by ice extent and negatively by air temperatures than estimated from the raw estimates from mark-recapture analyses. However, Bonferroni corrections for multiple comparisons lead to the conclusion that there is no correlation between the estimated survival rates and the environmental indices. To circumvent the lack of relationships, a stochastic

Monte Carlo method was chosen to examine whether the variability observed in historical data was sufficient to drive further population decline. Implicit in this method is an assumption that the historical distribution of survival rates adequately applies to future conditions (Coulson et al. 2001); this cannot be known with certainty. Nonetheless, an advantage of constraining the projections with historical distributions is that the assessment of future population trends is based on an observed range of behavior, rather than on correlations between demographic rates and environmental indices that may not remain constant under changing environmental conditions (Scheffer et al. 2009). Thus, the model projections may be considered as conservative measures of future population trends and risk of local extirpation.

115

Other factors that contribute to a decline in population size

The analysis of local population collapse considered here recognized only future variation in survival rates. The focus on the effect of variation in adult survival rates on extinction risk is useful for long-lived species like penguins because population growth rates are typically most sensitive, as measured by elasticity (de Kroon et al. 1986) to changes in adult survival rates (Sæther and Bakke 2000). The risk of local extinctions, however, may be substantially higher because of additional effects of stochastic variation in breeding success at small population sizes, depensatory dynamics arising from predation or other Allee affects that reduce per-capita growth rates under small population sizes, and catastrophic events that can rapidly decrease population sizes (Mangel and Tier 1994; Gerber and Hilborn 2001;

Ovaskainen and Meerson 2010).

With respect to Adélie penguins, changes in environmental conditions may contribute to such additional risk of local extirpation beyond the levels estimated here. In particular, while long-term reproductive success of Adélie penguins at the study colony has not exhibited a significant trend (Hinke et al. 2007), estimates of reproductive success during years of extreme snowfall (e.g., 2007, personal observation) produced the lowest chick production rates during the study period (Table 3-1). Trends for increases in the frequency and accumulation rates of snowfall in the Antarctic Peninsula region (Turner et al. 2005b; Thomas et al. 2008) can reduce reproductive success of Adélie penguins by burying occupied nests and or prohibiting access to the small rocks and pebbles that are necessary for nest construction

(Patterson et al. 2003; Boersma 2008). Additionally, at small population sizes, avian predators like brown skuas (Catharacta antarctica lonnbergi.) and giant petrels (Macronectes giganteus) may substantially reduce reproductive success (Emslie et al. 1995). As colonies

116 shrink, the ratio of peripherally to centrally located nests increases, and avian predators a generally more successful when depredating peripheral nests (Emslie et al 1995). Increased occurrences of reproductive failure owing to adverse environmental conditions or an increased per-capita predation rate could potentially reduce recruitment into the breeding population, further reducing population growth rates. Thus, smaller colonies of Adélie penguins are generally more at risk to collapse than larger colonies. Anecdotally, during the 2009/10 breeding season, an entire cohort of chicks in one small subcolony of gentoo penguins

(Pygoscelis papua) near the Copacabana colony was eliminated by skua predation and, in the following year, no adults bred in the subcolony (WZT, pers comm). While the risk of climate- induced extirpation in the near term under current conditions is low for Adélie penguins at the

Copacabana colony, future declines in abundance may enable other factors to assume a relatively more important role in changing the rate of population decline.

Emigration from the warming Antarctic Peninsula region could also contribute to the regional observations of population decline. While philopatry is a hallmark of the Adélie penguin life history (Ainley 2002) and adult males may return to the same nest location each year (Trivelpiece and Trivelpiece 1990), Dugger et al. (2010) demonstrated that, under adverse conditions associated with heavy sea ice and large icebergs in the Ross Sea, adults exhibited low dispersal rates ( > 3%) between adjacent colonies. Though relatively low, such dispersal is consistent with geologic records of colony abandonment and re-colonization associated with glaciation events around the Antarctic coast (Emslie et al. 2007). The dispersal rate of juvenile penguins, which may spend 3 to 4 years at sea before breeding for the first time, is also unknown. Such mobility among adults and uncertainty about juvenile dispersal suggests that degradation of conditions in the Antarctic Peninsula region could prompt a contraction of the Adélie penguin breeding range to more southern areas. The available mark- recapture data, however, are unable to identify whether such a range contraction could be due

117 to emigration because of limited search effort for tags anywhere other than the study colony.

Opportunistic sightings of banded individuals are rare, although at least one banded animal from the study site has been resighted in a large breeding colony on Paulet Island, some 200 km SE of the study colony (H. Lynch, pers. comm.). Future studies to investigate unique genetic markers may help resolve uncertainty about the role that emigration from the Antarctic

Peninsula plays in the current period of population decline.

From local decline to regional concern

The results presented here provide the first quantitative predictions of future population trends for Adélie penguins in the Antarctic Peninsula region. It should be noted that the Copacabana colony contains a small percentage of the global population of Adélie penguins, estimated to be near 2.5 million individuals (Ainley 2002). Thus, skepticism on scaling results about population declines from the study colony to the broader global population is warranted. In particular, the declines in Adélie penguin abundance that are evident in the Antarctic Peninsula region (Forcada et al. 2006; Lynch et al. 2008; Trivelpiece et al. 2011) contrast the generally stable to increasing populations of Adélie penguins elsewhere in Antarctica (Ainley 2002). Thus, the results are most relevant to the Antarctic

Peninsula region, which represents the northern-most reaches of the circumpolar range of

Adélie penguins, but home to roughly a third of their global population (Ainley 2002).

General declines in the number of breeding adults at long-term monitoring sites (e.g., Forcada et al. 2006; Ducklow et al. 2007; Carlini et al. 2007) and from opportunistic censuses throughout the Antarctic Peninsula region (Lynch et al. 2008), however, suggest that the specific details from demographic studies at the Copacabana colony may be generally applied across the region. Thus, declines in abundance of Adélie penguins in the Antarctic Peninsula

118 region are expected to continue, with a relatively high probability of >90% declines in abundance within three decades.

Implications for the Conservation of Antarctic Living Marine Resources

As noted above, one of the primary mandates of the Convention on the Conservation of Antarctic Living Marine Resources (CCAMLR), articulated in Article II of the Convention, is to minimize the risk of changes in the marine ecosystem due to harvesting that are not potentially reversible within two or three decades (CCAMLR 1980). The CCAMLR is chiefly concerned with managing a growing fishery for Antarctic krill, a staple of the Adélie penguin diet in the Antarctic Peninsula region (Polito et al. 2011). In a precautionary sense, a precipitous decline in predator population might therefore trigger a management action aimed to mitigate the negative population trends. However, the difficult question is whether any management action regarding krill harvesting would affect the population growth rate of

Adélie penguins. While functional relationships between krill abundance and indices of reproductive success and offspring growth among krill-dependent predators and krill have been described and suggest improved growth and offspring production with higher krill density (Reid et al. 2005), the relationships between krill abundance and the vital rates that largely influence population growth rates remain elusive. More commonly, survival rates of

Adélie penguins have been statistically linked to indices of physical conditions like regional ice extent (e.g. Ballerini et al. 2009; Emmerson and Southwell 2011) rather than indices of food availability. If variation in the size of Adélie penguin populations is attributable to changes in the availability of krill (Laws 1977; Trivelpiece et al. 2011), then a management action to minimize the effect of krill harvesting might be effective. However, climate-change induced increases in temperatures and consequent loss of sea ice will also impact the

119 productivity and distribution of Antarctic krill (Loeb et al. 1997), which depend on winter sea ice to complete their life cycle (Nicol 2006). Given the underlying physical changes in the environment and expectation for further degradation of the cyrosphere (Solomon et al. 2007) , a fisheries management action may be insufficient to slow or reverse the rate of decline among

Adélie penguins in the Antarctic Peninsula region.

ACKNOWLEDGMENTS

Many thanks to W. and S. Trivelpiece, and the numerous field biologists whose commitment to long-term studies in the Antarctic made this study possible. The U.S. Antarctic

Ecosystem Research Division at the Southwest Fisheries Science Center, National Oceanic and Atmospheric Administration and grants from the National Science Foundation to W.Z.

Trivelpiece provided logistical and financial support. Additional support was generously provided by the Lenfest Oceans Program at the Pew Charitable Trusts.

120

Tabel 3-1: Mean reproductive success (± 1 standard deviation) and breeding propensity of first time breeders (ages 0-6+) and adults. Reproductive success data for 1985 was not measured. Therefore, the average adult reproductive success is assumed for 1985 in all model runs.

Reproductive Success (BS) Breeding propensity (Pb) or stage Year Mean ± SD N Proportion N 1 All 0 0 0 0 2 All 0 0 0 0 3 All 0.28 ± 0.04 108 0.05 151 4 All 0.36 ± 0.02 306 0.33 400 5 All 0.48 ± 0.04 158 0.46 178 6+ All 0.52 ± 0.07 44 0.94 296 Adult (average) All 0.91 ± 0.31 2988 0.94 296 Adult 1985 NA NA 0.94 296 Adult 1986 1.17 ± 0.17 155 0.94 296 Adult 1987 0.59 ± 0.29 189 0.94 296 Adult 1988 0.91 ± 0.40 200 0.94 296 Adult 1989 0.55 ± 0.20 200 0.94 296 Adult 1990 0.31 ± 0.19 200 0.94 296 Adult 1991 0.73 ± 0.17 200 0.94 296 Adult 1992 0.75 ± 0.28 201 0.94 296 Adult 1993 1.18 ± 0.18 100 0.94 296 Adult 1994 0.95 ± 0.05 100 0.94 296 Adult 1995 1.26 ± 0.28 100 0.94 296 Adult 1996 0.62 ± 0.20 100 0.94 296 Adult 1997 1.49 ± 0.25 100 0.94 296 Adult 1998 1.39 ± 0.15 100 0.94 296 Adult 1999 0.95 ± 0.11 100 0.94 296 Adult 2000 1.28 ± 0.20 100 0.94 296 Adult 2001 0.50 ± 0.13 100 0.94 296 Adult 2002 0.92 ± 0.16 100 0.94 296 Adult 2003 0.87 ± 0.12 100 0.94 296 Adult 2004 1.01 ± 0.34 98 0.94 296 Adult 2005 1.03 ± 0.30 100 0.94 296 Adult 2006 0.92 ± 0.36 100 0.94 296 Adult 2007 0.41 ± 0.19 100 0.94 296 Adult 2008 1.06 ± 0.38 100 0.94 296

Aluminum Stainless steel Correction factor Survival Correction factor Survival Weights Final estimates Year Juvenile Adult Juvenile Adult Juvenile Adult Juvenile Adult Juvenile Adult Juvenile Adult 1985 1 0.48 0.42 0.84 NA NA NA NA 1.00 1.00 0.42 0.84 1986 1 0.48 0.40 0.84 NA NA NA NA 1.00 1.00 0.40 0.84 1987 1 0.81 0.42 0.96 NA NA NA NA 1.00 1.00 0.42 0.96 1988 1 0.67 0.63 0.92 NA NA NA NA 1.00 1.00 0.63 0.92 1989 1 0.31 0.74 0.71 NA NA NA NA 1.00 1.00 0.74 0.71 1990 1 0.31 0.35 0.71 NA NA NA NA 1.00 1.00 0.35 0.71 1991 1 0.71 0.49 0.94 NA NA NA NA 1.00 1.00 0.49 0.94 1992 1 0.71 0.57 0.94 NA NA NA NA 1.00 1.00 0.57 0.94 1993 1 0.59 0.40 0.90 NA NA NA NA 1.00 1.00 0.40 0.90 1994 1 0.49 0.44 0.84 NA NA NA NA 1.00 1.00 0.44 0.84 1995 1 0.66 0.53 0.92 NA NA NA NA 1.00 1.00 0.53 0.92 1996 1 0.36 0.34 0.75 NA NA NA NA 1.00 1.00 0.34 0.75 1997 1 0.66 0.68 0.92 NA NA NA NA 1.00 1.00 0.68 0.92 1998 1 0.71 0.39 0.94 1 NA 0.39 NA 0.99 1.00 0.39 0.94 1999 1 0.30 0.40 0.70 1 NA 0.40 NA 0.26 1.00 0.40 0.70 2000 1 0.56 0.46 0.88 1 0.60 0.46 0.81 0.46 0.46 0.46 0.84 2001 NA 0.52 NA 0.86 1 0.56 0.58 0.79 0.00 0.37 0.58 0.81 2002 NA 0.72 NA 0.94 1 0.76 0.32 0.88 0.00 0.10 0.32 0.89 2003 NA NA NA NA 1 0.37 0.60 0.68 0.00 0.00 0.60 0.68 2004 NA NA NA NA 1 0.59 0.74 0.80 0.00 0.00 0.74 0.80 2005 NA NA NA NA 1 0.48 0.38 0.74 0.00 0.00 0.38 0.74 2006 NA NA NA NA 1 0.78 0.59 0.89 0.00 0.00 0.59 0.89 2007 NA NA NA NA 1 0.74 0.58 0.87 0.00 0.00 0.58 0.87 2008 NA NA NA NA 1 0.71 0.58 0.86 0.00 0.00 0.58 0.86 Average 1.00 0.53 0.47 0.86 1.00 0.61 0.50 0.81 0.49 0.84 St. dev 0.00 0.16 0.12 0.09 0.00 0.14 0.13 0.07 0.13 0.09 121

122

A. Model structure

NB B

2 3 4 5

Breeding Survival transition

0 0 BSi*Sa,t*Pbi BSi*Sa,t*Pbi BSi*Sa,t*Pbi BSi*Sa,t*Pbi BSa,t*Sa,t*Pbα BSa,t*Sa,t*Pbα

Sa,t 0 0 0 0 0 0 0

0 Sa,t 0 0 0 0 0 0

0 0 Sa,t*(1-Pbi) 0 0 0 0 0 0 0 0 S *(1-Pb ) 0 0 0 0 nt+1 = a,t i nt 0 0 0 0 Sa,t*(1-Pbi) 0 0 0

0 0 0 0 0 0 Sa,t*(1-Pbα) Sa,t*(1-Pbα)

0 0 Sa,t*Pbi Sa,t*Pbi Sa,t*Pbi Sa,t*Pbi Sa,t*Pbα Sa,t*Pbα

Fig. 3-1: Diagram of matrix model structure and transition matrix parameterization. A). Model structure represented as a diagram with transitions between immature birds (ages 0 – 2), mature but virgin birds (3-5), and breeding (B) and non-breeding (NB) mature birds. B) The matrix model equation, with the transition matrix in parentheses. Transition to older age- classes and reproduction are modeled as functions of age specific survival (Si), age-specific breeding success (BSi) and age-specific breeding propensity (Pbi). n represents the vector of population size at time t for each i in the model.

123

Fig. 3-2: Model fits to number of nests (A) and model predictions of chick production (B). Data are plotted as solid diamonds. Model fit using uncorrected adult survival rates are plotted as dotted lines. Model fits using retention rates estimated from the double banding study are plotted as the dashed line. Model fits using estimated correction factors for adult survival are plotted as solid lines.

124

Fig. 3-3: Future projections under two future scenarios for survival of adults. A) Future conditions remain the same as historical (33% of years with poor survival). B) Future conditions with an increased frequency of poor survival (50%). Trajectories that declined >90% relative to the 2009 census are marked in gray. The overall median from all trajectories is marked in white.

125

Fig. 3-4: Probability of local-extirpation and mean population growth rates of Adélie penguins under future scenarios of environmental variability. A) Probability of local extirpation for juvenile, adult and full projection scenarios over a 30-year projection. The historical probability of years with poor survival is marked with the vertical dashed line. The effect of different degrees of auto-correlation in the survival data is displayed for positive (circles) and negative (triangles) autocorrelation values of r=0.2 or -0.2 (open points) and r=0.5 or -0.5 (closed points) on the adult projection scenario. B) Boxplot of population growth rates from 30-year projections under different expectations of the probability of poor survival events for juvenile, adult and full projection scenarios.

126

REFERENCES

Ainley DG (2002) The Adélie penguin: Bellwether of climate change. Columbia University Press, New York

Amstrup SC, Marcot BG, Douglas, DC (2007) Forecasting the range-wide status of polar bears at selected times in the 21st century. USGS Science Strategy to Support US Fish and Wildlife Service Polar Bear Listing Decision. US Geological Survey, Reston

Arnason AN, Mills KH (1981) Bias and loss of precision due to tag loss in Jolly-Seber estimates for mark-recapture experiments. Can J Fish Aquat Sci 38:1077-1095

Atkinson A, Siegel V, Pakhomov E, Rothery P (2004) Long-term decline in krill stock and increase in salps within the Southern Ocean. Nature 432:100-103

Ballerini T, Tavechia G, Olmastroni S, Pezzo F, Focardi S (2009) Nonlinear effects of winter sea ice on the survival probabilities of Adélie penguins. Oecologia 161:253-265

Beissinger SR, Westphal MI (1998) On the use of demographic models of population viability in endangered species management. J Wildlife Manage 62:821-841

Boersma PD (2008) Penguins as marine sentinels. Bioscience 58:597-607

Brook BW, Sodhi NS, Bradshaw CJA (2008) Synergies among extinction drivers under global change. Trends Ecol Evol 23:453-460

Carlini AR, Coria NR, Santos MM, Negrete J, Juares MA, Daneri GA (2009) Responses of Pygoscelis adeliae and P papua populations to environmental changes at Isla 25 de Mayo (King George Island). Polar Biol 32:1427-1433

Caswell H (2001) Matrix population models: Construction, analysis, and interpretation. Sinauer Associates, Inc. Sunderland

CCAMLR (1980) Convention on the Conservation of Antarctic Marine Living Resources. CCAMLR, Hobart

Conn PB, Kendall WL, Samual MD (2004) A general model for the analysis of mark-resight, mark-recapture, and band recovery data under tag loss. Biometrics 60:900-909

Coulson T, Mace GM, Hudson E, Possingham H (2001) The use and abuse of population viability analysis. Trends Ecol Evol 16:219-221

Croxall JP, Trathan PN, Murphy EJ (2002) Environmental change and Antarctic seabird populations. Science 297:1510-1514 de Kroon H, Plaisier A, van Groenendael J, Caswell H (1996) Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67:1427-1431

127

Ducklow HW, Baker K, Martinson DG, Quetin LB, Ross RM, Smith RC, Stammerjohn SE, Vernet M, Fraser W (2007) Marine pelagic ecosystems: the west Antarctic Peninsula. Philos Trans R Soc Lond B 362:67-94

Dugger KM, Ballard G, Ainley DG, Barton KJ (2006) Effects of flipper bands on foraging behavior and survival of Adélie penguins (Pygoscelis adeliae). The Auk 123:858-869

Duggers KM, Ainley DG, Lyver POB, Barton K, Ballard G (2010) Survival differences and the effect of environmental instability on breeding dispersal in an Adélie penguin metapopulation. Proc Natl Acad Sci USA 107:12375-12380

Emmerson L, Southwell C (2011) Adélie penguin survival: age structure, temporal variability and environmental influences. Oecologia doi:10.1007/s00442-011-2044-7

Emslie SD, Karnovsky N, Trivelpiece W (1995) Avian predation at penguin colonies on King George Island, Antarctica. Wilson Bull 107:317-327

Emslie SD, Coats L, Licht K (2007) A 45,000 yr record of Adélie penguins and climate change in the Ross Sea, Antarctica. Geology 35:61-64

Fraser WR, Hofmann EE (2003) A predator’s perspective on causal links between climate change, physical forcing and ecosystem response. Mar Ecol Prog Ser 265:1-15

Frederiksen M, Daunt F, Harris MP, Wanless S (2008) The demographic impact of extreme events: stochastic weather drives survival and populations dynamics in a long-lived seabird. J Anim Ecol 77:1020-1029

Forcada J, Trathan PN (2009) Penguin responses to climate change in the Southern Ocean. Global Change Biol 15:1618-1630

Forcada J, Trathan PN, Reid K, Murphy EJ, Croxall JP (2006) Contrasting population changes in sympatric penguin species in association with climate warming. Global Change Biol 12:411-423

Gerber LR, Hilborn R (2001) Catastrophic events and recovery from low densities in populations of otariids: implications for risk of extinction. Mammal Rev 31:131-15

Hildenbrandt H, Müller MS, Grimm V (2005) How to detect and visualize extinction thresholds for structured PVA models. Ecological Modelling 191:545-550

Hinke JT, Salwicka K, Watters GM, Trivelpiece S, Trivelpiece WZ (2007) Divergent responses of Pygoscelis penguins reveal a common environmental driver. Oecologia 153:845- 855

Jay CV, Marcot BG, Douglas DC (2011) Projected status of Pacific walrus (Odobenus rosmarus divergens) in the twenty-first century. Polar Biol 34:1065-1084

128

Jenouvrier S, Barbraud C, Weimerskirch H (2005) Long-term contrasted responses to climate of two Antarctic seabird species. Ecology 86:2889-2903

Jenouvrier S, Caswell H, Barbraud C, Holland M, Strœve J, Weimerskirch H (2009) Demographic models and IPCC climate projections predict the decline of an emperor penguin population. Proc Natl Acad Sci USA 106:1844-1847

Laidre KL, Stirling I, Lowry LF, Wiig Ø, Heide-Jørgensen MP, Ferguson SH (2008) Quantifying the sensitivity of arctic mammals to climate-induced habitat change. Ecol App 18(Suppl):S97-S125

Laws RM (1977) Seals and whales of the Southern Ocean. Philos Trans R Soc Lond B 279:81-96

Laws RM (1985) The ecology of the Southern Ocean. Am Sci 73:26-40

Loeb V, Siegel V, Holm-Hansen O, Hewitt R, Fraser W, Trivelpiece W, Trivelpiece S (1997) Effects of sea-ice extent and krill or salp dominance on the Antarctic food web. Nature 387:897-900

Maclean IMD, Wilson RJ (2011) Recent ecological responses to climate change support predictions of high extinction risk. Proc Natl Acad Sci USA doi:10.1073/pnas.1017352108

Mangel M, Tier C (1994) Four facts every conservation biologist should know. Ecology 75:607-614

Meredith MP, King JC (2005) Rapid climate change in the ocean west of the Antarctic Peninsula during the second half of the 20th century. Geophys Res Lett 32, L19604, doi:10.1029/2005GL024042

Nicol S (2006) Krill, currents, and sea ice: Euphausia superba and its changing environment. BioScience 56:111-120

Ovaskainen O, Meerson B (2010) Stochastic models of population extinction. Trends Ecol Evol 25:643-652

129

R Development Core Team (2010) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/

Reid K, Croxall JP, Briggs DR, Murphy EJ (2005) Antarctic ecosystem monitoring: quantifying the response of ecosystem indicators to variability in Antarctic krill. ICES J Mar Sci 62:366-373

Sæther B-E, Bakke Ø (2000) Avian life history variation and the contribution of demographic traits to the population growth rate. Ecology 81:642-653

Scheffer M, Bascompte J, Brock WA, Brovkin V, Carpenter SR, Dakos V, Held H, van Ness EH, Rietkerk M, Sugihara G (2009) Early warning signals for critical transitions. Nature 461:53-59

Schofield O, Ducklow HW, Martinson DG, Meredith MP, Moline MA, Fraser WR (2010) How do polar marine ecosystems respond to rapid climate change? Science 328:1520-1523

Smetacek V, Nicol S (2005) Polar ocean ecosystems in a changing world. Nature 437:362-368

Stammerjohn SE, Martinson DG, Smith RC, Yuan X, Rind D (2008) Trends in Antarctic sea ice retreat and advance and their relation to El Niño – Southern Oscillation and Southern Annular Mode variability. J Geophys Res 113, C03S90, doi:10.1029/2007JC004269

Thomas CD, Cameron A, Green RE, Bakkenes M, Beaumont LJ, Collingham YC, Erasmus BFN, de Siqueira MF, Grainger A, Hannah L, Hughes L, Huntly B, van Jaarsveld AS, Midgley GF, Miles L, Ortega-Huerta MA, Peterson AT, Phillips OL, Williams SE (2004) Extinction risk from climate change. Nature 427:145-148

Thomas ER, Marshall GJ, McConnell R (2008) A doubling of snow accumulation in the western Antarctic Peninsula since 1850. Geophys Res Lett 35: L01706, doi:10.1029/2007GL032529

Trivelpiece WZ, Trivelpiece SG (1990) Courtship period of Adélie, gentoo, and chinstrap penguins. In: Davis JS, Darby JT, (eds), Penguin Biology. Academic Press, Inc, San Diego

Trivelpiece WZ, Hinke JT, Miller AK, Reiss CS, Trivelpiece SG, Watters GM (2011) Variability in krill biomass links harvesting and climate warming to penguin population sin Antarctica. Proc Natl Acad Sci USA 108:7625-7628

130

Turner J, Colwell SR, Marshall GJ, Lachlan-Cope TA, Carleton AM, Jones PD, Lagun V, Reid PA, Iagovkina S (2005a) Antarctic climate change during the last 50 years. Int J Climatol. 25:279-294

Weimerskirch H (2002) Seabird demography and its relationship with the marine environment. In: Schreiber EA, Burger J (eds) Biology of Marine Birds. CRC Press, Boca Raton

White GC, Burnham KP (1999) Program MARK: Survival estimation from populations of marked animals. Bird Study 46(Suppl.):120-139

CHAPTER 4: Rapid climate change and life history: how plastic is the Adélie penguin?

131

132

ABSTRACT

Climate change is having major impacts on physical and biological systems in the

Antarctic. For the Adélie penguin, which breeds from 77.3ºS in the Ross Sea to 56ºS in the

South Sandwich Islands, different populations have been subject to different environmental condition over space and time. Understanding how life history traits from distinct breeding populations have varied across space and time may facilitate an assessment of the ability of the Adélie penguin to respond viably to ongoing climate change. To assess the ability of the

Adélie penguin to cope with recent rapid climate change in the Antarctic Peninsula region, I compared a suite of life history traits, including survivorship, fecundity, age at first breeding, and indices of breeding success, from long-term monitoring studies that were conducted at different times within the last 50 years and scattered across the latitudinal range occupied by

Adélie penguins. Similarities in life history traits associated with fecundity and parental investment in offspring across all populations suggested that the primary response of Adélie penguins to climate change was through a trade-off between survival and age at first breeding.

Specifically, a population from the South Shetland Islands exhibited the lowest adult survival rates and the youngest age of first breeding, consistent with predictions from life history theory that higher mortality rates can be offset by breeding earlier. However, the response exhibited by the Adélie penguins in the South Shetland Islands has been insufficient to maintain positive population growth rates. There, the rapid rate of climate change appears to have exhausted the ability of Adélie penguins to persist in natal habitats.

133

INTRODUCTION

The response of seabirds to variations in climate is mediated by the flexibility of their specific life history strategies. Studies of seabird responses to climate change have highlighted the sensitivity of survival rates (Sandvik et al. 2005; Frederiksen et al 2008; Jenouvrier et al.

2009), reproductive success (Sandvik and Erikstad 2008; Lewis et al. 2009), and the phenology of important events like reproduction and migration (Barbraud and Weimerskirch

2006; Gaston et al. 2009; Reed et al. 2009) to variation in air and sea temperatures, large scale atmospheric pressure anomalies, sea ice conditions, and extreme weather events. Other studies suggest that climate change may cause shifts in the distribution of breeding and foraging ranges of seabird (Wynn et al. 2007; Crawford et al. 2008) owing to changes in marine productivity. While such studies clearly demonstrate that seabirds do respond to climate change in significant ways, fewer studies consider whether plasticity in a broad suite of life history traits can interact to maintain population abundance and distributions (Forcada et al.

2008). Here, I compare demographic data from different populations of Adélie penguins separated by space and time to examine how life history traits have differed under various environmental conditions and then make inferences about the ability of the Adélie penguin to cope with climate change in the Antarctic Peninsula region, the most rapidly warming area in

Antarctica (Turner et al. 2005).

The extent of variation in life history traits within populations constrains the potential responses to environmental perturbations and ultimately determines whether a species can cope with rapid climate change. Life history strategies, defined as the suite of traits that interact to maximize the fitness of an organism, are evolutionary trade-offs in energy allocation (Stearns 1992; Bernardo 1994); investment favoring one trait generally incurs a

134

debt paid by another trait. Over time, natural selection acts to eliminate the allocation strategies that perform poorly in a given environment. Such local adaptation can be advantageous, in terms of fitness, for resident genotypes, but also deleterious if the environmental conditions supporting a specific life history trade-off change. Climate change represents one such risk for locally adapted life histories. In particular, in the Antarctic

Peninsula region, the current rate of climate change and expectations for further directional trends (Solomon et al. 2007; Bracegirdle et al. 2008) is expected to push systems beyond their range of historic variability. It is unclear whether locally adapted life histories strategies can respond quickly enough to keep pace (Visser 2008).

Viable responses to environmental variation are linked to the plasticity of life-history traits. Given the primary trade-off in life history theory between reproduction and survival, measuring the variation of those life history traits from distinct populations may allow an assessment of how populations can respond, or have responded, to climate change. Among seabirds, the extent of variation in fecundity and adult survival rates is typically small. Most seabirds lay only 1 or 2 eggs per breeding attempt, and very few, if any, individuals within the species deviate from that pattern (Bennett and Owens 2002). Consequently, adult survival rates are relatively high to compensate for low fecundity. However, population growth rates are sensitive to minor changes in survival rates of long-lived species (Woller et al. 1997;

Sæther and Bakke 2000; Weimerskirch 2002). Increased variability in environmental conditions that are associated with climate change can induce strong negative population growth rates in seabirds (e.g. Frederiksen et al. 2008) by simply increasing the variance in annual survival rates (Lewontin and Cohen 1969). In the long run, a life history that is dependent on stable adult survival and a relatively fixed, but low, fecundity may become maladaptive if the local environment shifts to a regime that exhibits frequent and large-

135

magnitude departures from the range of historical environmental variability. Thus, a conundrum: how does a long-lived seabird with low fecundity cope with rapid climate change that tends to reduce adult survival (Bernardo and Spotila 2006)?

In general, viable responses can be either migratory or adaptive (Visser 2008; Forcada and Trathan 2009). Here, I focus on potentially plastic responses that may arise from the life history tradeoff between survival and reproduction that could lead to adaptive responses; such plastic responses are the only means by which a species can truly persist under rapid global climate change. Predictions from life history theory provide a suite of possible responses to the seabird scenario outlined above and are derived from the core life history relationships represented schematically in Fig. 4-1. Specifically, if adult survival declines and changes in fecundity cannot compensate to maintain positive population growth rates, selection could favor strategies that reduce the costs of reproduction on future survival or increase the reproductive potential of those animals remaining. For example, the population might shift toward breeding at a younger age, thus securing breeding opportunities for individuals while circumventing the risk of early death. Alternatively, the population might shift towards strategies that minimize the cost to breeders of producing a chick, either by skipping breeding more frequently, reducing clutch size, or limiting the amount of parental care provided prior to independence of the chick. To examine these predictions, I review data from multiple-year demographic studies on Adélie penguin (Pygoscelis adeliae), monitored at different times and in different places throughout their Antarctic range. At the time each study was conducted, each breeding population was subject to different prevailing climatic conditions, providing the necessary contrast to examine how demographic traits may vary within a species. The comparisons serve to expand the scope of the long-term data collected on individual Adélie

136

penguin populations from around the continent to better inform inference on how penguins cope with ongoing climate change, particularly in the northern Antarctic Peninsula region.

The Adélie penguin

The general life history of the Adélie penguin is well characterized (Williams 1995;

Ainley 2002) and exhibits characteristics typical of seabirds. Briefly, the Adélie penguin inhabits of the marginal pack ice zone during the austral winter but returns to land during the austral summer to breed on ice-free beaches. Breeding colonies are widely distributed around

Antarctica and its offshore islands, ranging from 77.3ºS in the Ross Sea to 56ºS in the South

Sandwich Islands (Fig. 4-2). An Adélie penguin typically lives 10-16 yrs and the maximum age confirmed in the wild was 20 yrs (Ainley 2002). Sexual maturity is reached at age 3, but birds gradually recruit into the breeding population between the ages of 3 and 7 yrs (Ainley

2002). Females have a maximum clutch size of 2 eggs, but first-time breeders frequently only lay 1 egg (Ainley 2002). Survival rates during the first and second years of life are generally lower than for adults, and survival rates of all ages can be sensitive to large-scale environmental variability (Wilson et al. 2001; Jenouvrier et al. 2006; Hinke et al. 2007). In particular, variability in the physical environments and biological conditions in the marine habitats occupied by Adélie penguins (Loeb et al. 1997; Atkinson et al. 2004; Stammerjohn et al. 2008) has been correlated with Adélie population trends throughout its range (e.g., Fraser et al. 1992; Wilson et al. 2001; Jenouvrier 2006; Forcada et al. 2006; Hinke et al. 2007;

Emmerson and Southwell 2008). Currently, populations in the south Atlantic sector of the

Adélie penguin distribution, from the South Sandwich Islands to the western Antarctic

Peninsula region are declining from peak population sizes reached in the 1970s (Forcada et al.

137

2006; Lynch et al. 2008; Schofield et al. 2010; Trivelpiece et al. 2011), while populations in the south Pacific and Indian Ocean sectors have exhibited stable to increasing population during that time (Taylor and Wilson 1990; Wilson et al. 2001; Clarke et al. 2003; Jenouvrier et al. 2006).

Climate differences at the main study sites

Study sites with long-term demographic data on Adélie penguins range from Cape

Crozier in the southwestern Ross Sea to the South Shetland Islands off the northern tip of the

Antarctic Peninsula, spanning nearly the full latitudinal range breeding habitats (Fig. 4-2). On average, differences in temperature, precipitation, and winter sea-ice extent are evident in the main areas occupied by Adélie penguins over this broad geographical range. Relative to the

Ross Sea and East Antarctica, the northern Antarctic Peninsula region is warmer and receives more frequent and larger amounts of precipitation. The duration of winter sea-ice coverage is also shortest in the Antarctic Peninsula region, where the ice season has decreased by nearly two months since the mid 1970s (Stammerjohn et al. 2008). Since the middle of the 20th century, breeding populations of Adélie penguins in the northern Antarctic Peninsula region have also experienced rapid atmospheric warming while the populations in east Antarctica and the Ross Sea region have experienced colder and more stable temperatures (Turner et al.

2005). For Adélie penguins, these differences are important. Namely, temperature and precipitation conditions dictate the availability of snow-free breeding habitat (Patterson et al.

2003; Boersma 2008), and the extent and duration of winter sea-ice coverage determines the availability of suitable foraging habitat during winter and can impede movement to and from the colony during the breeding season (Emmerson and Southwell 2008; Ballard et al. 2010).

138

Local differences in environmental conditions due to geographical location are also influenced by large-scale atmospheric conditions. In particular, the Southern Annular Mode

(SAM), a measure of air pressure differences over Antarctica relative to lower latitudes, is a dominant driver of zonal difference in Antarctic climate (Thompson and Solomon 2002; Hall and Visbeck 2002). For penguins, the effect of the SAM is perhaps most important through its effects on winds (hence storm tracks, air temperatures, and precipitation), winter sea ice distributions, and sea surface temperatures (Hall and Visbeck 2002; Lefebvre et al. 2004).

Over the duration of the penguins studies considered here, the SAM exhibited a predominantly negative phase in the 1960s and a predominantly positive phase during the 1990s and 2000s

(Marshall 2003). A positive phase of the SAM, characterized by low pressure over the continent and higher pressure at lower latitudes, results in stronger southerly winds and is generally thought to favor cooling over most of continent (Thompson and Solomon 2002).

Due to the northward projection of the Antarctic Peninsula, however, increased westerly winds favor advection of warmer oceanic air into the region, resulting in substantial warming.

Thus, each of the sites of long-term studies of Adélie penguins considered here was studied under contrasting climatic conditions. In particular, a negative phase SAM and dry continental conditions prevailed in the Ross Sea during the late 1960s and early 1970s. In the 1990s, East

Antarctica was characterized by dry continental conditions and a positive phase of the SAM.

The northern Antarctic Peninsula region experienced a positive phase SAM with warmer, more maritime conditions during the late 1990s and early 2000s.

139

MATERIALS AND METHODS

Available data

The data sources used here (Table 4-1) are representative of multiple year demographic studies conducted across the latitudinal range of Adélie penguins over the last 50 yrs, from Cape Crozier (77º 3’S, 169º 23’E) at the southern-most edge of the Adélie breeding range, to Béchervaise Island (67º 35’S, 62º 49’E) in East Antarctica, to the King George

Island (62º10’S, 58º27’W) near the northern limit of currently occupied habitats (Fig. 4-2).

The oldest data used here derives from a study conducted during the 1960s and early 1970s at

Cape Crozier, in the Ross Sea. The Cape Crozier data were originally reported by Ainley and

Schlatter (1974) and Ainely and DeMaster (1980) and later revised and compiled by Ainley

(2002); the compilation from the earlier reports was used as the source of data for comparisons presented here. The data for Béchervaise Island were reported by Clarke et al. (2002) and

Clarke et al. (2003) and represent 12 years of demographic studies initiated in 1990. The demographic data for Admiralty Bay from 1997 to 2008 are presented for the first time, but portions of some of the reproductive indices (see below) have been reported previously (Hinke et al. 2007). Published data from additional topical studies at Edmonson Point (74º21’S,

165º10’E; Pezzo et al. 2007), in the western Ross Sea, and from a collection of small Adélie penguin colonies near the U.S. Long Term Ecological Research site at Palmer Station, hereafter referred to as Anvers Island (64º46’S, 64º04’W; Salihoglu et al. 2001), along the western Antarctic Peninsula, are included to strengthen comparisons with additional data.

Note that data on each life history trait are not available for all sites (Table 4-1).

140

From the published studies, I compiled data on age-specific survivorship, age-specific fecundity, age at first breeding, mean clutch size (number of eggs laid per nest), and mean reproductive success (number of chicks raised to the crèche stage per nest). When possible, I also compared trends in age at first breeding, reproductive success, and chick weights across sites. Life table estimates of net reproductive rates, generation times, and intrinsic population growth rates at each of the main study sites were also compared. Unless otherwise noted, comparisons between studies are made based on reported annual averages.

In most cases, similar methods were used at each study site to estimate the parameters outlined above. A notable exception, however, arises from different marking techniques for identifying individuals. For example, metal flipper bands were used at Cape Crozier and

Admiralty Bay while implanted passive radio transponders were used at Béchervaise Island.

The effects of such different identifying techniques on reproductive success and survival are likely to differ (Dugger et al. 2006). Therefore, despite a long history of study, comparisons across sites are necessarily rough in nature and should be interpreted with caution.

Life table analysis

A life table for Adélie penguins at Admiralty Bay has not been published previously, so the methods for data collection and analysis of the data from known-age penguins are provided here. The data on known-age individuals derive from annual studies of penguins banded as chicks (age 0) and monitored in all subsequent field seasons. I used data from the 1997/98 breeding season through the 2007/08 season, a period when the breeding population declined from 5612 to 2142 pairs. In these years, 500 chicks were marked with uniquely numbered stainless steel flipper bands. In the final year, only 250 chicks were banded due overall

141

population declines. The banded birds were resighted during all subsequent field seasons

(October through February) and annual presence/absence of each band in the study was compiled.

Reproductive success of all known-age breeders, estimated as the number of chicks crèched per nest, was monitored each breeding season. A successful crèche was determined at the first observation of the chicks being left unattended in the nest by either parent. Note that for the purpose of the life table, reproductive success is measured as the number of female fledglings per nest with a known-age female. Because the sexes of chicks were unknown at the time of crèche, a 1:1 sex ratio among chicks was assumed.

The fecundity schedule (bx) was determined from the data on age-specific reproductive success and the proportion of adult females from each age-class that attempted breeding. At Admiralty Bay, female birds between the ages of 3 and 8 yrs were recorded breeding, but only 13 females aged 6+ yrs were ever observed breeding. Thus, I averaged the data on female reproductive success for all ages ≥ 6 yrs. Estimates of the proportion of females breeding in each age-class were calculated as the proportion of the number of breeders observed in each age-class relative to number of birds in each age-class known to be alive based on the survivorship schedule.

To calculate survivorship, the proportion of each cohort still alive each year after hatching, for the life table analysis, I first used program MARK (White and Burnham 1999) to estimate survival probabilities from the annual presence/absence data. A global model with time- and age-varying survival rates and recapture rates was first fit to the data to assess goodness of fit. A bootstrapping algorithm, implemented in MARK, identified over-dispersion in the data ( cˆ = 1.62) and subsequent model selection therefore proceeded based on corrected likelihoods and quasi-AIC criteria (Lebreton et al. 1992). Given the goal of comparing vital

142

rates of Adélie penguins at Admiralty Bay with vital rates from previously published studies, the model set was restricted to age- and stage-structured models that included eight age- classes or up to three stages. The stage-based models included models with three stages (age- class3) with a juvenile (survival from age 0 to 2), a pre-breeder stage (age 2-3) and an adult stage. An alternative two stage model (age-class2) considered only a juvenile stage (survival from age 0 to 2) and an adult stage (survival from age 2+). Parameterization of the recapture models was either a full age-based model or restricted to a 2 stage model for recapture for juveniles (age1 and age 2) and adults (age3+). The stage-based classifications represent useful simplifications of the age-structure to describe the periods between initial banding and the typical age of first return to a natal colony, survival of immature animals prior to sexual maturity, and survival of sexually mature adults. Each of the age- and stage-based models was fit to the data and model averaging, based on AIC weights, was used to estimate survival rates from the best fitting models.

A main assumption in mark-recapture studies is that marks are not lost over time.

However, band loss and band-induced mortality can cause mark-recapture estimates of survival to be biased low (Nelson et al. 1980). Preliminary analysis confirmed this was the case in the data. Specifically, the estimated survival adult rates predicted a population growth rate of r = -0.48, lower than the observed decline in the population of r = -0.10. To correct the mark-recapture estimates of adult survival, a Leslie matrix model was constructed and fit to census data by estimating a band retention rate that was necessary to correct survival rates to account for the observed population trend. The retention rate can be considered a global correction for potential band loss and band-induced mortality. Corrected stage-specific survival rates, with stages indexed by a, were estimated as

143

∧ Sa S a = . Eq. 1 δ

The corrected survival rates were constrained to be less than 1 with a logit transformation.

During the fitting process, updated survival rates were also applied to the fertility coefficients in the Leslie matrix as

∧ ∧ F x = Fx * S a Eq. 2

where Fx is the age-specific fecundity reported in Table 4-2. The survivorship schedule (li) for each age-class (i) was then calculated from the corrected survival probabilities as

li−1 li = ∧ S a−1 Eq. 3

given that l0=1. The estimation was made with sum of squares minimization between actual and estimated population size from 1997 through 2007.

RESULTS

The distribution of age at first breeding was shifted toward younger birds at Admiralty

Bay relative to Cape Crozier (Fig. 4-3a). The mean age at first breeding at Cape Crozier was estimated to be 6.2 ± 0.05 yrs for males and 5.0 ± 0.1 yrs for females (Ainley 2002). The mean age of first time breeders at Admiralty Bay is 4.3 ± 0.14 yrs for males and 3.9 ± 0.14 yrs for

144

females. The average difference (± 95% confidence interval) in ages at first breeding between sites was 1.9 ± 0.3 years for males and 1.4 ± 0.4 years for females. To check whether the younger age distribution of first-time breeders is a result of on-going decline in the Admiralty

Bay population, the full data set (1982 to current) was examined to look for a trend in the age at first breeding. The shift to younger breeders appears to have occurred in Admiralty Bay before monitoring began in the mid 1980s and has not trended since (Fig. 4-3b).

Average egg production was near the maximum of 2eggs per female across all sites

(Fig. 4-4a). The number of chicks surviving to the crèche stage per nest was more variable across sites (Fig. 4-4a), with marginally higher chick production in the northern Antarctic

Peninsula region than at other locations. Over time, the number of chicks crèched per nest at

Admiralty Bay, Anvers Island, and Béchervaise Island exhibited trends (Admiralty Bay: F1,13

= 2.19, P = 0.16; Anvers Island: F1,14 = 3.2, P = 0.10; Béchervaise Island: F1,11 = 0.02, P =

0.88; Fig. 4-4b).

Body mass of chicks at the beginning of the crèche period and immediately prior to fledging generally suggested that parents at all sites produced chicks of similar weight. Chick body mass at the time of crèche, which generally occurs between the ages of 18 and 27 days

(Ainley and Schlatter 1972) were similar at both Cape Crozier and Admiralty Bay and averaged 1.6 ± 0.2 and 1.7 ± 0.01 kg, respectively (Fig. 4-5a). Just prior to departure from their natal colony, chicks weights averaged 2.8 to 3.1 kg at all study sites (Fig. 4-5b).

Life table analysis

Mark-recapture estimates of survival probabilities at Admiralty Bay indicated that an additive models of year and 3 stages (juveniles, pre-breeders, and adults) or 2 age-classes

145

provided equivalent fits to the data, together accounting for 70% of the model weight (Table

4-2). Model results were therefore averaged based on the AIC weights, since no single model best explained the variation in survival. Because of the time effect in the model-averaged results, a geometric average of the stage-specific survival probabilities was used in the life table analysis.

As birds aged, the percentage of females engaged in breeding at Admiralty Bay increased from 15% at age 3 To 86% for age 6+. Mirroring the age-specific increase in breeding propensity was an increase in reproductive success from 0.19 females creched·nest-1 for age 3 breeders up to 0.52 female creched·nest-1 for older birds (Table 4-3).

The model fitting procedure suggested that retention rates, δ, of 0.635 provided the best fit to the data (Fig. 4-6). Using the final estimates of survival rates for juveniles (0.44), pre-breeders (0.89) and adults (0.84), the survivorship schedules for Admiralty Bay (Table 4-

3) suggested that different mortality patterns operate within the different populations of Adélie penguins around the continent (Fig. 4-7). The breeding population at Admiralty Bay had the lowest survival rates for all stages, resulting in the most rapid decline in survivorship across study populations. The highest juvenile survival rates occurred at Béchervaise Island and contributed to the highest survivorship across all ages, despite the occurrence of the highest adult survival rates occurred at Cape Crozier (Fig. 4-7).

The life tables reported for breeding populations at Cape Crozier and Béchervaise

Island (Clarke et al. 2003) indicated that the Cape Crozier population had a generation time of

10.2 yrs, a net reproductive rate of 0.62, and a negative population growth rate (r = -0.046) in the late 1960s (Table 4-4). Penguins at Béchervaise Island had a generation time of 9.5 yrs, net reproductive rate of 1.03. The Béchervaise Island population was the only population in the study that exhibited a positive population growth rate (r = 0.003). Estimates for these

146

demographic parameters were all lowest at Admiralty Bay (Table 4-4), with a negative population growth rate (r = -0.129), shortest generation time (G = 8.85) and a low net reproductive rate (Ro = 0.29).

DISCUSSION

This review of data from Adélie penguin populations collected at different times and from different locations around the Antarctic continent illustrates that the main differences between Adélie penguin populations were related to local survival rates of adults and juveniles and the age at first breeding. Indices of egg production, reproductive success, and the mean weights of chick at crèche and fledgling were similar across space and time, indicating that life history traits related to paternal investment in offspring among Adélie penguins are relatively fixed within the species. The trade-off between adult survival rates and the distribution of breeding ages, however, is consistent with life history theory which predicts the positive correlation between survival rates and first breeding ages (Bennett and Owens 2002).

Moreover, the differences in breeding ages and survival across sites suggest that plastic responses in breeding age to local conditions are evident. In a general sense, this suggests that the demographic characteristics of a ‘typical’ penguin population are site dependent. More specifically, the apparent inflexibility of life history traits that could lead to increased productivity, but higher variability in adult survival rates across populations highlights the vulnerability of Adélie penguins to rapid climate change.

The shift in first-breeding ages can be viewed as a plastic, risk-prone decision to offset lower expected survival rates and the consequent reduction in the expected number of breeding opportunities. Among Adélie penguins, first-time breeders are generally

147

unsuccessful, but those that are successful tend to have longer reproductive life spans (Ainley

2002). Among Adélie penguins, breeding at any age may be a strong filter that retains only individuals of higher quality for future breeding (Lescroël et al. 2009). Thus, by breeding early when generation times are short, females that are successful may increase the chances of having relatively longer reproductive life spans. It is interesting to note that the shift to younger breeders at Admiralty Bay occurred prior to the period of monitoring in this population. Although speculative, the occupation of the northern habitats in the Antarctic

Peninsula and South Atlantic sector region, a habitat intrinsically warmer than the continental regions further south which typify the main population centers of Adélie penguins, may have implied generally lower survival rates and required younger birds to breed for the populations to persist in. If so, the occupation of the northern breeding ranges may have exhausted the phenotypic plasticity of the Adélie penguin, leaving them highly vulnerable to recent climate warming.

Without compensatory shifts in fecundity, earlier breeding appears to be the primary option available to offset increased mortality rates. However, the effectiveness of the shift to younger first-time breeding at buffering populations from variable survival rates may be insufficient to maintain stable population growth rates. Caswell and Hastings (1980) demonstrated that shifts in either age at first breeding or changes in fecundity can have the same effect on population growth rates. Importantly, the necessary magnitude of such responses depends on the population growth rate and the average adult survival rate and were related according to

148

q  λ  kq =    S 

where λ is the population growth rate, S is the survival rate, q identifies the unit shift in development time, and kq represents the factor by which fecundity must increase to match the effect of changing development time by q-units (Equation 14 in Caswell and Hasting 1980).

When the ratio of population growth rate to adult survival rate is high, Caswell and Hastings

(1980) noted that there should be strong pressure to shift breeding ages. At Admiralty Bay, the ratio is 0.88/0.84 = 1.05. This implies that a 1 to 2 unit shift in breeding age would require a 5 to 10% increase in fecundity to achieve the same effect on the population growth rate.

Although seemingly small, the fixed clutch size of seabirds appears to circumvent that prospect of such increases in net fecundity, and thus allow (or necessitate) breeding age to drift in order to maintain viable populations. For the data from Admiralty Bay, all things equal in the life table, increasing fecundity up to 33% would only result in a 3% increase (0.88 to

0.9) in population growth rate. That small increase in growth rate given a rather large change in fecundity clearly does not compensate for the markedly lower survival rates, further suggesting the limited ability of this declining population of Adélie penguins to cope with recent climate change.

One characteristic of potential importance for future monitoring of Adélie penguins arises from the similar body mass attained at the time of crèche and fledging at each of the study sites. Similarities across space and time, irrespective of local environmental conditions, suggest that a critical body mass appears to be conserved and likely reflects optimal conditions for chick survival. Life history theory suggests that one method to counter changes in survival is to reduce parental investment in offspring. If provisioning chicks incurs a cost on future survival of the parent, then producing large chicks relative to the average may result in

149

increased mortality rates for parents and subsequent selection against that strategy. Parents that produce chicks with below-average weights may also suffer because of higher mortality rates of chicks and a subsequent decreased fitness of the parental strategy. A gradual elimination of both parental strategies from a population under conditions of environmental duress would be expected to result in maintenance of the critical chick weight while reducing variation in chick weights. There was no evidence for such a decline in the variance of chick weights at our study site (data not shown). Nonetheless, monitoring for declines in the mean or variance of chick masses could reflect deterioration of the environment and its ability to support Adélie penguins.

For any organism, viable responses to environmental change can involve migration away from unsuitable areas or in situ adaptations that allow persistence in established areas.

The lack of evidence for changes in the distributions of most life history parameters of sufficient magnitude to maintain stable population growth rates in the northern population suggest that migratory movement to more southern areas is the only viable option that Adélie penguins in the Antarctic Peninsula may have right now. There are very few data available to document such large-scale movements; search effort for banded birds beyond the study colonies is essentially zero. Of all the birds banded at Admiralty Bay since the late 1970s, only one has ever been reported from somewhere other than King George Island; a lone 3 yr old bird was observed on Paulet Island (63.5°S, 55.75°W, some 200 km SE of the natal colony) during the 2008/09 austral summer (H. Lynch, pers comm). Dispersal distances of Adélie penguins in the Ross Sea during a period of population decline are similar in scale (Ainley and

DeMaster 1980) and estimates of annual emigration rates under poor environmental approach

3.5% at small colonies that are similar in size to the study colonies in Admiralty Bay (Dugger et al. 2009). Thus, emigration should not be assumed to be zero in the Antarctic Peninsula

150

region. Throughout the northern extent of the Adélie range, populations are declining rapidly

(Forcada et al. 2006; Hinke et al. 2007; Lynch et al. 2008). Estimating the contribution of emigration to those local declines is a major research challenge and will likely require improved monitoring techniques and increased search effort at potential refuge colonies.

Conclusion

Given the current rates of atmospheric warming, loss of sea ice, and consequent changes in the Antarctic Peninsula food web (Atkinson et al. 2004; Loeb et al. 1997; Ducklow et al. 2007), the Adélie life history strategy may no longer support viable populations in this region. It is likely that the range of environmental conditions expected to occur in the near future, including increased temperatures, less sea ice, and more frequent precipitation

(Solomon et al. 2007; Bracegirdle et al. 2008) in the Antarctic Peninsula region will move conditions beyond a window to which the Adélie penguin is adapted. Despite an apparent plastic response to breed earlier when adult survival rates are low, the magnitude of observed responses by Adélie penguins to ambient environmental conditions appear insufficient to reverse the significant population declines in the rapidly changing Antarctic Peninsula environment.

ACKNOWLEDGEMENTS

Many thanks to W Trivelpiece and S Trivelpiece and numerous field assistants for their long-term dedication to monitoring Adélie penguin populations on King George Island.

Comments by L. Emmerson improved the manuscript. Data from the Palmer Station Long

151

Term Ecological Research site were made available online by W. Fraser with support by the

Office of Polar Programs, NSF Grants OPP-9011927, OPP-9632763 and OPP-0217282. The

U.S. Antarctic Ecosystem Research Division of the National Oceanic and Atmospheric

Administration and grants from the National Science Foundation to W. Trivelpiece provided logistical and financial support for this work. Additional support was generously provided by the Lenfest Oceans Program at the Pew Charitable Trusts.

Table 4-1: Source of published data and the parameters available from the original studies of Adélie penguin populations around Antarctica.

Age at Study first Clutch Reproductive Chick Fledge Colony Name Location period Survival breeding size success weights weights Source 77º 31’S, Cape Crozier 1961-1976 X X X X X X Ainley 2002 169° 23’ E Edmonson 74° 21’ S, 1995-2005 X X Pezzo et al. 2007 Point 165° 10’ E a Béchervaise 67° 35’ S, a b Clarke et al. 2003 , 1990-2001 X X X Island 62° 49’ E Clarke et al. 2002b c 64° 46’ S, c d LTER Data Zoo , Anvers Island 1988-2006 X X 64° 04’ W Salihoglu et al. 2001d Admiralty 62° 10’ S, 1997-2008 X X X X X X This paper Bay 58° 27’ W

152

153

No. Rank Survival model Recapture model parameters ΔQAICc weight 1 age-class3+time age-class3*time 25 0 0.37 2 age-class2+time age-class*time 24 0.198 0.33 3 age-class1*time age-class*time 30 0.478 0.29 4 age-class+time age-class*time 30 7.316 0.01 5 age-class*time age-class*time 36 10.076 0.002

154

Table 4-3: Life table analysis of Adélie penguins at Admiralty Bay. The survivorship schedule, denoted l(x), continues a monotonic decrease between ages of 6 and 17; all other values constant between ages 6 and 17. We limit the dimensions of the life table at the oldest observed Adélie penguin age, which was 17, at Admiralty Bay (WZT, unpublished data). The fecundity schedule is denoted b(x). Survival probabilities are denoted g(x) and are corrected with an estimated retention rate of 0.68.

Prop. Mean no. g(x) ˆ xg )( Fertility Age breeding offspring b(x) raw corrected l(x) coef. 0 0.00 0.00 0.00 0.44 0.44 1.00 0.00 1 0.00 0.00 0.00 0.44 0.44 0.44 0.00 2 0.00 0.00 0.00 0.57 0.89 0.19 0.00 3 0.15 0.19 0.03 0.54 0.84 0.17 0.02 4 0.29 0.37 0.11 0.54 0.84 0.14 0.09 5 0.44 0.36 0.16 0.54 0.84 0.12 0.13 6 - 17 0.86 0.52 0.45 0.54 0.84 0.10 … 0.37

155

Table 4-4: Population properties derived from life-table analyses. Copacabana results are based on 17 age-classes, all other sites based on 19 age-classes. Ro is the net reproductive rate, G is the generation time, and r is the intrinsic rate of population growth.

Admiralty Cape Béchervaise Parameter Bay Crozier Island Juvenile survival 0.44 0.51 0.69 Pre-breeder survival 0.89 NA NA Adult survival 0.84 0.89 0.86 Ro 0.290 0.62 1.03 G 8.850 10.2 9.5 r -0.129 -0.046 0.003

156

Fig. 4-1: Core life history relationships and their correlations, denoted with a positive or negative signs, with adult survival rate.

157

Fig. 4-2: Map of Antarctica with location of the three main study sites at Admiralty Bay, Cape Crozier, and Béchervaise Island. Other data were taken from Anvers Island and Edmonson Pt. Colony location and abundance information is from Ainley (2002). The base map was modified from a version published by Australian Antarctic Data Centre, March 2009 and is available for download at http://data.aad.gov.au/aadc/mapcat/display_map.cfm?map_id=13580 (accessed 6/10/2009).

158

Fig. 4-3: Comparison of age distributions for first-time male and female breeders at Cape Crozier and Admiralty Bay across space (A) and across time at Admiralty Bay (B).

159

Fig. 4-4: Average reproductive success of Adélie penguins across space and time at selected breeding colonies. A) Mean breeding indices (mean and standard error) for Cape Crozier, Edmonson Pt, Béchervaise Island, Admiralty Bay, and Palmer Station. B) Annual mean numbers of chicks crèched per nest at Béchervaise (Δ), Admiralty Bay (●), and Palmer Station (○). Trend lines are plotted for reference; none exhibit statistically significant trends. Sites with missing data are denoted NA.

160

Fig. 4-5: Average mass of chicks at crèche and fledging across space and time at selected breeding colonies. A) Mean body mass of Adélie penguin chicks from 2 chick broods at the beginning of the crèche period (chicks aged 18-27 days, after Ainley and Schlatter 1972). B) Annual mean body mass of Adélie penguin chicks immediately prior to fledging. Error bars represent standard deviations. Sites with missing data are denoted NA.

161

Fig. 4-6: Adélie penguin census data (•) at Admiralty Bay and best fitting Leslie matrix model (solid line), with a retention rate of 0.635 applied to pre-breeder (survival from age 2 to 3) and adult stage (age 3+) survival rates. Dashed lines are model predictions based on retention rates of 0.64 and 0.63.

162

Fig. 4-7: Survivorship of Adélie penguins from Béchervaise Island (dashed line), Cape Crozier (thin line) and Admiralty Bay (thick line).

163

REFERENCES

Ainley DG (2002) The Adélie penguin: bellwether of climate change. Columbia University Press, New York

Ainley DG, DeMaster DP (1980) Survival and mortality in a population of Adélie penguins. Ecology 61:522-530

Ainley DG, Schlatter RP (1972) Chick raising ability in Adélie penguins. Auk 89:559-566

Atkinson A, Siegel V, Pakhomov E, Rothery P. (2004) Long-term decline in krill stock and increase in salps within the Southern Ocean. Nature 432:100-103

Ballard G, Toniolo V, Ainley DG, Parkinson CL, Arrigo KR, Trathan PN (2010) Responding to climate change: Adélie penguins confront astronomical and ocean boundaries. Ecology 91:2056-2069

Barbraud C, Weimerskirch H (2006) Antarctic birds breed later in response to climate change. Proc Natl Acad Sci USA 103:6248-6251

Bennett PM, Owens IPF (2002) Evolutionary ecology of seabirds. Life histories, mating systems, and extinction. Oxford University Press, Oxford

Bernardo J (1994) Experimental analysis of allocation in two divergent natural salamander populations. Am Nat 143:14-38

Bernardo J, Spotila JR (2006) Physiological constraints on organismal response to global warming: mechanistic insights from clinally varying populations and implications for assessing endangerment. Biol Lett 2:135-139

Bracegirdle TJ, Connolley WM, Turner J (2008) Antarctic climate change of the twenty first century. J Geophys Res 113:D0310, doi:10.1029/2007JD008933

Boersma PD (2008) Penguins as marine sentinels. BioScience 58:597-607

Caswell H, Hastings A. (1980) Fecundity, developmental time, and population growth rate: an analytical solution. Theor Popul Biol 17:71-79

Clarke J, Emmerson LM, Townsend A, Kerry KR (2003) Demographic characteristics of the Adélie penguin population on Béchervaise Island after 12 years of study. CCAMLR Sci 10:53-74

Clarke J, Kerry K, Irvine L, Phillips B (2002) Chick provisioning and breeding success of Adélie penguins at Béchervaise Island over eight successive seasons. Polar Biol 25:21-30

Crawford RJM, Tree AJ, Whittington PA, Visagie J, Upfold L, Roxburg KJ, Martin AP, Dyer BM (2008) Recent distributional changes of seabirds in South Africa: is climate having an impact? Afr J Mar Sci. 30:189-193

164

Ducklow HW, Baker K, Martinson DG, Quetin LB, Ross RM, Smith RC, Stammerjohn SE, Vernet M, Fraser W (2007) Marine pelagic ecosystems: the west Antarctic Peninsula. Philos Trans R Soc B 362:67-94

Dugger KM, Ainley DG, Lyver POB, Barton K, Ballard G (2010) Survival differences and the effect of environmental instability on breeding dispersal in the Adélie penguin metapopulation. Proc Natl Acad Sci USA 107:12375-12380

Dugger KM, Ballard G, Ainley DG, Barton KJ (2006) Effects of flipper bands on foraging behavior and survival of Adélie penguins (Pygoscelis adeliae). Auk 123:858-869

Emmerson L, Southwell C (2008) Sea ice cover and its influence on Adélie penguin reproductive performance. Ecology 89:2096-2102

Forcada J, Trathan PN (2009) Penguin responses to climate change in the Southern Ocean. Global Change Biol. 15:1618-1630

Forcada J, Trathan PN, Murphy EJ (2008) Life history buffering in Antarctic mammals and birds against changing patterns of climate and environmental variation. Global Change Biol 14:2473-2488

Forcada J, Trathan PN, Reid K, Murphy EJ, Croxall JP (2006) Contrasting population changes in sympatric penguin species in association with climate warming. Global Change Biol 12:411-423

Gaston AJ, Gilchrist HG, Mallory ML, Smith PA (2009) Changes in seasonal events, peak food availability, and consequent breeding adjustment in a marine bird: a case of progressive mismatching. Condor 111:111-119

Hall A, Visbeck M (2002) Synchronous variability in the Southern Hemisphere atmosphere, seas ice and ocean resulting from the annular mode. J Climate 14:3043-3057

Hinke JT, Salwicka K, Watters GM, Trivelpiece S, Trivelpiece WZ (2007) Divergent responses of Pygoscelis penguins reveal a common environmental driver. Oecologia 153:845-855

Jenouvrier S, Barbraud C, Weimerskirch H (2006) Sea ice affects the population dynamics of Adélie penguins in Terre Adélie. Polar Biol 29:413-423

165

Lebreton J-D, Burnham KP, Clobert J, Anderson DR (1992) Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecol Monogr 62:67-118

Lefebvre W, Goosse H, Timmermann R, Fichefet T (2004) Influence of the Southern Annular Mode on the sea ice-ocean system. J Geophys Res 109:C09005. Doi:10.1029/2004JC002403

Lescroël A, Dugger KM, Ballard G, Ainley DG (2009) Effects of individual quality, reproductive success and environmental variability on survival of a long-lived seabird. J Anim Ecol 78:798-806

Lewis S, Elston DA, Daunt F, Cheney B, Thompson PM (2009) Effects of extrinsic and intrinsic factors on breeding success in a long living seabird. Oikos 118:521-528

Lewontin RC, Cohen (1969) On population growth rate in a randomly varying environment. Proc Natl Acad Sci USA 62:1056-1060

Loeb V, Siegel V, Holm-Hanson O, Hewitt R, Fraser W, Trivelpiece W, Trivelpiece S (1997) Effects of sea-ice extent and krill or salp dominance on the Antarctic food web. Nature 387:897-900

Lynch HJ, Naveen R, Fagan R (2008) Census of penguin, blue-eyed shag Phalacrocorax atriceps and southern giant petrel Macronectes giganteus populations of the Antarctic Peninsula region, 2001-2007. Mar Ornithol 36:83-97

Marshall GJ (2003) Trends in the Southern Annular Mode from observations and reanalysis. J Climate 16:4134-4143

Nelson LJ, Anderson DR, Burnham KP (1980) The effect of band loss on estimates of survival. J Field Ornithol 51:30-38

Pezzo F, Olmastroni S, Volpi V, Focardi S (2007) Annual variation in reproductive parameters of Adélie penguins at Edmonson Point, Victoria Land, Antarctica. Polar Biol 31:39-45

Reed TE, Warzybok P, Wilson AJ, Bradley RW, Wanless S, Sydeman WJ (2009) Timing is everything: flexible phenology and shifting selection in a colonial seabird. J Anim Ecol 78:376-387

166

Sæther B-E, Bakke Ø (2000) Avian life history variation and the contribution of demographic traits to the population growth rate. Ecology 81:642-653

Salihoglu B, Fraser WR, Hofmann EE (2001) Factors affecting fledging weight of Adélie penguin (Pygoscelis adeliae) chicks: a modeling study. Polar Biol 24:328-337

Sandvik H, Erikstad KE, Barrett RT, Yoccoz NG (2005) The effect of climate on adult survival in five species of North Atlantic seabirds. J Anim Ecol 74:817-831

Sandvik H, Erikstad KE (2008) Seabird life histories and climatic fluctuations: a phylogenetic-comparative time series analysis of North Atlantic seabirds. Ecograhpy 31:73-82

Schofield O, Ducklow HW, Martinson DG, Meredith MP, Moline MA, Fraser WR (2010) How do polar marine ecosystems respond to rapid climate change? Science 328:1520-1523

Stammerjohn SE, Martinson DG, Smith RC, Yuan X, Rind D (2008) Trends in Antarctic annual sea ice retreat and advance and their relation to El Nino-Southern Oscilation and Southern Annular Mode variability. J Geophys Res 113, C03S90, doi:10.1029/2007JC004269

Stearns SC (1992) The evolution of life histories. Oxford University Press, Oxford

Taylor RH, Wilson PR (1990) Recent increase and southern expansion of Adélie penguin populations in the Ross Sea, related to climatic warming. New Zealand J Ecology 14:25-29

Thompson DWJ, Solomon S (2002) Interpretation of recent southern hemisphere climate change. Science 296:895-899

Trivelpiece WZ, Hinke JT, Miller AK, Reiss CS, Trivelpiece SG, Watters GM. (2011) Variability in krill biomass links harvesting and climate warming to penguin populations in Antarctica. Proc Natl Acad Sci USA 108:7625-7628

Turner J, Colwell SR, Marshall GJ, Lachlan-Cope TA, Carleton AM, Jones PD, Lagun V, Reid PA, Iagovkina S (2005) Antarctic climate change during the last 50 years. Int J Climatol 25:279-294

Visser M (2008) Keeping up with a warming world; assessing the rate of adaptation to climate change. Proc R Soc Lond B 275:649-659

Weimerskirch H (2002) Seabird demography and its relationship with the marine environment. In: Schreiber EA, Burger J (eds) Biology of Marine Birds. CRC Press, Boca Raton

167

White GC, Burnham KP (1999) Program MARK: survival estimation from populations of marked animals. Bird Study 46(Suppl.):120-139

Williams TD (1995) The Penguins. Oxford University Press, Oxford

Wilson PR, Ainley DG, Nur N, Jacobs SS, Barton KJ, Ballard G, Comiso JC (2001) Adélie penguin population change in the Pacific sector of Antarctica: relation to sea ice extent and the Antarctic Circumpolar current. Mar Ecol Prog Ser 213:301-309

Woller RD, Bradley JS, Croxall JP (1992) Long-term population studies of seabirds. Trend Ecol Evol 7:111-114

Wynn RB, Josey SA, Martin AP, Johns DG, Yésou P (2007) Climate-driven range expansion of a critically endangered top predator in northeast Atlantic waters. Biol Lett 3:529-533