NORTH PACIFIC RESEARCH BOARD BERING SEA INTEGRATED ECOSYSTEM RESEARCH PROGRAM FINAL REPORT

Transcription

1 NORTH PACIFIC RESEARCH BOARD BERING SEA INTEGRATED ECOSYSTEM RESEARCH PROGRAM FINAL REPORT Three-dimensional spatial distribution of Black-legged Kittiwakes Rissa Tridactyla and Thick-billed murres Uria lomvia and their prey provides insights into seabird population trends at the Pribilof Islands. NPRB BSIERP Project B63 Final Report Rosana Paredes 1, David B. Irons 2, and Daniel D. Roby 3 1 Department of Fisheries and Wildlife, 104 Nash Hall, Oregon State University, Corvallis, OR USA. (250) , rparedes.insley@gmail.com. 2 U.S. Fish and Wildlife Service, Anchorage, AK 99503, USA. November

2 55 ABSTRACT Seabirds breeding at St. Paul Island, located farther from alternative foraging habitats, may be less resilient than those at St. George Island to cope with food shortages predicted by climate change. We tested this hypothesis by analyzing data on foraging ecology, nutritional stress and breeding performance of two piscivorous seabirds, the black-legged kittiwake (surface-feeder) and thick-billed murre (divingfeeder) nesting at two Pribilof Islands with contrasting population trends. St. Paul Is. (declining populations) is three times farther from oceanic habitats than St. George Is. (stable populations). Our approach combined measures of three-dimensional spatial use of marine habitats, diets, corticosterone levels, foraging and breeding performance, and remote sensing of sea-surface height anomalies. During three cold years with expected low prey availability near the Pribilof Islands, kittiwakes and murres made short-day and long-overnight trips with feeding peaking at night or twilight in deep but not shallow regions. Birds from St. George foraged more on oceanic higher-energy prey (myctophids, squid), whose availability was apparently enhanced by deep-water oceanographic features (shelf break, eddies) at night. Birds from St. Paul foraged mainly on the Middle Shelf, increasing foraging and diving effort during daytime (murres) or making extreme overnight trips to the basin (kittiwakes) to compensate for reduced prey. The increase in foraging effort of birds from St. Paul buffered chick survival, however, this came at a cost of elevated adult nutritional stress. Given that nutritional stress is a good predictor of adult survival, our results suggest that this may explain the contrasting population trajectories of the Pribilof colonies under chronic food shortages in the Shelf KEYWORDS Kittiwakes, thick-billed murres, eastern Bering Sea, foraging and diving behavior, biologging, nutritional stress, population trends, eddies, diets, habitat bathymetry CITATION Paredes, R., D. B. Irons, and D.D. Roby Three-dimensional spatial distribution of Black-legged Kittiwakes (Rissa Tridactyla) and Thick-billed murres (Uria lomvia) and their prey provides insights into seabird population trends at the Pribilof Islands. NPRB BSIERP Project B63 Final Report, 119 p. 86 2

3 87 TABLE OF CONTENTS Abstract 2 Table of contents 3 List of Tables 5 List of Figures 6 Study Chronology 8 Introduction 12 Objectives 16 Chapter 1. Usage of oceanic habitats by a surface-feeding seabird with stable population buffered food shortages in shelf habitats Abstract 20 Introduction 22 Methods 24 Results 32 Discussion 36 Chapter 2. Three-dimensional spatial use of marine habitats provides insight into the contrasting population trends of a deep-diving seabird in the Bering Sea Abstract 63 Introduction 64 Methods 67 Results 74 Discussion 76 Conclusions 105 BSIERP and Bering Sea Project connections 110 3

4 Management or policy implications 111 Publications 111 Poster and oral presentations at scientific conferences or seminars 112 Outreach 114 Acknowledgements 115 Literature cited

5 136 LIST OF TABLES Table 1.1. Summary statistics of data-logger deployment and the total number of trips conducted by black-legged kittiwakes at the two Pribilof Islands 47 Table 1.2. Sex differences on trip distances and habitat usage of kittiwakes at the Pribilof Islands, Values in table are mean ± SE 48 Table 1.3. The effects of colony (St. Paul = STP; St. George = STG), habitat (shelf and basin), year (2008, 2009, 2010), and sex on the foraging trip distance of black-legged kittiwakes nesting at the Pribilof Islands. GLMMs with individual as random factor in all tests. Statistically significant (p < 0.05) relationship is shown in bold (see text for post-test analyses) 49 Table 1.4. Breeding and foraging performance of kittiwakes from St. Paul and St. George islands in Values are mean ± SE, with sample size in parentheses 50 Table 2.1. Summary of data loggers deployed in Thick-billed murres at the Pribilof Islands during 2008, 2009 and Table 2.2 Diving effort of thick-billed murres nesting at St. Paul and St. George Islands according to type of trip and foraging habitat in Mean ± SE 92 Table 2.3. Foraging and breeding performance of non-tagged thick-billed murres from St. Paul and St. George islands in Values are mean ± SE, with sample size in parentheses. (*) indicate significance between islands

6 162 LIST OF FIGURES Figure 1.1. Foraging behavior of chick-rearing black-legged kittiwakes nesting at the Pribilof Islands a) GPS tracking (St. Paul = red; St. George = blue) during 2008 (n = 34 tracks), 2009 (n = 58) and 2010 (n = 70); and b) Kernel densities of foraging locations of black-legged kittiwakes at the Pribilof Islands during 2008, 2009, and a) Percentages of volume contour represent total foraging range (95% = green; and 75%= yellow) and core feeding areas (50%= red). Solid color = St. George, pattern color= St. Paul. Each island had birds (1-3 trips/day) per year 51 Figure 1.2. Foraging locations of black-legged kittiwakes from St. Paul and St. George Islands according to time of day and marine habitat: shelf and basin of the southeastern Bering Sea shelf. The rectangle indicates the dark (black), twilight (grey), and daylight (white) periods 53 Figure 1.3. b) Core feeding areas (50% contours) of birds that traveled to the shelf and to basin habitats. Patterns: red = St. Paul; black = St. George; Years: 2010 =, 2009 =, 2008 = 54 Figure 1.4. Maximum trip distance according to shelf and basin habitats. Means ± SE; number of trips inside bars. Significance between colonies (P < ) is indicated by the asterix (see details in Table 3)- 56 Figure 1.5. Eddy basin association with kittiwake foraging a) Examples of individual tracks and foraging locations (circles) overlaid on eddy fields (blue = cyclonic eddies, orange = anticyclonic) matched by date (± 3 days) of data collection b) Frequencies of foraging and random locations according to distance to nearest eddy perimeter (vertical line); negative values indicate locations inside features 57 Figure 1.6. Prey species consumed by tracked kittiwakes during trips to shelf and basin habitats. Percentages are based on the total number of trips/samples (n) to each habitat. Other fish : eelpout, rockfish, and gadids; Crustaceans : amphipods or/and euphausiids; Others : amphipods and sea nettle. Empty samples from stomach lavages were only found from birds that traveled to the shelf 58 Figure 1.7. Diet composition of chick-rearing kittiwakes from St. George (2008 = 41 samples; 2009 = 26; 2010 = 51) and St. Paul Island (2008 = 32; 2009 = 29; 2010 = 35). Categorization of prey by domain location (shelf and basin) was based of diet of tracked birds (Figure 6) 59 Figure 1.8. Acoustically measured biomass density (g/m 2 ) of juvenile pollock integrated from 100 m to the surface. Data were collected from mid-july to mid-august in each year over 110, 10-km long 6

7 transects that were randomly placed within 200 km of St. George Island. Map surfaces were generated using minimum curvature interpolation 60 Figure 1.9. Relation between eddy kinetic energy and myctophid occurrence in black-legged kittiwake diets at the Pribilof Islands (St. Paul = white circles; St. George = black) a) Example of EKE (cm 2 s -2 ) averaged over July 2010 (color) in the bird foraging region. Gray shading denotes the shelf (<200 m depth) b) Correlation between EKE and proportion of myctophids in historical diets. Note that this relationship was found significant only when myctophids were present in bird diets 61 Figure Corticosterone levels of chick-rearing kittiwakes at St. Paul and St. George Islands during and b) Annual eddy kinetic energy in the basin area based in foraging range of birds in the basin habitat. Decreasing stress levels of St. George birds coincide with increase of EKE between 2008 and Figure 2.1. Geographic and bathymetric location of the study colonies, Pribilof Islands (St. Paul and St. George) in the southeastern Bering Sea Continental Shelf, Alaska 94 Figure 2.2 GPS tracks of thick-billed murres at the Pribilofs Islands during 2008, 2009 and 2010 (a). Examples of daytime trips (yellow) and overnight trips (pink and purple). Vertical lines indicate dive locations within tracks 95 Figure Figure 3. Foraging trip frequency (a) and maximum distance (b) of thick-billed murres nesting at the Pribilof Islands according to foraging habitat (Shelf and Slope) and type of trip (day and overnight) during 2009 and (*) Denotes significance between colonies (P < 0.023). Number of trips inside bars. 97 Figure 2.4. Inter-annual differences in the frequency (A) and dive depth (B) of thick-billed murres nesting at St. Paul and St. George Islands according to daylight periods (based in sunset and sunrise) during 2008, 2009 and Figure 2.5. Examples of dive profiles of thick-billed murres from St. Paul and St. George Island during daytime and overnight trips to the Bering Sea Shelf and Slope. Maximum distance of each trip is shown in the top corner of each graph. Deep dives of the trip to Slope occurred on the Shelf during the in and outbound part of the trip 99 Figure 2.6. Hourly diving activity and mean depth of thick-billed murres nesting at St. Paul and St. George Islands during complete trips (day and overnight) to the Shelf (Middle and Outer) and Slope in The Outer Shelf was final destination of overnight trips or transit area from/to the Slope by birds 7

8 from St. George. Color bars indicate daylight periods: black = nocturnal, grey = crepuscular and white = day 100 Figure 2.7. Prey delivered to chicks by thick-billed murres at St. Paul (n = 278) and St. George Island (n = 264) between 2008 and A) Proportion of total prey species and schematic lipid content based on Van Pelt et al. (1998) and Whitman (2010). Osmeridae includes capelin, eulachon, and smelt. Gadids lipid content was assumed to be similar to juvenile pollock. B) Frequencies of prey size by year relative to adult bill-gape length 102 Figure 2.8. Percentage of occurrence of the main prey species consumed by adult thick-billed murres at St. Paul and St. George Islands during 2008 (n = 57 samples), 2009 (n = 78), and 2010 (n = 95) 103 Figure 2.9. Corticosterone levels of thick-billed murres rearing chicks at St. Paul (2008 = 58 individuals, 2009 = 87, 2010 = 240) and St George (2008 = 58, 2009 = 89, 2010 = 96) Islands during 2008 and Means ± SE

9 246 STUDY CHRONOLOGY FYI 2008 March-April Dr. Paredes was hired as a post-doc fellow by principal investigators, Dr. Irons and Dr. Roby, to lead the Seabird Telemetry project-b63. April-May 2008 protocols were written in collaboration with other BSIERP and Patch Dynamics Study (PDS) projects for data collection, GPS casing were tested, equipment/supplies ordered for 1 st field season; and housing arranged at the Pribilof Islands. June-July field-work activities were coordinated with other BSIERP and PDS projects and crew members were trained in all field activities July-August field-work was conducted at St. Paul and St. George Islands; deployment of GPS data loggers with either activity loggers or TDRs to thick-billed murres and kittiwakes. In addition, diet and blood samples, and body measurements were taken of tagged birds. We also provided bird locations for adaptive sampling of prey at sea (PDS B67). September Progress report submitted to NPRB and Pribilof communities. October-December 2008 results summarized and presented in the BSIERP PI meeting at Girdwood. January- February 2009 GPS data loggers were given maintenance, GPS casing re-tested and 2008 presented at the Alaska Marine Science Symposium (AMSS) and Pacific Seabird Group meeting (PSG). FYI 2009 In 2009, the Pribilof Island project joined efforts with the Bogoslof Island project (PDS B77) lead by David Irons and Ann Harding; so all activities were coordinated and standardized among sites. March-April Field-crew members for the 2009 season were selected; and housing ensured at each island. Progress report submitted to NPRB and Pribilof communities. April-May 2009 protocols were written (based in 2008-reviewed protocol) in collaboration with other BSIERP and PDS projects for data collection, and equipment/supplies ordered for the 2 nd 9

10 field season. June-July field-work activities were coordinated with other BSIERP and PDS projects and crew members were trained in all field activities. July-August field-work was conducted at St. Paul, St. George and Bogoslof Islands; deployment of GPS data loggers with either activity loggers or TDRs to thick-billed murres and kittiwakes. In addition, diet and blood samples, and body measurements were taken of tagged birds. We also provided bird locations for adaptive sampling of prey at sea (PDS B67). D. Roby presented 2009 results at the International Ornithological Conference. September Progress report submitted to NPRB and Pribilof communities. October-December 2009 data results summarized and presented in the AMSS and PSG. January- February 2010 GPS data loggers were given maintenance, and results presented at the AMSS and PSG meeting in FYI 2010 March-April Field-crew members for the 2009 season were selected; and housing ensured at each island. Progress report submitted to NPRB and Pribilof communities. April-May 2010 protocols were written (based in 2009-reviewed protocol) in collaboration with other BSIERP and PDS projects for data collection, and supplies ordered for the 3 rd field season. June-July field-work activities were coordinated with other BSIERP projects and crew members were trained in all field activities. July-August field-work was conducted at St. Paul and St. George Islands; deployment of GPS data loggers with either activity loggers or TDRs to thick-billed murres and kittiwakes. In addition, diet and blood samples, and body measurements were taken of tagged birds. D. Roby presented 2009 results at the International Ornithological Conference. September Progress report submitted to NPRB and Pribilof communities. A. Harding and R. Paredes presented results at the 1 st World Seabird Conference. October-December 2010 results were summarized and presented at the AMSS Progress 10

11 report submitted to NPRB and Pribilof communities. GPS data, master file of capture birds, and metadata was delivered to NPRB. January- February Analysis of data was ongoing and inclusion of other projects data was coordinated with other co-authors for preparation of manuscripts FYI 2011 March-April Data data of kittiwakes and murres were analyzed. Progress report was sent to NPRB and Pribilof communities. May data presented at the International Marine Conservation Congress and The Ecosystem Studies of Sub-Arctic Seas (ESSAS) meeting. June-August Preparation and submission of two manuscripts on 2009 comparison of the Pribilofs and Bogoslof Islands. September-October Progress report submitted to NPRB and Pribilof communities. October-December 2010 results were summarized and presented at the AMSS GPS data, master file of capture birds, and metadata was delivered to NPRB. January- February Data analysis for preparation of manuscript on black-legged kittiwake Pribilof data, which was presented at the AMSS FYI 2012 March-April progress report submitted to NPRB and Pribilof communities. May-September The Pribilof manuscript of kittiwakes was revised after co-authors comments for submission to PLOSOne in December. The second draft of the manuscript of Pribilofs of murres has been included in this report after co-authors revision. The two submitted manuscripts on the Pribilof and Bogoslof comparison were revised and one was accepted for publication in Marine Ecology Progress Series in August. Data and metadata of black-legged kittiwakes forage locations and dive parameters of murres were sent to NPRB. 11

12 September Non-Cost extension was granted by NPRB from September 30 to November This extension allowed finishing analysis of 2010 data from other PDS projects. October-November The second manuscript on the Pribilof and Bogoslof comparison was accepted for publication in Deep Sea Research II. The proofs of the MEPS manuscript were sent back to journal in November. Data and metadata of dive locations of murres were sent to NPRB. Final report B63 was submitted to NPRB in November

13 350 INTRODUCTION Seabirds have been widely recognized for their ability to indicate changes in marine ecosystems (Piatt et al. 2007, and references therein). Changes in ocean conditions due to climate (Wanless and Harris 1998, Suryan et al. 2002, Frederiksen et al. 2005) or overfishing (Jackson et al. 2001) can disrupt energy transfer between trophic levels and affect the food supply of seabirds. Responses to these ecosystem changes may be reflected in seabird diet composition (Hatch and Sanger 1992, Bryant et al. 1999, Suryan et al. 2002, but see Renner et al. 2012), foraging behavior (trip distance and duration: Burger and Piatt 1990, Suryan et al. 2000), nutritional stress (assessed by measurements of the stress hormone, corticosterone, Kitaysky et al. 2000, 2007), nesting success (Jodice et al. 2006), and adult survival (Kitaysky et al. 2010, Goutte et al. 2010). Multi-pronged studies that link the foraging distribution and diet of seabirds to the nutritional status, and performance of breeding individuals (reviewed by Barrett et al. 2007, Burger and Shaffer 2008) can provide a better understanding of marine habitats and predatorprey interactions, and should provide a more powerful indicator of changes in the marine ecosystem than any one parameter alone (Piatt et al. 2007). For example, although breeding performance (e.g. chicks fledged/nest) can reflect food availability, these measures can be confounded by both species-specific differences in ability to compensate for food shortage and variation in predator risk among years and colonies (Piatt et al. 2007). Population responses of long-lived seabirds are sensitive to long-term changes in food availability and typically reflect oceanographic changes on inter-annual and decadal scales (Thompson and Ollason 2001, Irons et al. 2008). However, nutritional status of breeding individuals reflects changes in food availability during the reproductive season and determines post-reproductive survival of adult birds, providing a mechanistic link between changes in prey availability and population processes in seabirds (Kitaysky et al. 2007, 2010) The marine environment is characterized by patchy and ephemeral food resources, which vary greatly in abundance, temporal and spatial distribution, and energy value among different habitats (Maravelias 1999, Williams et al. 2001). According to optimal foraging theory (Emlen 1966), the tactics adopted by an individual should maximize the net rate of energy intake, and the spatial and temporal use of marine habitats by predators is therefore expected to reflect the availability of prey. For example, seabirds concentrate their foraging in hot spots that coincide with both transient (e.g. eddies and tides) and permanent oceanographic features (e.g. shelf breaks, seamounts and fronts) where prey tend to be concentrated (Skov et al. 2008, Bost et al. 2009). Higher-abundance fish assemblages have been found 13

14 more often in areas with high eddy activity (measured by eddy kinetic energy; Drazen et al. 2011) indicating these features enhance prey availability for surface-feeding predators in basin habitats (Dell et al. 2012). The tendency of seabird colonies to be close to predictable or productive habitats (Croxall and Wood 2002, Piatt and Springer 2003, Laidre et al. 2008), and the repetitive use of these oceanographic features by chick-rearing birds (Irons et al. 1998) highlight the importance of nearby predictable prey. Local and regional changes in ocean conditions and human activities may affect marine habitats differently; therefore the accessibility to stable prey availability (Mattern et al. 2009) or diverse foraging options may benefit seabirds during reproduction As central-place foragers, seabirds are constrained to commute between the foraging areas and the nest site to feed their chick (Orians and Pearson 1979). Some seabirds have dual or bimodal foraging strategies, short vs. long trips (Chaurand and Welmerskirch 1994, Saraux et al. 2011) equivalent to day vs. overnight trips in some species (Paredes et al. 2005, Zavalaga et al. 2011). Because of time constraints of central-place foragers, the functionality of these strategies is commonly attributed to chick provisioning vs. self-feeding (Welcker et al. 2011). Seabirds adapted for visual feeding usually concentrate foraging during daylight hours (Shealer 2002), and some species also forage at crepuscular and nocturnal hours overlapping with the vertical migration of their prey to sea surface (Regular et al. 2010). Diel vertical migrations (DVM) of fish and zooplankton are nearly ubiquitous in marine systems (Hays et al. 2003); and the magnitude and timing of this migration can vary by taxa (Schabetsberger et al. 2000) and ocean habitat with different bathymetry (Sims et al. 2005). High concentrations of prey and marine predators have been found more often near the surface at night in productive topographic features such as seamounts (Johnson et al. 2008), and more often in ocean basins than in shallow shelf habitats (Dias et al. 2012). Thus, timing foraging at places where vertical migration of prey is more intense would be advantageous for both restoring adult reserves and ensuring a meal for the offspring The Bering Sea is a highly productive ecosystem, supporting large numbers of fish, seabirds, and marine mammals. A number of native coastal communities have cultures based on the marine resources of the Bering Sea, and it also sustains one of the largest fisheries in the United States (Stabeno et al. 2012b). The Bering Sea has experienced considerable climate shifts in recent decades (Hare and Mantua 2000). Climate regime shifts often force fish communities to transition between alternate species (Litzow et al. 2006), and major ecosystem reorganization followed the late 1970 s shift (Anderson and Piatt 1990, Conners et al. 2002). Given the economic and ecological significance of the Bering Sea, a number of 14

15 research projects have focused on the impact of climate change on the Bering Sea ecosystem (Wiese et al. 2012; Hollowed et al. 2012, Hunt et al. 2011, Mueter et al. 2011). The Middle Shelf is the region of the Bering Sea that is most susceptible to climate change because of it s variable marginal ice zone and associated cold pool (Stabeno et al. 2012a). Less frequent shift temperature regimes (Stabeno et al. 2012b) are likely to impact populations of seabirds and northern fur seals Callorhinus ursinus at the Pribilof Islands due to prolonged reduction of prey; fish catches were lower in the cold pool region during cold than warm periods (Stevenson and Lauth 2012). The seabird populations at these islands have either been relatively stable (St. George Island) or declining (St. Paul Island) since 1975 (Byrd et al. 2008). High quality forage fish species, such as capelin Mallotus villosus, have concurrently either disappeared or decreased in the diet of piscivorous seabirds and fur seals, while others, such as walleye pollock Theragra chalcogramma have increased (Sinclair et al. 2008, Renner et al. 2012). The timing of these dietary shifts suggests that changes in forage fish availability may influence the population decline of marine predators breeding at the Pribilof Islands Our current knowledge of the foraging behavior of piscivorous seabirds and their prey at the Pribilof Islands came from studies in the 1980 s from Hunt et al. (1981, 1986, 1996), Schneider et al. (1982, 1984, 1990), and many others that followed this initiative (Decker and Hunt 1996, Kitaysky et al. 2000, Swartzman and Hunt 2000, Takahashi et al. 2008, Jancke et al. 2008, Motohiro et al. 2010, Kokubun et al. 2010). Concurrent long-term data in seabird productivity, population size, and diets collected as part of the monitoring program by the USFWS Alaska Maritime National Wildlife Refuge have provided invaluable information of prey preferences that can be link to reproductive and population processes at the Pribilof Islands (Byrd et al. 2008a, 2008b, Sinclair et al. 2008, Renner et al. 2012). Until this study, however, seabirds nesting at the Pribilof Islands have never before been tracked at sea and, therefore, foraging location was a critical data gap for a better understanding of relationships between prey availability, habitat and population processes. In this study, we compared the inter-annual spatial and vertical distribution of two seabirds nesting at St. Paul and St. George, two geographically associated islands in the Pribilof group but where apex predators may respond to different environmental regimes. The maximum edge of the winter ice on the Bering Sea shelf is generally nearer to St. Paul than to St. George; and the productive edge of the Bering Sea shelf ( Green Belt ; Springer et al. 1996) is three times closer to St. George than St. Paul. To the extent that the influence of ice is greater in the vicinity of St. Paul, seabirds nesting on that island might be differentially affected by the loss of that influence if future warming reduces the incidence of ice in the area, whereas 15

16 the proximity of St. George to the productive Shelf break may buffer predicted food shortages (Byrd et al. 2007; Byrd et al. 2008a). Given that chronic reductions in food supplies can directly affect seabird fertility (Cury et al. 2011) and adult survival mediated by nutritional stress (Kitasky et al. 2010, Goutte et al. 2010) the geographic location of St. Paul and St. George is likely to be an important factor influencing their population growth (Byrd et al. 2008a). We have chosen to examine a diving-feeding seabird, the thick-billed murre (Uria lomvia) and a surface-feeding seabird, the black-legged kittiwake (Rissa tridactyla) because information obtained from different feeding ecotypes could help to characterize effects of climate on foraging resources in marine habitats. We hypothesized that seabird breeding populations at St. Paul, farther located from alternative habitats, might be less resilient than those at St. George to cope with food shortages predicted by climate change. To test this hypothesis we integrated tracking and behavioral data of foraging seabirds with concurrent data on nutritional stress, diets, parental activity budgets, fledging and reproductive success, juvenile pollock biomass/distribution, and energy content of prey types from other BSIERP and Patch Dynamics projects. We also used remote sensing data to assess relative myctophid availability in the Basin using eddy kinetic energy (EKE) and long-term seabird diets. Our study period, was characterized by cold-conditions in the Middle Shelf with low variability in sea-ice extent (Stabeno et al. 2012), and lower forage fish availability compared to warm years (Hollowed et al. 2012). Although these three cold years did not allow for examining the differential effects of sea-ice retreat at St. Paul between cold and warm years, they provided a good opportunity to examine responses of seabirds to prolonged poor foraging conditions, which is key information if foraging conditions near St. Paul is affected due to warming scenarios

17 475 OBJECTIVES ) Determine the foraging locations and trip duration and distance of black-legged kittiwakes and thickbilled murres nesting at both Pribilof islands (St. Paul and St. George) between 2008 and 2010 to address local and inter-annual differences in prey availability. To achieve this goal we studied both species at both islands (see Fig. 2.1 for study area) simultaneously during three years. Specifically we: a) Tracked chick-rearing birds used Global Positioning System (GPS) data loggers in combination with activity loggers in kittiwakes and time-depth recorders (TDR) in murres. Our methods (page and 68-71) and results (pages and 73-75; Figs and 2.2-6) for trip parameters, foraging locations (shelf, slope, basin) and diving behavior (murres; Table 2.2) are shown in Chapter 1 and 2 for kittiwakes and murres respectively. The deployment of GPS in murres proved to be challenging, however, we obtained tracking data for both colonies each year, although we had few tracks in We also obtained dive in all years except at St. Paul in However, it was not until 2009 that we were able to deploy both GPS and TDR in birds at St. George and at both colonies in 2010 (see Table 1.1 and 2.1 for details). To address possible effects of double tagging in diving behavior, we compare dive depth between birds with TDR and birds with GPS and TDR in 2008 (page 71). b) Observed tracked and control birds at the colony to provide an independent measure of colony/year performance and to address possible instrument effect. Our methods (pages and 69) and results (pages 33 and 75-76; Table 1.4 & 2.3; Fig. 2.7) on trip duration, chick-feeding rates, and murre chick diets are shown in Chapter 1 and 2 for kittiwakes and murres respectively. c) Analyzed data on diets of tracked and control birds and juvenile pollock densities and distribution within birds foraging ranges collected by the at-colony BSIERP B65 and at-sea PDS B67 respectively. We also used historical kittiwake diets from the Alaska Maritime National Wildlife Refuge in combination with eddy kinetic energy as proxy of myctophid availability on the basin (Figure 1.9). We measured seabird diet composition (prey type) and availability (abundance and/or distribution) likely to reflect changes in prey availability. Our methods (pages 28-29) and results in diet composition, juvenile pollock density and abundance and eddy kinetic energy (pages 32-33; Figs. 1.8, 1.9, 2.7; 2.8) are shown in Chapter 1 and 2. Although we did not measure prey quality in this report, we analyzed the 17

18 energy content of 2009-kittiwake diet samples in Paredes et al. (2012) based in (Whitman 2011; B77 project). We also used data published in squid and krill abundance and distribution in the study area in (Benoit-Bird et al. 2011) and energy-content of preys according to regions (Whitman 2010) for interpretation of our results in foraging patterns ) Determine relationships between; 1) foraging effort and marine habitat use and 2) diets, stress levels, reproductive success and chick survival that can be used to interpret differences in population trends. To achieve this objective, we integrated results from objective 1 on the spatial and temporal use of marine habitats, foraging and diving effort, diets, pollock density/distribution, myctophid availability, and chick-feeding rates, with data on breeding performance (fledging success), and nutritional stress of adults of both species at each colony collected concurrently by the at-colony BSIERP B65 and stress hormones PDS 67 projects. Our methods (page 31 and 71) and results (pages 35 and 77; Table 1.4 and 2.3; Figs. 1.10; 2.9) of the colony/year comparison of these two variables are shown in Chapter 1 and 2 for kittiwakes and murres respectively. Given differences in proximity of the two colonies to the continental shelf-edge, we predicted that St. Paul birds (declining population) would forage predominantly on the shelf, whereas St. George birds (stable population) would more often commute to more productive oceanic waters. We further predicted that St. Paul birds would have higher foraging (trip distance) and diving (dive depth, frequency) effort provision their chicks at lower rates and quality prey, experience higher levels of nutritional stress, and have lower fledging success than birds breeding on St. George because of the low availability of forage fish on the shelf during cold years (Hollowed et al. 2012) BSIERP hypothesis addressed 2. Climate and ocean conditions influencing water temperature, circulation patterns and domain boundaries impact fish reproduction, survival and distribution, the intensity of predator-prey relationships and the location of zoogeographic provinces through bottom-up processes. Specifically: 2b. Reduced cold pool extent will increase overlap of inner domain forage fish and outer domain piscivores. This prediction could not be tested because the cold pool extended over most of the Middle Shelf in the 18

19 three study years ( 2d. Sporadic reversals to cold conditions (e.g., 1999) will have strong effects on the subarctic community and result in increased interannual variability in abundance and pelagic productivity of piscivorous fish, seabirds and marine mammals. This prediction was partially supported because our study period was part of a prolonged cold regime (Stabeno et al. 2012). 4. Climate and ocean conditions influencing circulation patterns and domain boundaries will affect the distribution, frequency and persistence of fronts and other prey-concentrating features and thus the foraging success of marine birds and mammals largely through bottom-up processes. Specifically: 4a. Climate-ocean changes will displace predictably located, abundant prey (hot spots) necessary for successful foraging by central place (seabirds and fur seals while nurturing young) and hot spot (baleen whales, walrus) foragers. We found partial support for this prediction (see below). 4b. Central place foragers will shift their diet, foraging locations or rookery locations to increase foraging opportunities (based on differential foraging success). We found support for this prediction (see below)

20 Chapter 1: Usage of oceanic habitats by a surface-feeding seabird with stable population buffered food shortages in shelf habitats Manuscript in preparation for submission process to PLOS ONE Running head: Foraging in shelf and oceanic habitats Rosana Paredes 1*, Rachael A. Orben 2, Ann M. A. Harding 3, David B. Irons 4, Daniel D. Roby 1, Robert M. Suryan 5, Rebecca C. Young 6, Kelly Benoit-Bird 7, Carol Ladd 8, Heather Renner 9, Scott Heppell 1 ; Richard A. Phillips 10, Alexander Kitaysky Department of Fisheries and Wildlife, 104 Nash Hall, Oregon State University, Corvallis, OR USA 2 Ocean Sciences Department, University of California Santa Cruz, Long Marine Lab, Santa Cruz, CA 95060, USA 3 Alaska Pacific University, Environmental Science Department, 4101 University Drive, Anchorage, AK 99508, USA 4 U.S. Fish and Wildlife Service, Anchorage, AK 99503, USA 5 Hatfield Marine Science Center, Oregon State University, Newport, OR 97365, USA 6 Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, Alaska, , USA 7 College of Earth, Ocean, and Atmospheric Sciences, 104 CEOAS Admin Bldg Oregon State University, Corvallis, OR 9733, USA 8 Pacific Marine Environmental Laboratory, NOAA, Seattle, WA 98115, USA 20

21 Alaska Maritime National Wildlife Refuge, U.S. Fish and Wildlife Service, Homer, AK 99603, USA 10 British Antarctic Survey, Natural Environment Research Council, High Cross, Madingley Road, Cambridge CB3 0ET, UK *Corresponding author: Rosana Paredes. rparedes.insley@gmail.com Abstract Diverse habitats are expected to raise the stability of ecosystems and its components, however this ecological concept has rarely been tested in marine ecosystems. We tested that proximity of breeding colonies to diverse foraging habitats may buffer seabirds from negative effects of environmental variability on food resources. We compared the foraging ecology, nutritional stress and breeding performance of black-legged kittiwakes (Rissa tridactyla) breeding at two colonies on the Bering Sea shelf with contrasting population trajectories, and located at different distances from productive pelagic habitats. We also measured forage fish densities on the shelf and assessed myctophid availability in the basin using eddy kinetic energy and historical data sets of kittiwake diets. The three study years ( ) were characterized by cold oceanographic conditions and low availability of forage fish on the shelf. In the basin, myctophid consumption of kittiwakes increased in years of high eddy kinetic energy, and foraging locations were more often found near or inside mesoscale eddies than random. As expected by colony location, birds from St. Paul (declining population) farther from oceanic habitats were less able to exploit the basin than those from St. George (stable population) without increasing foraging effort. As a result, kittiwakes at St. Paul ate more shelf forage fish or low quality prey while St. George birds more high-quality oceanic prey in all years. Chick-feeding frequencies and fledging success did not differ between colonies, although long-distance trips were associated with higher nutritional stress of parents at St. Paul than at St. George in 2010, when the former commuted to the basin. These results suggest that a prolonged reduction in forage fish availability on the Bering Sea continental shelf is likely to cause shifts in foraging behavior, reduce adult survival and contribute to the decline of kittiwake populations only at St. Paul. In contrast, kittiwakes at St. George appear to be buffered against reproductive costs to a much greater extent by their closer proximity to predictable prey in the oceanic realm

22 INTRODUCTION Diversity in all its forms number of genes, species, communities, habitats can be expected to give rise to stability of ecosystems and ecosystem components (MacArthur 1955). Ecosystem stability may depend on the ability of communities to contain species or functional groups that are capable of differential responses to environmental changes (McCann 2000). Understanding the factors that govern the stability of populations and communities has gained increasing importance as habitat fragmentation and environmental perturbations continue to escalate due to human activities (Steiner et al. 2011). Here, we examined whether the diversity-stability ecological concept applies to a typical marine top-predator the surface-feeding black-legged kittiwake (Rissa tridactyla) that breeds in continental shelf regions of the Eastern Bering Sea. Characterized by an extensive seasonal sea-ice advance and retreat, the eastern Bering Sea is a diverse and highly productive shelf ecosystem that sustains one of the most important commercial fisheries in the United States ( Declines of seabird and marine mammal populations have been observed at St. Paul, one of the Pribilof Islands, while numbers at St. George are stable after an initial decline over the last 30 years (Byrd et al. 2008; Testa 2011; Satherthwaite et al. 2012). Some of the recent efforts of the multidisciplinary Bering Sea Integrated Ecosystem Research Program have been dedicated to understanding the causes of population declines for a variety of species in an effort to improve prediction of climate impacts on the ecosystem (Wiese et al. 2012) The marine environment is characterized by patchy and ephemeral food resources, which vary greatly in abundance, temporal and spatial distribution, and energy value among different habitats (Maravelias 1999; Williams et al. 2001). According to optimal foraging theory (Emlen 1966), the tactics adopted by an individual should maximize the net rate of energy intake, and the spatial and temporal use of marine habitats by predators is therefore expected to reflect the availability of prey. For example, seabirds concentrate their foraging in hot spots that coincide with both transient (e.g. eddies and tides) and permanent oceanographic features (e.g. shelf breaks and fronts) where prey tend to be concentrated (Skov et al. 2008; Bost et al. 2009). Higher-abundance fish assemblages have been found more often in areas with high eddy activity (measured by eddy kinetic energy; Drazen et al. 2011) indicating these features 22

23 enhance prey availability for surface-feeding predators in basin habitats (Dell et al. 2012). The tendency of seabird colonies to be close to predictable or productive habitats (Croxall and Wood 2002; Piatt and Springer 2003; Laidre et al. 2008), and the repetitive use of these oceanographic features by chick-rearing birds (Irons et al. 1998) highlight the importance of nearby predictable prey. Local and regional changes in ocean conditions and human activities may affect marine habitats differently; therefore the accessibility to stable prey availability (Mattern et al. 2009) or diverse foraging options may benefit seabirds during reproduction As central place foragers, seabirds must commute between their breeding site and foraging grounds (Orians and Pearson 1979), and foraging theory predicts that individuals should balance time and distance traveled to meet energy requirements for both adults and chicks (Schoener 1971). Costs associated with an increase in foraging effort due to a reduction in food availability are well documented. An increase in time spent foraging and dietary shifts to lower quality prey have been linked to reduced chick survival and overall breeding success (Abraham and Sydeman 2004; Pinaud et al. 2005), lower body condition (Weimerskirch et al. 2001; Harding et al. 2011), higher levels of physiological stress (Kitaysky et al. 2007; 2010), and reduced survival of parents (Oro and Furness 2002; Olsson and van der Jeugd 2002). There is a tradeoff between current and future reproductive effort for long-lived iteroparous organisms such as seabirds. They are predicted to restrain their short-term investment in reproduction if it is likely to reduce survival and have a negative impact on lifetime reproductive output (Charlesworth 1980). Support for this prediction is, however, equivocal, with some studies showing that parents prioritize their own survival over success in the current reproductive attempt (Monaghan et al. 1989), and others that they will accept some long-term costs (Reid 1987; Hanssen et al. 2005) In this study, we compared the behavioral, physiological and reproductive responses to food shortages of black-legged kittiwakes breeding at two shelf-based colonies, St. Paul and St. George, that differ in their proximity to deep-water foraging habitat in the Bering Sea Basin. The three study years ( ) were characterized by cold oceanographic conditions (Stabeno et al. 2012) providing a good framework for determining the importance of the basin as a foraging habitat for seabirds. Compared to warm years, key forage fish species for kittiwakes, such as juvenile pollock (Theragra chalcogramma; Sinclair et al. 2008), are less abundant on the continental shelf because they disperse offshore (Hollowed et al. 2012). Higher levels of nutritional stress of piscivorous seabirds at the Pribilof Islands during colder years has been attributed to poor foraging conditions on the continental shelf because kittiwakes are expected to 23

24 forage in the vicinity of their colonies (Benowitz-Frederick et al. 2008; Satterthwaite et al. 2012). However, information on factors affecting the usage and foraging conditions over the basin habitat is lacking. This is needed to fully interpret and predict the consequence of changes in food availability on the shelf as fish species distribution in the Bering Sea continues to change in response to climate change (Mueter and Litzow 2008). We used multiple methods of data collection including GPS tracking, stress levels, fish surveys, nest site monitoring, and remote sensing for identification and estimation of eddy activity in the basin We tested the hypothesis that the proximity of breeding colonies to diverse foraging habitats may buffer seabirds from the negative effects of environmental variability on food resources (Byrd et al. 2008). Given differences in proximity of the two colonies to the continental shelf-edge, we predicted that St. Paul birds (declining population) would forage predominantly on the shelf, whereas St. George birds (stable population) would more often commute to more productive oceanic waters. We further predicted that St. Paul birds would provision their chicks at lower rates, experience higher levels of nutritional stress, and have lower fledging success than birds breeding on St. George because of the low availability of forage fish on the shelf during cold years (Hollowed et al. 2012) Methods Study Site and Species We studied black-legged kittiwakes simultaneously at two Pribilof Islands, St. Paul (104 km 2, 57 7 N W) and St. George (90 km 2, N W), located on the continental shelf of the southeastern Bering Sea, in 2008, 2009 and 2010 (Figure 1). Most field activities were underten during the chick-rearing period, during the last week of June through the first week of August. Kittiwake reproductive success surveys were initiated prior to egg laying in early-june and concluded at fledging at the end of August as part of the ongoing Alaska Maritime National Wildlife Refuge (AMNWR) monitoring program (see Byrd et al. 2008). 24

25 Black- legged kittiwakes nest colonially on sea cliffs and tend to lay clutches of 1 or 2 eggs. Kittiwakes are monogamous with both parents incubating and feeding the chicks, and are sexually dimorphic with respect to size (males are larger and heavier than females; Jodice et al. 2000). Kittiwakes capture prey by plunge diving (< 1m); main prey items include juvenile pollock, sandlance (Ammodytes hexapterus) and myctophids (Myctophidae; Sinclair et al. 2008, Renner et al. 2012). The total population of black-legged kittiwakes at St. Paul (~15,000 individuals) and St. George (~72,000 individuals) in 2005 was estimated from full-island counts in Hickey and Craighead (1977) adjusted with population trend data reported by Byrd et al. (2008b). Neritic waters of the southeastern Bering Sea continental shelf experience annual changes in temperature associated with different oceanographic regimes (cold vs. warm; Stabeno et al. 2012). These changes are associated with sea-ice extent and cold pool distribution on the shelf, resulting in shifts in zooplankton communities and the abundance and distribution of prey for top predators (Hunt et al. 2011; Coyle et al. 2011). Our study encompassed three cold years, when the overall abundance of juvenile pollock was lower or when schools were located farther from the Pribilof Islands (Hollowed et al. 2012; Andrews, unpublished data). Other main prey species of kittiwakes at the Pribilof Islands, such as sandlance, were also poorly represented in the National Marine Fisheries Service Resource Assessment and Conservation Engineering surveys (RACE; Lauth 2010) and this study. Therefore, we focused in juvenile pollock for estimation of fish density on the shelf. The deep oceanic waters of the Bering Sea shelf break and basin (> 200 m) differ from the shelf in oceanography and fish and zooplankton communities (Stabeno et al. 1999). Mesoscale eddies are features characteristic of the Bering Sea basin especially near the shelf break canyons (Schumacher and Stabeno 1994; Mizobata et al. 2008; Ladd et al. 2012). These hotspots of biological production are known to last from days to months in the Bering Sea basin, and provide an important foraging arena for northern fur seals (Callorhinus ursinus) breeding at the Pribilof Islands (Sterling 2009; Nordstrom et al. 2012). Elevated concentrations of forage fish such as myctophids (Murhling et al. 2007; Drazen et al. 2011) and juvenile pollock: Schumacher and Stabeno 1994) have been found to be more often in association with mesoscale eddies than other areas without features Study design We used tracking and activity data for the analysis of foraging behavior, and classified marine habitat 25

26 very broadly into shelf or neritic, and basin or oceanic (includes shelf break). Prey availability was assumed to reflect the direct estimates of juvenile pollock densities on the shelf, and the indirect estimates of myctophid availability based on the historical data of kittiwake diet (10 years frequency of occurrence) and temporal overlap of eddy kinetic energy. For this analysis, we first examined whether kittiwakes forage in basin eddies, and, second, whether annual changes in EKE correlated with the occurrence of myctophids. Kittiwake diet composition was used to determine prey quality and quantity taken in both habitats, and fitness benefits associated with the foraging strategies observed on both islands were assessed on the basis of (a) chick feeding frequencies of non-tracked birds, (b) fledging success (chicks fledged/nest with chicks), (c) breeding success (chicks fledged/nest with eggs), and (d) circulating levels of the steroid hormone corticosterone (CORT) in blood plasma as a measure of the nutritional stress of parents. CORT levels have been shown to be elevated when food availability is low (Kitaysky et al. 2010), and in individuals with depleted endogenous energy reserves (Kitaysky et al. 1999) Field-data Collection In total, 155 adult kittiwakes with chicks aged 5-25 days were captured at their nests using 8-m noose poles and snare traps. Global positioning system (GPS) data loggers (GiPSy 2 and 3 - Technosmart; flat antenna with 250mA battery; size = 41x14x7 mm; weight = 10-12g) were attached to the dorsal surface of four central rectrices using white Tesa tape #4651. The GPS loggers were encapsulated in a shrink tube casing (2 g; 4FT IC8725 ¾ inches clear; Frigid North, AK, USA) prior to deployment to ensure waterproofing, and set to record at a 1 or 180 second sampling interval. Activity (saltwater immersion) loggers (British Antarctic Survey; Mk13 =1.8 g; MK9 = 2.5g) attached to plastic color bands were simultaneously deployed on all GPS birds for determination of feeding behavior. Total instrument weight was 3.4% of average kittiwake body mass at the two colonies (416 ± 2.0g, n = 292). No effect of instrumentation was found on the trip duration and stress levels of kittiwakes from the two colonies in 2009 (Paredes et al Recapture effort started 2 days after deployment; the majority of birds (85%) were recaptured within 2-4 days, and some (n = 17) after 5 to 17 days. At first capture, birds were blood-sampled for subsequent corticosterone analysis (see below), weighed using a 500 g Pesola balance (± 5 g), banded with a USFWS metal band, fitted with loggers, and temporarily marked with livestock paint crayons to facilitate individual identification. Upon recapture, devices were retrieved, birds were again blood-sampled and weighed, and diet samples collected. Blood samples (< 100 µl) were collected from the brachial vein within three minutes of capture according to a standardized technique for corticosterone analysis 26

27 (Benowitz-Fredericks et al. 2008), and for sex determination (Fridolfsson & Ellegren 1999). Samples were centrifuged, and plasma and red cells were preserved frozen for later laboratory analysis at the University of Alaska Fairbanks. Diet samples were obtained from opportunistic regurgitations of captured adults or chicks and in 2008, the stomach contents of adults obtained using water offloading (Wilson 1984; Paredes et al.2012). All samples types were pooled for analysis. Additional samples were obtained from untagged birds to increase sample size; we had a total of 58 and 51 samples for St. Paul and St. George respectively for the three years. Diet samples were weighed using a 100 g balance (± 1 g) immediately after collection and preserved with 70% ethanol for laboratory analysis Yearly, on each island 6-17 plots of kittiwakes, each with 7-45 nests, were monitored every 3-5 days to determine fledging and breeding success (Drummond et al. 2011; Larned et al. 2011), to determine fledging and breeding success A subset of nests (n = per island) with uninstrumented birds was observed from dawn to dusk ( h) for determination of chick feeding frequencies during a total of 3-7 days at both colonies in 2009 and Observations were conducted during early, mid, and late chick-rearing to account for any difference in provisioning behavior associated with chick age. Details of fish data collection and analysis can be found in Benoit-Bird et al. (2011). Briefly, ship-based sampling of the environment and potential prey took place in a 200 km radius of St. George Island from mid-july to mid-august in 2008 and 2009, along 10-km long transects according to a stratified random design. Multi-frequency acoustic data were combined with targeted trawls to estimate the distribution of potential prey around the colonies. 38 khz integrated acoustic scattering identified juvenile pollock, and was combined with the target strength-weight relationship for pollock in the Bering Sea (Traynor 1996) to estimate the average biomass density (g/m 2 ) of pollock for each transect, apportioned to either young of the year or age-1+ pollock using trawl data Laboratory and data analysis Foraging behavior Location data from GPS units were first filtered following a forward-backward 27

28 speed method (McConnell et al. 1992). We used a maximum speed of 80 km/h (the maximum instantaneous velocity recorded from the highest quality GPS fixes [dilution of precision of 1, indicating a wider angular separation of satellites and the highest positional accuracy] was 71 km/h) to cull erroneous positions (median = 5%, range 0 19%, of locations were removed from each individual). Most of the erroneous locations were near the colony, likely due to satellite signal interference from adjacent rock cliffs or vegetation. Not all GPS units recorded positions at the same time intervals, and some units occasionally malfunctioned with larger time gaps in obtaining positions. Due to these complications, we spatially interpolated all tracking data at 10 m linear intervals for time gaps of 30 min or less between consecutive locations. We developed an automated routine using Matlab (The Mathworks, Inc.) to identify and measure central place foraging trips (distance, duration and azimuth) based on specific criteria. We defined a foraging trip based on the activities of birds carrying GPSs and direct observations of adults returning with prey to provision chicks. These criteria required a bird to leave a 0.5 km buffer around the nest site for a minimum of 30 min (the lower 25 th percentile of observed foraging trip durations at all islands) to be included in analyses of foraging activity. We measured the straight-line distance and calculated the trip duration. In situations where a complete round-trip was not recorded because of GPS failure before the bird returned to the colony, we only include the maximum distance estimate in our analysis if the individual returned to within 75% of the maximum distance from the colony for before the track was lost. The bird had to return to within 1 km of the colony for a trip to be considered complete and included in calculations of trip duration. Mean trip distance (range = km, n = 307) of individual kittiwakes was highly correlated with mean trip duration ( h; Pearson = 0.809; P < ) so only the trip distance was included in further analyses. Marine habitats were classified as shelf (neritic) and basin (oceanic); analysis of maximum track location identified a few trips to the shelf break in each year (2009 = 5, 2010 = 5), and these trips were included with the basin habitat category for the analysis. Diurnal differences in trip distance coincided with habitat use; trips to basin occurred primarily overnight and to the shelf throughout the day (Paredes et al. 2012). Therefore, time of day was not included in the models. We used three general linear mixed models (GLMMs) with individual as a random factor to test differences in trip distance. Model 1 included colony, sex and year as fixed factors. There were no differences between the sexes in trip distance by colony and neither by year, except for 2009 (female = 15, male = 18) so this factor was excluded in the habitat analysis (Table 3). The other two models included habitat (shelf and basin) with year (model 2) or colony (model 3) as fixed factors. For the colony comparison, we only used 2010 because there was enough sample size of birds from both colonies using both habitats. 28

29 The spatially interpolated central place foraging trips were combined with activity logger data by subsampling the GPS record to one point every 10 minutes. A time lag of 8 minutes previous and 2 minutes following the activity record was chosen to allow the greatest number of matched points. GPS locations and wet periods were matched by time to determine potential foraging locations. Because activity loggers were attached to a leg band, birds resting on sea surface for bathing or preening were likely to register as completely wet. Therefore, we assumed that birds were not foraging when they were wet for 90% or greater of their time during a ten-minute period if this period was also associated with a previous or later event of similarly high wet values. The marine habitat of foraging locations was determined using ArcGIS and frequencies estimated per island and time of day. Kernel densities of all foraging locations were made for each colony per year using a single smoothing factor (href = m) using least square validation in Home Range Tools for ArcGIS. Because in 2010 we had twice the birds tracked than either 2008 or 2009, we randomly sampled birds in 2010 to standardize sample sizes (15 birds). To avoid trip replication of individuals that had longer tracking periods, we chose trips that were completed by each individual within the first or last 24 h period allowing finishing a trip that started within this period (e.g. basin trips) in all years. We also made kernel densities of foraging locations of trips to each habitat using a single smoothing factor ( m) to visualize overlap of core-feeding areas (50% contours)between islands and years. Foraging locations were also used to estimate distance to the closest eddy perimeter to determine the use of eddy features by black-legged kittiwakes (see eddies below) Diet composition Prey items were measured (length and weight) and identified to the lowest taxonomic level possible. Lengths of partially digested pollock were estimated from otolith size to determine age classes (see Renner et al. 2012). Total frequency of occurrence was calculated as the percentage of all samples containing prey remains in which a prey item occurred. Diet of tracked birds was analyzed according to the furthest foraging habitat used prior to the day of recapture. Percentages of prey species were calculated based on the total number of trips per samples to each habitat Eddy associations and Eddy kinetic energy Gridded sea-surface height anomalies (SSHA) from were downloaded from AVISO ( The ref merged dataset (from consists of delayed-mode, merged data from two satellite missions, Topex/Poseidon and ERS followed by Jason-1 and Envisat. The mapped altimetry data set includes one map every 7 days with a 1/3 spatial resolution on a Mercator grid (Ducet et al. 2000). Eddies were 29

30 identified following Chelton et al. (2011) and visualized using ArcGis software. We determined the association between kittiwake foraging locations and mesoscale eddies over basin waters (> 200 m deep) using eddy data from Chelton et al. (2011). This 18 year ( ) dataset was derived from the sea surface height fields in Version 3 (DT-2010) of the AVISO reference series and included eddies with lifetimes of 4 months or longer ( Eddies were continuously tracked and associated metrics provided at 7-day temporal and 0.25 degree spatial resolutions. Eddy metrics that we used in our analyses included: centroid, radius (used to define perimeter), and whether cyclonic or anticyclonic. Each kittiwake foraging location (see criteria for identifying foraging locations (above) over basin waters, was matched (± 3 days) to the nearest weekly mapped eddy fields. We determined distance to the perimeter of the closest eddy, whether the foraging location was inside or outside of the eddy, and whether the eddy was cyclonic or anticyclonic. We also generated an equal number of normally distributed random locations within the perimeter of the basin foraging locations (defined used convex hull function in Matlab) for respective colonies. We compared the frequency distributions of distance to nearest eddy perimeter to those of foraging locations categorized as inside, near (< 20 km) and outside ( km) the cyclonic or anticyclonic eddy perimeter Eddy kinetic energy (EKE) calculated from SSHA was used as a measure of summer mesoscale variability (Ladd et al. 2012). Assuming geostrophy, the EKE (per unit mass) is estimated from SSHA (η ) as: EKE = [ U 2 g ʹ + V g ʹ ], ʹ U g = g f Δηʹ, Δy V g ʹ = g f Δηʹ, Δx where U g and V g are the geostrophic velocity anomalies, f is the Coriolis parameter, and denotes the time average. Mean EKE was calculated for July of each year delimited by a polygon constrained over the basin and constructed using six corners of maximum bird-track locations for all three tracking years. This was calculated for all island-year data with myctophids in kittiwake diets from , which resulted in 10 years of data including the three years kittiwakes were tracked. The percentage of occurrence of myctophids in each island-year was correlated with EKE values in the same years. We tested the hypothesis that there is no correlation between EKE and myctophid occurrence and the alternative one- 30

31 tailed hypothesis of a positive relationship between the two variables, based on the results of other fish- eddy studies (Drazen et al. 2011; Dell et al. 2012) Corticosterone levels and breeding performance Plasma baseline CORT concentrations (ng ml -1 ) were measured using radioimmunoassay based on the methods of Kitaysky et al. (1999). Each sample was equilibrated with 2000 counts per minute (cpm) of titrated CORT prior to extraction with 4.5 ml distilled dichloromethane. After extraction, the percent titrated hormone recovered from each individual sample (average hormone recovery was >87%) was used to correct final CORT concentrations. Samples, in duplicates, were reconstituted in phosphate-buffered saline gelatin buffer and combined with antibody and radiolabeled. Dextran-coated charcoal was used to separate antibody-bound hormone from unbound hormone. Sensitivity of the assay was 7.8 pg/tube and inter-and intra-assay variability were less than 6% and 2%, respectively. Corticosterone values were log-transformed before analysis using GLMMs to test maximum likelihoods with colony and year as the main factors and individual as a random factor. Chick feeding frequencies, calculated as number of feeds per hour at each nest, were also compared between islands and among years using GLMMs, with individual as a random factor. Fledging success and breeding success were analyzed using GLMMs, with colony and year as the fixed factors and plot as a random factor Statistical analyses were carried out using PASW Statistics 18 and R (version ). We used parametric tests (general linear and mixed models) to compare groups if the residuals met the assumptions for the general linear model (homogeneity and normality). Pearson correlations were used to relate distance and duration of foraging trips, and EKE and myctophid occurrence. Multiple comparisons were undertaken using the post-hoc Tukey HSD test. We used Chi-Square tests with Yates correction for comparing frequencies. Means were expressed ± SE. All comparisons were two-tailed except for the correlation between EKE and myctophid occurrence (one-tailed), and differences were considered significant when P <

32 Results Foraging behavior The majority of GPS and activity logger units deployed on chick-rearing birds were recovered (96%, n = 161), and 75% of these had sufficient data for analysis. About 10% of nests failed during tag deployment each year (Table 1) Usage of shelf & basin foraging habitats Inter-colony comparisons There were differences in the use of marine habitats between islands (χ 2 1= 13.55, P = ; Figure 1a); birds from St. George had more trips to the basin than St. Paul birds in the three years (31%, vs. 13%, n = 153 total trips), and St. Paul birds foraged more often in the shelf habitats (89% vs. 68%, n = 154; Figure 1a). The slight overlap in foraging areas (95% kernel contours; 2008 = 0%; 2009 = 6%; 2010 = 12%) occurred primarily in the basin (Figure 1b). On the shelf, birds from St. Paul foraged more often in the north, northwest and northeast directions (75%, n = 133), and those from St. George to east, northeast, and northwest (71%, n = 106). Most trips (94%, n =68) to the basin were southwest to west in both islands in all years instead of due south. Both colonies had similar diurnal (24 h) distribution of feeding activity. In the basin, birds actively foraged at all hours with more feeding activity (56%; n = 1748) during dark and twilight hours ( ) and a peak in feeding activity during the morning twilight (Figure 2). Over the shelf, feeding activity peaked between 1400 and 1800, and most foraging detections (80%; n = 3542) occurred during daylight hours ( ) Inter -annual comparisons Females and males from both islands foraged in both marine habitats, although there were a higher number of trips by females to the basin and more trips by males to the shelf in 2008 (χ 2 1= 5.823, P = ) and 2009 (χ 2 1= 9.62, P = ). There was no sex bias in habitat use observed in 2010 (χ 2 1= 0.01, P = 1.0; Table 2). Foraging areas, particularly in the basin area, were more spread out in 2010 than 32

33 or 2009 (Figure 1b). Birds that traveled to the basin had some core feeding areas (50% contours) near or over the shelf break (2008 and 2010). Areas used in consecutive years (hot spots) on the shelf were found mainly near each island. In the basin, there were several persistent hot spots (n = 6; km 2 ) used by birds from St. George during 2-3 years (n = 3 hot spots); or from both islands in a year (n = 3; Figure 3) Trip distance In model 1 (sex, year, and colony), we only found significant interaction effects of colony* year and sex*year on trip distance (Table 3). Birds nesting at St. George traveled further distances on average than those at St. Paul in 2008 (F 1, 39 = , P = 0.001) and 2009 (F 1, 77 = 4.716, P = 0.033), but not in 2010 (F 1, 185 = 0.520, P = 0.472; Figure 2). On average, females made longer-distance trips than males in 2009 (F 1, 77 = 6.446, P = 0.013), and no sex differences were found in 2008 and 2010 (P > 0.05). In model 2 (habitat and year), we found significant interaction effect of both factors on trip distance (Table 3). Trips to the basin were significantly longer than those in the shelf in each year (2008: F 1, 39 = , P < ; 2009: F 1, 77 = , P < ; 2010: F 1, 185 = , P < ; Figure 5). Trips to the basin were significantly longer in 2010 (F 2, 236 = 6.022, P = 0.003) than 2008 and 2009 (post-hoc tests: P < 0.05). Trips to the shelf were significantly longer in 2009 (F 2, 65 = 4.638, P = 0.013) than in the other two years (post-hoc tests: P < 0.05). In model 3 (habitat and colony), we found a significant interaction effect of both factors (Table 3). Birds from St. Paul made significantly longer trips to the basin than those from St. George; however, they did not differ on trip distances to the shelf (Table 3 & Figure 5). In 2009, 8 of the 13 birds from St. Paul that foraged on the shelf traveled to a patch of the juvenile pollock found (see below) northwest of the island shelf: ( o azimuth; Figure 8), and these trips were significantly longer than trips conducted in other directions (F 1, 35 = , P < ) Eddy associations Bird tracks and foraging locations overlapped with both eddy types (Figure 6a). There was a high frequency of foraging locations (77%) inside or near (< 20 km) than outside (> 20 km) perimeters of cyclonic eddies. There were fewer foraging locations inside or near anticyclonic eddies (23%) than outside; however, there was a peak (53%) at less than 60 km from the perimeter (Figure 6b). Frequencies of locations inside, near (<20 km from perimeter) and outside (> 20 km) eddies differed between foraging and random locations (cyclonic: χ 2 2 = 40.4; p < ; anticyclonic: χ 2 2= 38.5; p < ). There was 33

34 proportionately higher number of foraging locations near (< 20 km) the cyclonic (χ 2 1 = 14.08; p = ) and anticyclonic (χ 2 1 = 7.98; p = ) than those of random locations (Figure 6a&b) Diet and habitat-prey relationships Diet analysis of the tracked birds (n = 41) showed that age-1 pollock, sandlance, eelpouts (Lycodes sp.), rockfish (Sebastes sp.) and fish offal were obtained on the shelf, while myctophid species (Stenobrachius leucopsarus; Nannobrachium regale; S. nannochir) were obtained primarily in the oceanic habitats (n = 21; Figure 7). Euphausiids (Thysanoessa raschii) and amphipods (Themisto libelulla; T. pacifica) were obtained in combination with forage fish species either on the shelf or basin. Squid (Gonatopsis borealis) were only obtained in the basin in combination with myctophids (Figure 6). Juvenile pollock of both age-classes were completely absent from the diet of kittiwakes, except in 2009 where St. Paul kittiwakes fed on some age-1 pollock (Figure 6). Sandlance was only found in St. Paul diets in 2008 and 2009, and myctophids were the main forage fish found in St. George diets during all years. The high occurrence of myctophids in St. Paul diet in 2010 coincides with an absence of other forage fish (Figure 7) Pollock and myctophid availability The average biomass density of juvenile pollock, including both young of the year and those in their second year, within 200 km of St. George Island was similar between the two years (2008: 43.4 g/m 2, 95%CI ; 2009: 41.7 g/m 2, 95% CI ). This biomass, however, was differentially distributed both in space and across year classes. Age-1+ juvenile pollock were virtually absent in 2008, but occurred in two discrete patches within the study area in The distances from St. Paul Island to these patches were km (n =2 transects) and km (n = 3 transects) for the large and small patches respectively (Figure 8). The higher variance in pollock biomass density observed in 2009 than in 2008 is indicative of the highly aggregated spatial distribution observed in juvenile pollock of both year classes within the study area (Figure 8) Using long-term kittiwake diet data (10 years) at the Pribilof Islands, we found a significant positive correlation between myctophid occurrence and eddy kinetic energy EKE (one-tailed Pearson correlation: 34

35 , P = 0.027; n = 28; Figure 9). In 2003, there was an absence of myctophids in kittiwake diets, and although EKE was the highest (132 cm 2 s- 2 ), birds from St. Paul fed mainly on sandlance (90%, n = 22). Of the three study years, July EKE averaged over the kittiwake foraging region was the lowest in 2008 (47.3 cm 2 s- 2 ), intermediate in 2009 (73.7 cm 2 s- 2 ), and highest in 2010 (93.6 cm 2 s- 2 ; Figure 10b). Myctophid occurrence in kittiwake diets increased from 2008 to 2010 at St. Paul, and from 2008 to 2009 at St. George Island but not in 2010 (Figure 7) Reproductive performance There was no difference in chick feeding frequency between colonies (F 1, 79 = 0.458; P = 0.501) or years (F 1, 79 = 0.877; P = 0.352; Table 4). Fledging success did not differ between colonies (F 1, 725 = 0.771; P = 0.380) but did among years (F 2, 725 = 4.831; P = 0.008; Table 4). Kittiwakes fledged chicks at higher rates in 2010 than 2008 at both islands (post-hoc tests: p < 0.001), but there was no difference in fledging success between 2009 and 2010 or 2009 and 2008 (P > 0.05). Breeding success was influenced by both colony and year (F 2, 1479 = 6.317; P = 0.002; Table 4), and factors were therefore analyzed independently. Birds from St. Paul had higher breeding success than their counterparts at St. George in 2010 (F 1, 506 = ; P < ), with no differences in 2008 and 2009 (P > 0.05). Birds from both islands had lower breeding success in 2009 (post-hoc tests: P < ) than in 2008 and 2010, but there were no differences between 2008 and 2010 (post-hoc tests: P > 0.05) Stress levels We found an interactive effect of colony and year on CORT levels (F 2, 367 = 3.265; P = 0.039), thus factors were analyzed independently. There were no colony differences in 2008 and 2009 (P > 0.05), but St. Paul birds had higher stress levels than St. George birds in 2010 (F 1, 190 = 10.80; P = ; Figure 10). Within colonies, there was no difference in the stress levels of birds from St. Paul between years (F 1, 156 = 1.280; P = 0.28), whereas a negative trend was observed at St. George although this trend was not significant (F 1, 211 = 2.79; P = 0.068), with lower CORT levels in 2010 than 2008 (Figure 10a), which coincided with a positive trend of EKE between (Figure 10b)

36 DISCUSSION We tested the hypothesis that the proximity of breeding colonies to diverse foraging habitats may buffers seabirds from negative effects of environmental variability on food resources. We predicted that birds from the declining breeding population on St. Paul Island, located farther from the oceanic habitat, would be more vulnerable than those from the stable population at St. George Island to climate effects on food availability on the shelf regions (Byrd et al. 2008). Under reduced forage fish availability on the shelf, birds from St. Paul were less able to exploit oceanic waters without increasing foraging effort, which was reflected in their diets. In the basin, kittiwakes foraged more often in proximity to mesoscale eddies, which appeared to enhance availability of rich-lipid myctophids. There was no difference in chick-feeding frequencies and fledging success between colonies; however, birds from St. Paul made longer trips and had higher levels of nutritional stress than birds at St. George in a year when birds from both colonies concentrated feeding in oceanic waters Foraging patterns and prey availability in shelf and basin habitats As expected, given the cold oceanographic conditions characterizing the three study years (Hunt et al. 2011; Stabeno et al. 2012), at-sea survey data indicated that prey availability on the shelf was low. Juvenile pollock, the main forage fish for kittiwakes breeding at the Pribilof Islands (Sinclair et al. 2008, Renner et al. 2012), were patchily distributed on the shelf near the islands in 2008 and 2009, and age-0 fish were more abundant than the larger, more energy-rich, age-1 fish. Although we did not measure juvenile pollock abundance in 2010, other studies found that age-1 pollock were extremely rare near the Pribilof Islands in that year (Andrews, unpublished data). Other forage fish species, such as capelin, were scarce between July-August on the southeastern Bering Sea shelf (this study, Lauth 2010; Hollowed et al. 2012). Kittiwake diet composition reflected the low availability of forage fish species on the shelf during the three study years, with nil or low representation of shelf species (i.e. juvenile pollock and sand lance); instead, the diet was dominated by oceanic species (i.e. myctophids). However, even though age-0 pollock abundance was lower in 2008 and 2009 than reported in warm years (Hollowed et al. 2012), the 36

37 near absence of pollock in kittiwake diet at the Pribilofs was somewhat unexpected given its historical importance as one of the major food resources (Sinclair et al. 2008, Renner et al. 2012). Pollock shoals may have been too deep during daytime (Benoit-Bird et al. 2012), and not readily abundant during their vertical migration for surface feeding seabirds to access as suggested by the low frequency of foraging locations in the shelf at night and twilight periods (Figure 3). During 2009, St. Paul birds accessed the large aggregation of juvenile pollock found northwest of the island but only targeted the larger age-1 pollock which had substantially higher energy densities and energy content per individual (Whitman 2010). Birds from St. George traveled equivalent distances to where pollock patches were found but to the basin where they fed on myctophids, a prey species with higher energy content than age-1 pollock (Whitman 2010). Despite low forage fish availability in the continental shelf regions, birds tracked from both colonies concentrated their daytime foraging on the shelf, feeding relatively close to the breeding colonies in all three years. The diet during the day on the shelf of tracked individuals from both islands was largely comprised of lower quality food (e.g. fish offal) and/or smaller prey (e.g. euphasiids). Only some individuals (mainly from St. Paul) fed on shelf-based forage fish. Presumably daytime trip distances were constrained by the parent s need to frequently provision their offspring (Suryan et al. 2002; Paredes et al. 2012), resulting in foraging on these low quality but closely available prey items. Trips to the basin were primarily overnight and longer than those over the shelf, and all tracked birds concentrated feeding in oceanic waters, mostly during twilight periods, returning with either lipidrich myctophids, or a combination of myctophids and invertebrates (e.g. euphausiids and amphipods). In contrast, although tracked birds returned with different prey species after foraging in the shelf they also had occasionally empty stomachs. These results highlight the contrasting foraging conditions of both marine habitats, and also the importance of the shelf for foraging kittiwakes at both Pribilof Islands during reproduction. Analysis of feeding locations indicate that when traveling to the basin kittiwakes concentrated feeding in oceanic waters (Figure 3), and more frequently in proximity to eddy boundaries (<20 km) than those of random locations (Figure 6b). Bird s feeding locations were more often found inside or near eddy edges of cyclonic eddies, and km from edges of anticyclonic eddies. For surface-feeder kittiwakes, high densities of myctophid assemblies may be more available inside or near the perimeter of cyclonic eddies at night than those of anticyclonic eddies (Murhling et al. 2007). Our findings are in line with consistent use of these features by northern fur seals (Sterling et al. 2009; Nordstrom 2012). The overlap in core-feeding areas observed in the basin between birds from both islands in 2009 and 2010 and a 37

38 repetitive use of the same foraging areas among years by St. George birds suggest that kittiwake targeted such predictable hot spots. Eddy activity appears to influence availability of myctophids in the basin as suggested by the positive relationship found between eddy kinetic energy and myctophid occurrence in the long-term dataset of kittiwake diets. Myctophids and other fish species have been found to be more abundant in areas with high eddy kinetic energy (Drazen et al. 2012; Dell et al. 2012). Birds from both colonies foraged in oceanic habitat in 2010 when EKE was the highest but also when shelf-based forage fish was absent from bird diets and in low abundance based on at-sea sampling. Thus, eddy activity does not appear to drive kittiwake choice to forage in the basin or shelf; however, when birds foraged in the basin, the proportion of myctophids in the diet was positively correlated with eddy kinetic energy, which could have ameliorated nutritional stress in birds from St. George (Figure 10) Colony and sex associations with marine habitats As we predicted based on the geographic location of the colonies, St. Paul birds foraged more often in neritic waters on the shelf whereas birds from St. George foraged consistently in deep oceanic waters. Birds from St. Paul foraged more often on shelf-based forage fish in 2008 and 2009, and switched to prey associated with oceanic waters in 2010 when shelf species were less available. The consistent targeting by birds from St. Paul of a patch of age-1 pollock on the shelf in 2009 highlights the importance of high quality or abundant forage fish for kittiwakes breeding at this colony. In contrast, although St. George kittiwakes also fed on the shelf, their primary source of forage fish (myctophids) was the basin and was associated with nocturnal feeding. These observed foraging patterns are expected to differ during warm years (Benowitz-Fredericks et al. 2008), with birds from both islands conducting shorter distance trips and foraging primarily over the shelf because of the higher availability of prey in these years. The tracking results from this study coupled with long-term dietary preferences recorded at the Pribilofs (Sinclair et al. 2008; Renner et al. 2012) suggest, however, that oceanic habitat may be used to boost acquisition of predictable high-quality prey by birds at St. George Island in all years. We found that the use of marine habitats and distance travelled differed between sexes according to the year. Although females foraged more often in basin habitats than males in 2008 and 2009, no sex differences were observed in Differences in parental roles reported for the black-legged kittiwake (Jodice et al and references therein) may in part explain these results. Conducting long trips to access high-quality prey may be a typical strategy for females, whereas the trip durations of males may be 38

39 more constrained by the need to spend time at the colony protecting their nests from squatters or intruders (Paredes and Insley 2010). Although sex differences in parental duties may influence the foraging strategies of males and females, habitat differences in prey availability often have an over-riding role (Ishikawa and Watanuki 2002; Rey et al. 2012). For example, both males and females foraged in a patch of age-1 pollock northwest of St. Paul in 2009, and both sexes from the two islands foraged in oceanic habitats when availability of myctophids was high. Similarly, no sex differences in foraging habitat selection were found in 2009 at Bogoslof Island in the southwestern Bering Sea, where food availability is considered to be higher than at the Pribilofs (Paredes et al. 2012). Other seabird studies have found that sex differences in foraging behavior are more common when food is limited (Ishikawa and Watanuki 2002). Together, these results suggest that the colony and sex differences in foraging patterns observed in our study might be facultative, related ultimately to the relative availability of prey in the shelf compared with the basin Foraging and breeding performance and adult stress Differences between colonies in trip distance among years were mainly influenced by the frequency of trips to the basin; the ones measured in this study were the longest trips yet recorded for kittiwakes during chick-rearing from any colony (up to 345 km; Suryan et al. 2002; Kortzerca et al. 2010). Birds from St. Paul made shorter trips than those from St. George in 2008 and 2009, but longer to the basin in 2010 when they foraged extensively in oceanic waters. The increase in foraging effort (distance traveled) by kittiwakes from both colonies did not affect feeding frequencies or chick survival; however it coincides with high stress levels for birds at both islands. Not surprisingly, given the importance of the basin for St. George birds, a relationship between EKE and nutritional stress levels was found only at St. George (Figure 10). The decline in stress levels of birds nesting at St. George coincided with an increase in EKE among years, whereas stress levels of birds nesting on St. Paul were consistently high across years. Given that elevated CORT levels reduce over-winter survival in kittiwakes (Kitaysky et al. 2010; Goutte et al. 2010), our results support past studies showing that kittiwakes may accept some long-term consequences of increased reproductive effort if it allows them to fledge chicks successfully despite generally low food availability (Golet et al. 2004). Although the fledging success of birds from both colonies was greater in the year of highest EKE and hence estimated myctophid consumption, the higher stress levels of parents at St. Paul highlights the additional costs associated with longer flights to the basin. Our results also help in interpretation of the long-term population decline of kittiwakes observed on St. Paul, which is paradoxical in that it contradict breeding success (Renner unpublished data). We also 39

40 found that breeding success tend to be higher for kittiwakes at St. Paul than those at St. George and was significant in The higher nutritional stress in the shelf-feeding population, combined with (1) equal fledgling success (2) increased potential overwinter mortality of adults presents a plausible mechanism to explain differences in population trajectories at the Pribilof Islands; a long-term decline in adults during the non-breeding season as opposed to decreased reproductive success (Satterthwaite et al. 2012). Adult birds on St. George (stable population) are buffered to a greater extent by their proximity to productive oceanic waters In conclusion, we found that the foraging patterns and stress levels of kittiwakes nesting at the Pribilof Islands were affected by changes in forage fish availability in the Bering Sea continental shelf. Under conditions of reduced forage fish availability on the shelf, the high availability of myctophids, apparently enhanced by eddy activity in oceanic habitats, may buffer chick survival at both Pribilof colonies, but at a greater cost to birds breeding on St. Paul farther located from oceanic waters. Although fluctuations in oceanographic conditions should be a normal occurrence for long-lived predators breeding on the continental shelf of the Bering Sea, if the prolonged cold-warm regimes observed in recent decades continue into the future, there are likely to be long-term repercussions for several species. Our results give support to the ecological concept that a diversity of available foraging habitats, and a broad forage base are beneficial for the stability of seabird populations Acknowledgments This collaborative study was part of the Bering Sea Integrated Ecosystem program funded by The North Pacific Research Board. We are grateful for the enthusiastic assistance and excellent work of crew members on St George: Brie Drummond, Dean Kildaw, Rob Massangale, Nathan Banfield, Caroline Poli, Vijay Patil, Rolanda Steenweg, Donald Lyons, Ram Papish, Chris Barger and Rob Marshall and on St Paul: John Warzybok, Ine Dorreteijn, Dan Cushing, Kerrith McKay, Ana Santos, Tom Harten, and Alexis Will. In addition, Alexis Will did a great job with data entry and preliminary analysis of behavioral data. We thank Zhenya Kitaiskaia for conducting hormonal assays and Kathy Turco for expert diet analyses. Thanks to Dudley Chelton and Peter Gaube (Oregon University) for providing unpublished data and advice on eddy analysis. We thank Karin Holser (St. George Island Institute), Sally and Chris Merculief (Traditional Tribal Council of St. George Island), Phil Zavadil and Debbie Lestenkof (Aleut Community of St. Paul Island) and Priscilla Wohl and Arina Purcella (Northern Forum) for logistical and financial 40

41 assistance. Thanks to Karen Brenneman and Michelle St. Peters (USFWS-Anchorage) for invaluable expeditor assistance. Bruce Robson, Steve Insley, Scott Shaffer, Jeremy Sterling and Carlos Zavalaga provided technical and academic assistance at early stages of telemetry project. This manuscript benefit from discussions with Jeremy Sterling, Steve Okkonen, and Dudley Chelton about Bering Sea eddies. This study was funded by NPRB BSIERP project B63 and B77 to David Irons and Dan Roby, project B65 to Heather Renner, and project B77 to Alexander Kitaysky. This is contribution No. XXX of the North Pacific Research Board References Abraham CL, Sydeman WJ (2004) Ocean climate, euphausiids and auklet nesting: inter-annual trends and variation in phenology, diet and growth of a planktivorous seabird, Ptychoramphus aleuticus. Mar Ecol Prog Ser 274: Benoit-Bird KJ, Kuletz K, Heppell S, Jones N, Hoover B (2011) Active acoustic examination of the diving behavior of murres foraging on patch prey. Mar Ecol Prog Ser 443: Benowitz-Fredericks ZM, Shultz MT, Kitaysky AS. (2008) Stress hormones suggest opposite trends of food availability for planktivorous and piscivorous seabirds in two years. Deep-Sea Res II 55: Bost CA, Cotté, C, Bailleul F, Cherel Y, Charrassin JB, et al (2009) The importance of oceanographic fronts to marine birds and mammals of the southern oceans. J Mar Syst 78: Byrd GV, Schmutz JA, Renner HM (2008) Contrasting population trends of piscivorous seabirds in the Pribilof Islands: a 30 year perspective. Deep-Sea Res II 55: Charlesworth B (1980) The evolution of age-structured populations. Cambridge University Press. Cambridge, UK Coyle KO, Eisner LB, Mueter FJ, Pinchuk AI, Janout MA, et al (2011) Climate change in the southeastern Bering Sea: impacts on pollock stocks and implications for the oscillating control hypothesis. Fish Oceanogr 20: , [doi: /j x]. Croxall JP, Wood AG (2002) The importance of the Patagonian Shelf for top predator species breeding 41

42 at South Georgia. Aquatic Conserv: Mar. Freshw. Ecosyst 12: doi: /aqc.480 Dell J, Wilcox C, Hobday AJ (2011) Estimation of yellowfin tuna (Thunnus albacares) habitat in waters adjacent to Australia s East Coast: making the most of commercial catch data. Fish Oceanogr 20: Drazen, JC, De Forest LG, Domokos RK (2011) Micronekton abundance and biomass in Hawaiian waters as influenced by seamounts, eddies, and the moon. Deep-Sea Research 58: Drummond BA, Larned AL, Thomson G (2011) Biological monitoring at St. Paul Island, Alaska in 2010 with additional observations at Otter and Walrus Islands. U.S. Fish and Wildl. Serv. Rep., AMNWR 2011/03. Homer, Alaska. Ducet, N, Le Traon PY, Reverdin G (2000) Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and-2. J. Geophys. Res. - Oceans 105, Emlen JM (1966) The role of time and energy in food preference. Am Nat 100: Fridolfsson A-K, Ellegren H (1999) A simple and universal method for molecular sexing of non-ratite birds. J Avian Biol 30: Golet GH, Schmutz JA, Irons DB, Estes JM (2004) Determinants of reproductive costs in the long-lived black-legged kittiwake: a multiyear experiment. Ecol Monogr 74: Goutte A, Angelier F, Welcker J, Moec B, Clément-Chastel C, et al (2010) Long-term survival effect of corticosterone manipulation in black-legged kittiwakes. Gen Comp Endocr 167: Hanssen SA, Hasselquist D, Folstad I, Erikstad KE, (2005) Cost of reproduction in long-lived bird: incubation effort reduces immune function and future reproduction. Proc R Soc Lond B 272: Harding AMA, Welcker J, Steen H, Hamer KC, Kitaysky AS, et al. (2011) Adverse foraging conditions may impact body mass and survival of a high Arctic seabird. Oecologia 167: Hickey JJ, Craighead FL (1977) A census of seabirds on the Pribilof Islands. In: Environmental Assessment of the Alaskan Continental Shelf. Final Reports of Investigators, vol. 2. BLM/NOAA/OMPA, Boulder, pp Hollowed AB, Barbeaux SJ, Cokelet ED, Farley E, Kotwicki S, et al (2012) Effects of climate 42

43 variations on pelagic ocean habitats and their role in structuring forage fish distributions in the Bering Sea. Deep Sea Res II 65-70: Hunt GL Jr, Coyle KO, Eisner L, Farley EV, Heintz R, et al (2011) Climate impacts on eastern Bering Sea food webs: a synthesis of new data and an assessment of the Oscillating Control Hypothesis. ICES J. Mar. Sci. doi: /icesjms/fsr036 Irons DB (1998) Foraging area fidelity of individual seabirds in relation to tidal cycles and flock feeding. Ecology 79: Ishikawa K, Watanuki Y (2002) Sex and individual differ- ences in foraging behavior of Japanese cormorants in years of different prey availability. J Ethol 20: Jodice PGR, Roby DD, Suryan RM, Irons DB, Turco KR, Brown ED, Thedinga JF, Visser GH (2006) Increased energy expenditure by a seabird in response to higher food abundance. Mar Ecol Prog Ser 306: Kitaysky AS, Wingfield JC, Piatt JF (1999) Food availability, body condition and physiological stress response in breeding black-legged Kittiwakes. Funct Ecol 13: Kitaysky AS, Piatt JF, Wingfield JC (2007). Stress hormones link food availability and population processes in seabirds. Mar Ecol Prog Ser 352: Kitaysky AS, Piatt JF, Hatch SA, Kitaiskaia AV, Benowitz-Fredericks ZM, et al. (2010) Food availability and population processes: severity of nutritional stress during reproduction predicts survival of long-lived seabirds. Funct Ecol 24: Kotzerka J, Hatch SA Garthe SE (2010) GPS tracking devices reveal foraging strategies of black-legged kittiwakes. J Ornithol 151: Ladd C, Stabeno PJ, O'Hern JE (2012) Observations of a Pribilof eddy. Deep-Sea Res. I 66, doi: /j.dsr , Laidre KL, Heide-Jørgensen MP, Nyeland J, Mosbech A, Boertmann D (2008) Latitudinal gradients in sea ice and primary production determine Arctic seabird colony size in Greenland. Proc R Soc Lond B 275: Larned AL, Sapora SF, Drummond BA (2011) Biological monitoring at St. George Island, Alaska in U.S. Fish and Wildl. Serv. Rep., AMNWR 2011/02. Homer, Alaska. 43

44 Lauth RR (2010) Results of the 2009 eastern Bering Sea continental shelf bottom trawl survey of groundfish and invertebrate resources. U.S. Dep. Commer., NOAA Tech. Memo. NMFS- AFSC-204, 228 pp. MacArthur RA (1955) Fluctuations of animal populations and a measure of community stability. Ecology 36: Mattern T, Houston DM, Lalas C, Setiawan AN, Davis LS (2009) Diet composition, continuity in prey availability and marine habitat keystones to population stability in the Snares penguin (Eudyptes robustus). Emu 109: McCann KS (2000) The diversity-stability debating. Nature 405: Marvelias CD (1999) Habitat selection and clustering of a pelagic fish: effects of topography and bathymetry on species dynamics. Can J Fish Aquat Sci 56: Monaghan P, Uttley JD, Burns MD, Thaine C, Blackwood J (1989) The relationship between food supply, reproductive effort and breeding success in Arctic Terns. Sterna Paradisaea. J Anim Ecol 58: Muhling B, Beckley L, Olivar M (2007) Ichthyoplankton assemblage structure in two meso-scale Leeuwin Current eddies, eastern Indian Ocean. Deep-Sea Res. II 54: Nordstrom CA, Battaile BC, Cotté C, Trites AW (2012) Foraging habitats of lactating northern fur seals are structured by thermocline depths and submesoscale fronts in the eastern Bering Sea. Deep Sea Res II: doi / j.dsr Olsson O, van der Jeugd HP (2002) Survival in king penguins Aptenodytes patagonicus: temporal and sex-specific effects on environmental variability. Oecologia 132: Orians GH, Pearson NE (1979) On the theory of central place foraging. Pages in D.J. Horn, R.D. Mitchell, and G.R. Stairs, editors. Analysis of ecological systems. Ohio State University Press, Columbus, Ohio, USA. Oro D, Furness RW (2002) Influences of food availability and predation on survival of kittiwakes. Ecology 83: Paredes R, Insley SJ (2010) Sex-biased aggression and male-only care in Brünnich s guillemots and Razorbills. Ibis: 152:

45 Paredes R, Harding AM, Irons DB, Roby DD, Suryan RM, et al (2012) Proximity to multiple foraging habitats enhances seabirds resilience to local food shortages. Mar Ecol Prog Ser 471: Piatt JF, Springer AM (2003) Advection, pelagic food webs and the biogeography of seabirds in Beringia. Mar Ornithol 31: Pinaud D, Cherel Y, Weimerskirch H (2005) Effect of environmental variability on habitat selection, diet, provisioning behaviour and chick growth in yellow-nosed albatrosses. Mar Ecol Prog Ser 298: Reid WV (1987) The cost of reproduction in the Glaucous-winged Gull. Oecologia 74: Renner HM, Meuter F, Warzybok JA, Drummond BA, Sinclair EH (2012) Patterns of change in diets of seabirds in the Pribilof Islands: relationships with environmental conditions and chick survival. Deep-Sea Res II. doi: / j.dsr Rey AR, Pütz K, Scioscia G, Lüthi B, Schiavini A (2012). Sexual differences in the foraging behaviour of Magellanic Penguins related to stage of breeding. Emu 112: Satterthwaite WH, Kitaysky AS, Mangel M (2012) Linking climate variability, productivity and stress to demography in a long-lived seabird. Mar Ecol Prog Ser 454: Schoener TW (1971) Theory of feeding strategies. Annu Rev Ecol Syst 2: Sinclair EH, Vlietstra LS, Johnson DS, Zeppelin TK, Byrd GV, et al (2008) Patterns in prey use among fur seals and seabirds in the Pribilof Islands. Deep-Sea Res II 55: Skov H, Humphreys E, Garthe S, Geitner K, Hamer K, et al (2008). Application of habitat suitability modelling to tracking data of marine animals as a means of analyzing their feeding habitats. Ecol Model 212: Stabeno PJ, Schumacher JD, Salo SA, Hunt GL Jr Flint M (1999) Physical environment around the Pribilof Islands. In: Dynamics of the Bering Sea. T.R. Loughlin and K. Ohtani (eds). Fairbanks: University of Alaska Sea Grant, AK-SG-99-03, pp Stabeno P, Moore S, Napp J, Sigler M, Zerbini A (2012) Comparison of warm and cold years on the southeastern Bering Sea shelf and some implications for the ecosystem. Deep-Sea Res II 65 70:

46 Steiner CF. Stockwell RD. Kalaimani V. Aqel Z Dispersal Promotes Compensatory Dynamics and Stability in Forced Metacommunities. Am Nat 178: Sterling JT (2009) Northern fur seal foraging behaviors, food webs, and interactions with oceanographic features in the eastern Bering Sea. Ph.D. Dissertation, University of Washington, WA Suryan RM, Irons DB, Kaufman, M, Jodice, PGR, Roby DD, et al (2002) Short-term fluctuations in forage fish availability and the effect on prey selection and brood-rearing in the black-legged kittiwake. Mar Ecol Prog Ser 236: Testa JW (2011) Fur seal investigations, U. S. Dep. Commer., NOAA Tech. Memo. NMFS- AFSC-226, 80 ppp. 789± 822. Traynor JJ (1996) Target-strength measurements of walleye pollock (Theragra chalcogramma) and Pacific whiting (Merluccius productus) ICES J Mar Sci 53: Weimerskirch H, Prince PA, Zimmermann L (2001) Influence of environmental variability on breeding effort in a long-lived seabird, the yellow-nosed albatross. Behav Ecol 12: Wiese FK, Van Pelt TI, Wiseman WJ (2012) Bering Sea Linkages. Deep-Sea Res II 65-70:2-5. Williams A, Koslow JA, Last PA (2001) Diversity, density and community structure of the demersal fish fauna of the continental slope off western Australia (20 to 35 S). Mar Ecol Prog Ser 212: Wilson RP (1984) An improved stomach pump for penguins and other seabirds. J Field Ornithol 55: Whitman L (2010) Variation in the energy density of forage fishes and invertebrates from the Southeastern Bering Sea. Ph.D Dissertation. Oregon State University. Corvallis, OR

47 Table 1. Summary statistics of data- logger deployment and the total number of trips conducted by black- legged kittiwakes at the two Pribilof Islands. St. Paul St. George Birds with GPS/activity data- loggers- deployed Data- logger units recovered Bird abandoned/nest failed Data- logger units with sufficient data for analysis Total number of trips

48 Table 2. Sex differences on trip distances and habitat usage of kittiwakes at the Pribilof Islands, Values in table are mean ± SE Female Male Female Male Female Male Trip distance (km) 57.7± ± ± ± ± ± 7.1 Number of trips % Shelf %Basin Number of birds Significant differences are shown in bold (P < 0.001)

49 Table 3. The effects of colony (St. Paul = STP; St. George = STG), habitat (shelf and basin), year (2008, 2009, 2010), and sex on the foraging trip distance of black- legged kittiwakes nesting at the Pribilof Islands. GLMMs with individual as random factor in all tests. Statistically significant (p < 0.05) relationship is shown in bold (See text for post- test analyses) Effect df F Error df P Outcome Model 1 Colony Sex Year Sex x colony Sex x year F>M (2009) Colony x year STG>STP (08&09) STP> STG (2010) Sex x year x colony Model 2 Habitat Basin > Shelf Year Habitat x year Basin: 2010>2008 Shelf: 2009>2008 & 10 Model 3 Colony Habitat (2010) Colony x habitat STP > STG basin STP = STG shelf 49

50 Table 4. Breeding and foraging performance of kittiwakes from St. Paul and St. George islands in Values are mean ± SE, with sample size in parentheses island year island x year St. Paul St. George St. Paul St. George St. Paul St. George P P P Chick feeding rate 0.10 ± ± ± ± (feeds h - 1 ) (18) (17) (23) (25) Fledging success 0.39 ± ± ± ± ± ± (chicks fledged per nest w/chicks) (229) (110) (29) (14) (235) (114) (10>08) Breeding success 0.27 ± ± ± ± ± ± (chicks fledged per nest w/eggs) (339) (154) (358) (130) (327) (181) Significant results are shown in bold. Fledgling success was higher in 2010 than 2008 in both islands (post- hoc tests < 0.001). Breeding success was significantly lower in 2009 than 2008 and 2010 in both islands (post- hoc tests p < 0.006). St. Paul s breeding success was higher than St. George s in 2010 (p < ) only.

51 a) b) 2008 Shelf 2008 Basin

52 Figure 1. Foraging behavior of chick- rearing black- legged kittiwakes nesting at the Pribilof Islands a) GPS tracking (St. Paul = red; St. George = blue) during 2008 (n = 34 tracks), 2009 (n = 58) and 2010 (n = 70); and b) Kernel densities of foraging locations of black- legged kittiwakes at the Pribilof Islands during 2008, 2009, and a) Percentages of volume contour represent total foraging range (95% = green; and 75%= yellow) and core feeding areas (50%= red). Solid color = St. George, pattern color= St. Paul. Each island had birds (1-3 trips/day) per year

53 Number of foraging locations Shel N = :00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22: Time of day Number of foraging locations Basin N = :00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22: Time of day Figure 2. Foraging locations of black-legged kittiwakes from St. Paul and St. George Islands according to time of day and marine habitat: shelf and basin of the southeastern Bering Sea shelf. The rectangle indicates the dark (black), twilight (grey), and daylight (white) periods

54 Shelf trips

55 Basin trips 1445! Figure 3. Core feeding areas (50% contours) of birds that traveled to the shelf and to basin habitats. Patterns: red = St. Paul; black = St. George; Years: 2010 =, 2009 =, 2008 =

56 " 200" 2008" 2009" *" C. Ba rge r Trip"Distance"(km)" 150" 100" 2010" *" 50" 0" 27" 37" 69" 5" 27" 74" 2" 1" 17" 7" 14" 27" 29" 38" 86" 12" 41" 101" St."Paul" St."George" St."Paul" St."George" St."Paul" St."George" 1455 Shelf" Basin" Total" Figure 4. Maximum trip distance according to shelf and basin habitats. Means ± SE; number of trips inside bars. Significance between colonies (P < ) is indicated by the asterix (see details in Table 3)

57 1467 a) b) Cyclonic Anticyclonic Figure 5. Eddy basin association with kittiwake foraging a) Example of individual tracks and foraging locations (circles) overlaid on eddy fields (blue = cyclonic eddies, orange = anticyclonic) matched by date (± 3 days) of data collection b) Frequencies of foraging locations according to distance to nearest eddy perimeter (vertical line); negative values indicate locations inside features. 57

58 1493 SHELF&(&n&=&19)& Age$1& pollock& 17%& Empty& 17%& Fish&offal& 22%& Other&fish& 22%& Crustaceans&& 10%& 1494 Sandlance&&& 11%& BASIN+(n+=+22)+ Others+ 9%+ Empty+ 0%+ Myctophids+ &+ crustaceans+ 27%+ Myctophids+ 46% Myctophids+ &+squid+ 18% Figure 6. Prey species consumed by tracked kittiwakes during trips to shelf and basin habitats. Percentages are based on the total number of trips/samples (n) to each habitat. Other fish : eelpout, rockfish, and gadids; Crustaceans : amphipods or/and euphausiids; Others : amphipods and sea nettle. Empty samples from stomach lavages were only found from birds that traveled to the shelf

59 %"Occurrence" 100" 80" 60" 40" 20" 0" 2008" 2009" 2010" St."George" age+0" age+1" Sandlance" Fish"offal" Myctophid" Other"fish" Euphaussid"Amphipod" 1508 shelf" basin" shelf"or"basin" %"Occurence" 100" 80" 60" 40" 20" 0" St."Paul" 2008" 2009" 2010" age+0" age+1" Sandlance" Fish"offal" Myctophid"Other"fish"Euphaussid"Amphipod" "Pollock" shelf" basin" shelf"or"basin" Figure 7. Diet composition of chick-rearing kittiwakes from St. George (2008 = 41 samples; 2009 = 26; 2010 = 51) and St. Paul Island (2008 = 32; 2009 = 29; 2010 = 35). Categorization of prey by domain location (shelf and basin) was based of diet of tracked birds (Figure 6)

60 Figure 8. Acoustically measured biomass density (g/m 2 ) of juvenile pollock integrated from 100 m to the surface. Data were collected from mid-july to mid-august in each year over 110, 10-km long transects that were randomly placed within 200 km of St. George Island. Map surfaces were generated using minimum curvature interpolation

61 a) b)

62 Figure 9. Relation between eddy kinetic energy and myctophid occurrence in black-legged kittiwake diets at the Pribilof Islands (St. Paul = white circles; St. George = black) a) Example of EKE (cm 2 s -2 ) averaged over July 2010 (color) in the bird foraging region. Gray shading denotes the shelf (<200 m depth) b) Correlation between EKE and proportion of myctophids in historical diets. Note that this relationship was found significant only when myctophids were present in bird diets

63 a) 1569 b) Figure 10. Corticosterone levels of chick-rearing kittiwakes at St. Paul and St. George Islands during and b) Annual eddy kinetic energy in the basin area based in foraging range of birds in the basin habitat. Decreasing stress levels of St. George birds coincide with increase of EKE between 2008 and

64 Chapter 2: Three-dimensional spatial use of marine habitats provides insight into the contrasting population trends of a deep-diving seabird in the Bering Sea Running head: Habitat usage of murres with contrasting populations Rosana Paredes 1*, David B. Irons 2, Daniel D. Roby 1, Yann Tremblay 3, Rebecca Young 4, Ann M. A. Harding 5, Rachael A. Orben 6, Robert M. Suryan 7, Heather Renner 8, Alexis Will 4, Alexander Kitaysky Department of Fisheries and Wildlife, 104 Nash Hall, Oregon State University, Corvallis, OR USA. 2 U.S. Fish and Wildlife Service, Anchorage, AK 99503, USA. 3 Institut pour la Recherche et le Développement (IRD) UMR Exploited Marine Ecosystems. France. 4 Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, Alaska, , USA. 5 Alaska Pacific University, Environmental Science Department, 4101 University Drive, Anchorage, AK 99508, USA. 6 Ocean Sciences Department, University of California Santa Cruz, Long Marine Lab, Santa Cruz, CA 95060, USA. 7 Oregon State University, Hatfield Marine Science Center. Newport, OR 97365, USA. 8 Alaska Maritime National Wildlife Refuge, U.S. Fish and Wildlife Service, Homer, AK 99603, USA

65 Abstract Divergent long-term population trajectories of thick-billed murres (Uria lomvia), at the Pribilof Islands are paradoxical because the remarkable flexibility in foraging behavior of this dive-pursuit seabird often buffers reproductive success. We hypothesized that the oceanographic location of St. Paul Is. (declining population), a shelf-colony farther from alternative oceanic habitats, limits adults food supplies resulting in increased foraging effort and nutritional stress during food shortages on the continental shelf when compared to St. George Is. (stable population). Our study encompassed three cold years ( ) with expected reduced food availability for piscivorous seabirds breeding on the Pribilof Islands. Our approach combined measures of spatial (GPS-tracking) and vertical (TDR-diving) use of foraging habitats, diets of chicks and adults, adult time budgets (nest attendance, and chick feeding frequency), fledging success and adult nutritional stress. We found that birds from St. Paul foraged entirely on the Middle Shelf (100 m depth), and dove deeper and longer during extended daytime trips compared to their counterparts on St. George. Although there were no colony differences in total distance traveled and diving effort overnight, birds from St. George fed on rich-lipid prey probably more abundant during nocturnal hours on the Outer Shelf and Slope ( 200 m). In contrast, there was negligible nocturnal diving activity in the Middle Shelf suggesting that birds from St. Paul were more restricted for self-feeding. As predicted, chicks from both colonies were fed as often and with equivalent quality and size of prey in both colonies resulting in similar fledging success. Nevertheless, the increased foraging effort and lower quality prey of birds from St. Paul mirrored higher levels of nutritional stress in two of the three study years. Given that nutritional stress appears to be a good predictor of adult survival, we propose that chronic reduction in food availability on the Middle Shelf may in part explain the divergent population trends of thick-billed murres at the Pribilof Islands

66 1635 Introduction The southeastern Bering Sea is a highly productive ecosystem, supporting large numbers of fish, seabirds, and marine mammals. A number of native coastal communities have cultures based on the marine resources of the Bering Sea, and it also sustains one of the largest fisheries in the United States (Stabeno et al. 2012; Weise et al. 2012). The Bering Sea has experienced considerable climate shifts in recent decades (Hare and Mantua 2000). Climate regime shifts often force fish communities to transition between alternate species (Litzow et al. 2006), and major ecosystem reorganization followed the late 1970 s shift (Anderson and Piatt 1990, Conners et al. 2002). Given the economic and ecological significance of the Bering Sea, a number of research projects have focused on the impact of climate change on the Bering Sea ecosystem (Hollowed et al. 2012; Hunt et al. 2011; Mueter et al. 2011). The Middle Shelf is the region of the Bering Sea that is most susceptible to climate change because of its annually varying marginal ice zone and cold pool extent (Stabeno et al., 2012a). Less frequent shift temperature regimes (Stabeno et al. 2012) are likely to impact populations of seabirds and northern fur seals (Callorhinus ursinus) at the Pribilof Islands due to prolonged reduction of prey; fish catches were lower in the cold pool region during cold than warm periods (Stevenson and Lauth 2012). The seabird populations at these islands have either been relatively stable (St. George Island) or declining (St. Paul Island) since 1975 (Byrd et al. 2008). High quality forage fish species, such as capelin (Mallotus villosus), have concurrently either disappeared or decreased in the diet of piscivorous seabirds and fur seals, while others, such as walleye pollock (Theragra chalcogramma) have increased (Sinclair et al. 2008; Renner et al. 2012). The timing of these dietary shifts suggests that changes in forage fish availability may influence the population decline of marine predators breeding at the Pribilof Islands. Although there is evidence that variation in food supply underlies many of the population changes observed in seabirds (Sandvik et al. 2005; Kitaysky et al. 2010), the mechanisms by which food availability influences population size are less understood because of the intensive long-term studies required to determine the recruitment, immigration, emigration and survival of these long-lived species (Satterthwaite et al. 2012). Moreover, growing evidence that the tradeoff between current and future reproduction can vary both among species (Weimerskirch et al. 2001; Hanssen et al. 2005), and among populations (Coulson 2002) suggests that knowledge of reproductive tradeoffs at given locations is required to fully interpret observed changes in population size. For example, whilst some studies confirm the assumption that long-lived seabird parents should reduce or terminate current reproductive investment 66

67 in order to safeguard future survival if breeding conditions are poor (Monaghan et al. 1989), a number of studies have shown that parents may increase foraging effort during reproduction even if there is an associated reduction in survival or future fecundity (Reid 1987, Golet et al. 1998, Hanssen et al. 2005). Recent studies have successfully linked the nutritional stress of adult seabirds (reflected in circulating levels of the hormone, corticosterone (CORT) with demographic parameters (Kitaysky et al. 2010; Satterthwaite et al. 2012). There is good evidence that exposure of animals to high levels of CORT over an extended period can reduce immune functionality (Saino et al. 2003), cognitive abilities of individuals (Kitaysky et al. 2003), and survival (Romero and Wikelski 2000; Brown et al. 2005; Kitaysky et al. 2007). Behavioral data has also been shown to be an effective parameter for identifying the proximate causes of population change (Lewis et al. 2006), with some measurements of foraging behavior providing a useful indication of prey availability (Piatt et al. 2007). As central-place foragers, seabirds are constrained to commute between the foraging areas and the nest site to feed their chick (Orians and Pearson 1979). Some seabirds have dual or bimodal foraging strategies, short vs. long trips (Chaurand and Welmerskirch 1994; Saraux et al. 2011) equivalent to day vs. overnight trips in some species (Paredes et al. 2005; Zavalaga et al. 2011). Because of time constraints of central-place foragers, the functionality of these strategies is commonly attributed to chick provisioning vs. self-feeding (Welcker et al. 2011). Seabirds adapted for visual feeding usually concentrate foraging during daylight hours (Shealer 2002), and some species also forage at crepuscular and nocturnal hours overlapping with the vertical migration of their prey to sea surface (Regular et al. 2010). Diel vertical migrations (DVM) of fish and zooplankton are nearly ubiquitous in marine systems (Hays et al. 2003); and the magnitude and timing of this migration can vary by taxa (Schabetsberger et al. 2000) and bathymetry of the ocean habitat (Sims et al. 2005). High nocturnal concentrations of prey and marine predators have been found more often near productive topographic features such as seamounts (Johnson et al. 2008), and more often in ocean basins than in shallow shelf habitats (Dias et al. 2012). Thus, timing foraging at places where vertical migration of prey is more intense would be advantageous for both restoring adult reserves and ensuring a meal for the offspring Thick-billed murres raise a single chick that needs to be fed 3-4 times/d, and brooded constantly to protect it from predators during its extremely short period of development in the nest (15-20 days; Harris and Birkhead 1985; Paredes et al. 2005). In spite of these time constraints, murres have shown some ability to buffer chick-feeding rates by allocating more time to foraging when food availability decreases (Burger 67

68 and Piatt 1990; Zador and Piatt 1999; Harding et al. 2006). Murres are flying seabirds adapted for deep diving; their short wings maximize underwater propulsion when pursuing prey, but result in high wing loading, making aerial flight very costly (Pennycuick 1987; Elliot et al. 2013). Given the energetic costs associated with flight and life history traits, food availability at commuting distance from the colonies is an important constraint for murres during reproduction. Foraging and diving behavior have been extensively studied in thick-billed murres (Paredes et al. 2005; Elliot et al. 2007, 2010; Takahashi et al. 2008; Motohiro et al. 2010), however no study to date has been able to measure simultaneously the spatial and vertical use of marine habitats. We studied thick-billed murres at the Pribilof Islands, St. Paul and St. George, both located on the southeastern Bering Sea Middle Shelf but at different distances from the productive continental Shelf break (Springer et al. 1996). Our study encompassed three cold years ( ; Stabeno et al. 2012) with expected low availability of key forage fish species in the Continental Shelf (Hollowed et al. 2012) and high levels of nutritional stress in piscivorous seabirds at the Pribilof Islands (Benowitz-Frederick et al. 2008; Satterthwaite et al. 2012), providing a good opportunity for examining foraging effort allocation of thick-billed murre colonies with divergent population trajectories. We hypothesized that the oceanographic location of St. Paul (declining population), a shelf-colony farther from alternative oceanic habitats, limits adults food stocks resulting in increased foraging effort and nutritional stress during food shortages on the continental Shelf compared to St. George (stable population). Our approach combined measures of spatial (GPS-tracking) and vertical (TDR-diving) use of foraging habitats (bathymetry), diets of chicks and adults, adult time budgets (nest attendance, and chick feeding frequency), fledging success and adult nutritional stress. We predicted that the higher flight costs (range and total distance traveled) associated with commuting to deeper waters in the Slope regions (Figure 1) for St. George birds during overnight trips will be outweighed by (a) access to more abundant prey than on the shelf, and (b) lower diving effort (shallow and short dives), resulting in lower levels of nutritional stress than at St. Paul. Given that murres need to feed their chicks often and during day-light hours, we predicted that parents from both St. Paul and St. George would (a) forage in near shelf regions during the day-time, and (b) feed on similar prey to the chicks resulting in similar fledging success between the two islands. Knowledge of the costs and benefits of the different foraging strategies will help identify mechanisms explaining the different population trends observed at the two islands

69 1728 Methods Study site and species Thick-billed murres were studied simultaneously at the Pribilof Islands, St. Paul (104 km 2, N W) and St. George (90 km 2, N W) between July and August in 2008, 2009 and Additional data on fledging success was obtained from June-September each year as part of the monitoring program of the Alaska Maritime National Wildlife Refuge. Both Pribilof Islands are located in the Middle Shelf ( m isobaths) of the Southeastern Bering Sea Shelf about 70 km from each other, however St. Paul is three times further from the adjacent Outer Shelf (~30 km; m) and the Slope (~ 90 km; >1000 m, Figure 1) than St. George. The edge of the continental shelf has been described as the Bering Sea Green Belt, with the highly productive habitat resulting from physical processes such as crosswise circulation and eddies in the Bering Slope Current, tidal mixing, and advection and upwelling that bring nutrients into the euphotic zone (Coachman and Walsh, 1981; Schumacher and Reed, 1992; Springer et al. 1996). High productivity along the shelf break supports large numbers of zooplankton, fish and squid (Radchenko, 1992; Sinclair and Stabeno, 2002), which in turn attracts high numbers of marine mammals and seabirds (Schneider, 1982; Piatt and Springer, 2003) Field data collection Data loggers We captured murres that were rearing 5-15 days old chicks. Only one member of a pair was captured during daylight hours using a telescoping noose-pole. At initial capture, each bird was weighed (± 1g) and one or two data loggers were attached; a Time-Depth Recorder (TDR), and a Global Positioning System (GPS; Table 1). In 2008, birds from both islands were deployed with TDRs and few birds (n = 4) with both data loggers. In 2009, both instruments were attached to birds on St. George and only GPSs on St. Paul. In 2010, both data loggers were deployed on birds from both islands. The TDR was attached to a plastic leg band (Protouch) using zip-ties (British Antarctic Survey, modified method), and the GPS logger attached to the central back feathers using black Tesa tape (Paredes et al. 2005). GPS loggers were water proofed prior to attachment using shrink heat tubing (4FT IC8725 ¾ inches clear; Frigid North, AK, USA), heated to mould tightly around the logger and minimize buoyancy. 69

70 Handling time took ca min, and less than 20min in all cases. Recapture efforts started approximately 24hrs post deployment, and birds were recaptured on average after 48 hours (St. Paul = 42 ± 3.6, n = 69; St. George = 54 ± 5.2, n = 69). Upon recapture, all instruments were removed, and body mass was measured again. All individuals were banded with a unique USFWS metal band prior to release. Lotek TDRs LAT 1500 and 2500 (mass = g, dimensions = 8 x 32-35mm) recorded depth and time every 1-3 sec, with an absolute pressure accuracy of +/- 1% of full scale. TechnoSmart Gipsy-2 (flat antenna: dimensions 23 x 15 x 6 mm; total mass = g; 500 ma battery) recorded date and time, position in decimal degrees (latitude, longitude) at intervals of 1-60 sec. Longer recording intervals were more often used at St. George to obtained complete trips to Slope. Diets Chick diets were determined by identification of prey delivered by adults. Murre parents deliver a single whole prey item, which is held lengthwise in the bill and is usually visible, allowing identification and estimating its relative size. Prey was identified during all-day watches dedicated to observations of chick diet, and ancillary observations made during all-day adult time-budget watches. A telescope with a zoom lens or 10x42 binoculars were used to identify prey items during the observations sessions. All prey were identified to the lowest possible taxonomic level and visually assigned to a size category relative to the gape length of the parent. The three size categories were a) small, (b) large, or (c) medium (equal to gape). Adult diet were determined using the water off-loading (lavage) method described in detail by Wilson (1984) and for murres by Ito et al. (2010), with the process repeated twice to ensure as complete emptying of the stomach as possible (Neves et al. 2006). Both tagged birds and non-tagged were lavaged, and tagged birds were only sampled during their second capture. Total numbers of birds diet-sampled at each island were: St. Paul n = 627, St. George n = 508. Diet results are presented as percent occurrence (defined as the percentage of samples containing at least one prey type). Parental Time budgets We measured parental attendance (bird-min h -1 ), chick-feeding frequency (feeds hr -1 ) and trip duration (day and overnight) at each colony. Marked birds at each colony were observed on a plot containing 7-15 breeding pairs of murres from sunrise to sunset, on 3-6 observation-days scheduled to match the early, mid and late chick-rearing phase (Birkhead and Nettleship, 1987). Half a day observations were included to record return times of overnight trips. Observations of tagged and control birds were simultaneously done in 2008 and 2009 to address data logger effects on bird behavior. During each session, we recorded each parent s arrival and departure time at the colony, and the time of food 70

71 delivery to chicks. Nest attendance was measured in bird-minutes per hour (Harding et al. 2007; 2013). Fledging Success Fledging success (number of chicks fledged/nests with chicks) was determined at each colony from data recorded during regular observations by the Alaska Maritime National Wildlife Refuge (Byrd et al. 2008b). The same nest-sites (20-30) grouped into plots (8-12) were observed in the same manner each year. Nests were checked every 3-6 days, and chicks were considered to have fledged if they disappeared from the nest-site more than 15 d after hatching (Byrd et al. 2008b). Corticosterone levels We measured circulating levels of baseline CORT from adults during the chickrearing period at each colony to determine the levels of nutritional stress incurred by parents. Tagged birds were sampled after deployment (St. Paul n = 65; St. George n = 51). We also sampled non-tagged birds ( control, St. Paul n = 180; St. George n = 192) captured once to determine data logger effect. All birds were sampled according to a standardized technique (Benowitz-Fredericks et al. 2008), with a blood sample (< 500 µl) collected within three minutes of capture. After blood collection, all samples were centrifuged, and plasma was preserved frozen for later analysis at the University of Alaska Fairbanks Data Analysis We used tracking data from GPS loggers and diving data from TDRs to determine colony and yeardifferences in the spatial and vertical foraging allocation. GPS locations were filtered and interpolated (see Harding et al for details) using a Matlab program to calculate the maximum trip distance (straight distance between the nest site and the most distant location), total distance traveled (cumulative distance between locations) and trip duration of complete trips in each year. In situations where a complete round-trip was not recorded because of GPS failure, we still included the maximum distance estimate in our analysis if the individual began returning to the colony (to within 75% of the maximum distance). Trips were categorized by time of day ( type : day and overnight) and habitat ( foraging location : shelf and slope) based on the most distant position. An overnight trip was defined as beginning on one calendar day and returning the next. Dive data were processed using a purpose built zero-offset correction algorithm and analysis program written in MATLAB (IKNOS-DIVE, IKNOS toolbox, Y. Tremblay unpubl.). A dive was deemed to have occurred when maximum depth was 2 m (Falk et al. 2000, Elliot et al. 2008). Dives were classified 71

72 according to daylight periods (day, crepuscular and nocturnal; Harding et al. 2013) in 2008, 2009 and Daily time of sunrise, sunset, and twilight at each island were obtained from the Astronomical Applications Department, U. S. Naval Observatory,Washington, DC Nocturnal dives were defined as occurring between evening and morning astronomical twilight, day dives between sunrise and sunset, and crepuscular dives between astronomical twilight and sunrise or sunset. Dive parameters including bottom time (time between the first and last inflection points at > 80 % of the maximum depth), post-dive interval (PDI), descend and ascend rates (vertical speed descending or ascending), dive efficiency [bottom time/(dive duration + PDI)] and wiggles at the bottom of a dive (oscillations that may indicate feeding events) were used to determine colony differences in diving effort allocation during individual types of trips (see below). The simultaneous deployment of GPS and TDRs in murres was not accomplished in both islands until the third study year (2010). In 2008, birds were deployed mainly with TDR and few birds with GPS and TDR (n = 4). In 2009, birds at St. Paul were deployed only with GPS, and both GPS and TDR at St. George. To address possible effect of double tagging on diving behavior, we compared the dive depth of birds with only TDR and those with two tags (GPS & TDR) in Using a mixed model with colony as fixed factor and individual nested with daylight period as random factor, we found that mean dive depth did not differ between birds that had single and double tags (GPS and TDR= 22.0 ± 6.9 m, n = 4 birds; TDR= 26.2 ± 2.7 m, n = 26 birds; F 1, = 0.328, P = 0.569) in both colonies (F 1, = 0.151, P = 0.699) therefore we used all birds for further analysis of diving behavior. Dive locations of complete trips were calculated by linear interpolation of the latitude and longitude vectors in 2009 (St. George) and 2010 (both islands) and used to 1) estimate the diving effort allocation (number of dives, dive depth, dive duration, descend and ascend rates, bottom time, PDI, dive efficiency; and wiggles at the bottom) according to type of trip, and 2) frequencies of dives and dive depth according to oceanographic location (Middle Shelf: 100 m isobaths, Outer Shelf: 200 m, and Slope: > 1000 m). We included complete trips of St. George birds in 2009 (n = 6) to increase sample size because there were no year-differences in dive depth between 2009 and 2010 at St. George (see below). We used the percentage of occurrence of prey species consumed by murre adults and proportion and size (small, medium, large) of prey delivered to chicks as a proxy of food quality between colonies and years. Diets of tracked birds were linked to foraging habitat (Middle and Outer Shelf and Slope) of the last trips and reported only if GPS was still recording when logger was retrieved. Comparisons of chick diets were done broadly as fish and squid; ~ 49% (n = 1135) of total prey delivered to chicks were identified. 72

73 Data from observations of non-tagged birds such as chick feeding frequencies and trip duration, as well as measures of fledging success, were used to determine colony and year differences in breeding and foraging performance. CORT levels of adults were used to determine differences in nutritional stress of parents during the chickrearing period. Only nests that had chicks when sampled were used for the analysis. Because there were no colony differences in CORT levels between tagged birds (see instrumentation effect) all birds sampled were used for the analysis. We used three approaches to address possible effect of instrumentation on birds behavior, 1) compare nest attendance, chick-feeding frequency and trip duration between control and tagged birds in 2008 and ) compare CORT levels between tagged (GPS and TDR) and control birds in 2010 and 3) report frequencies of nest abandonment of tagged birds between colonies and years Statistical Analysis We used mixed linear models (MLMs) for the analysis of foraging range, total distance traveled, trip duration and dive depth, corticosterone levels, fledging success, and chick-feeding frequencies. We used colony and year as fixed factors in all the models and type of trip (day and overnight) in foraging models. Individual or nest was used as random factor for measures of trip duration and chick frequency. Nest was nested within plot to control for possible effect of nest location within each island. Because thick-billed murres at the Pribilof Islands differ in dive depth according to daylight periods (day, crepuscular and night hours; Harding et al. 2013), we used individuals nested within daylight periods as a random factor to reduce variance within individual means for dive depth models. Sex was excluded from both trip and dive models because we did not find significant differences of males and females on trip distance (F 1, = 0.409, P = 0.409), trip duration (F 1, = 1.296, P = 0.206) and dive depth (F 1, = 0.142, P = 0.555), and neither interactive effects with year and colony (P > 0.05). Tracking data only included 2009 and 2010 because few birds had complete trips in 2008 (St. George= 2). Dive data only included 2008 and 2010 because there were data collected at St. Paul in We also used similar MLMs for the analysis of foraging trips according to habitat used (Shelf = both colonies, and Slope =St. George), which was further split into Middle Shelf (100 m isobath), Outer Shelf (200 m), and Slope (> 1000 m) for analysis of dive depth in For the colony comparison in diving effort during individual trips (day and overnight) in 2010, trip nested within individual was used as random factor and dive depth as covariate in 73

74 dive parameter models. We excluded post-dive intervals longer than 60 sec (29%, n = 4504) to avoid intervals between consecutive bouts (Tremblay et al. 2003). To examine logger effect we used MLMs that had logger effect (tagged and control) as a fixed factor in addition to colony, year and individual or nest as a random factor. Adjustments for multiple comparisons were done for all tests using Bonferroni. Frequencies of type of trips, number of dives and prey size delivered to chicks were compared between islands and years using Chi-Square tests. Data was log-transformed if the residuals did not meet the assumptions for the general linear model (homogeneity and normality). Statistical analysis was carried out using PASW Statistics 18 and R. Means are expressed ± SE of the mean. All comparisons are two-tailed, and differences were considered significant when P < Results Foraging behavior Foraging location In the three study years, murres from St. Paul foraged mainly in the Shelf regardless of time of day whereas birds from St. George foraged both on the Shelf and Slope (Figure 2 and 3). Trips to the slope by murres from St. George were mainly overnight (Figure 3A). Of eleven birds that had multiple trips at St. George (n = 11), overnight trips were made to the slope (n = 5), to the shelf (n = 4), and to both the shelf and slope (n = 1) in addition to daytrips on the shelf. Foraging distance We found birds foraged farther and covered more horizontal space during overnight trips than daytime trips (P< 0.001) regardless of year or colony (P>0.05; Figure 3B). Trip distance was affected by both colony and type of trip (F 1, = , P < 0.001). Birds from St. Paul traveled longer distances than those from St. George during daytime trips (F 1, = 8.586, P = 0.006), and shorter than their counterparts during overnight trips (F 1, = , P < 0.001). These overnight differences were mainly due to trips to the slope by birds from St. George, which traveled comparably long distances both in 2009 (81.5 ± 8.5 m, n = 9) and 2010 (108 ± 11.4 m, n =5; F 1,12 = 3.462, P = 0.087, Figure 3A). Total distance traveled was also affected by both colony and time of day (F 1,54.0 = 7.04, P =0.010). Birds from St. Paul covered more horizontal space than those from St. George during daytime trips (F 1,23.0 = , P =0.029) but no differences were found overnight (F 1,531.0 = 4.07, P =0.052, Figure 3B). During overnight trips, birds from St. George that foraged in the Slope region traveled longer distances than those that foraged in the Shelf region (F 1,12 = 25.80, P < ). Trip duration increased with trip distance 74

75 during daytime trips (r = 0.447, P = 0.007; n = 35) but not during overnight trips (shelf: r = 0.079, P = 0.677; n = 27, slope: r = 0.326, P = 0.392; n = 6). Day trip duration was longer at St. Paul than at St. George and no inter-colony differences were found overnight (Table 2) Diving behavior We recorded a total of 32,488 dives (2008 = 14,311; 2009 = 4,329; 2010= 13,848) with a minimum and maximum depth of 2 and 133 m respectively. We found distinct colony-patterns in the frequencies of dives according to daylight periods in both 2008 (χ 2 1 = , P < , n = 14311) and 2010 (χ 2 1 = 608.6, P < , n = 13848; Figure 3). Birds from St. Paul dove more often during daylight (68%, n = 8653), less during crepuscular (27%) and seldom during nocturnal hours (5%). Birds from St. George dove as frequently during nocturnal hours (32%) and crepuscular hours (26%) as during daylight hours (42%, n = 19,506). Similar results were found using a sub-set of dives from 19 complete trips of tracked birds per island in 2010 (χ 2 1 = 366.1, P < , n = 6,093 dives). The frequency of St. George s nocturnal dives decreased by 50 and 35% in 2010 compared to 2008 and 2009, respectively; and no inter-annual differences were found for St. Paul birds (Figure 4A). Dive depth did not differ between colonies (F 1,215.8 = 1.082, P =0.299) or years (F 1,215.8 = 0.067, P =0.796) when controlled for daylight periods, although the interaction between colony and year was marginally significant (F 1,215.8 = 3.890, P =0.050). Murres from St. George made deep dives during daylight hours only in 2008 (Figure 4B). Across years, birds from both colonies dove on average significantly deeper during daylight (F 2,216.6 = , P < 0.001; 42.3 ± 1.4 m, n = 14,268), intermediate at crepuscular (19.8 ± 1.5 m, n = 7,271) and shallower at nocturnal hours (9.2 ± 1.8 m, n = 6,620; post-hoc tests: P < , Figure 4B). The analysis of diving effort during complete trips in 2010 showed that compared to St. George, birds from St. Paul dove deeper and longer during daytime trips; and also dove deeper during overnight trips but made significantly fewer dives (Model 1; Table 2; Figure 5). On average, St. Paul birds spent longer at the bottom of dives and had faster ascent rates but equivalent descent rates to those at St. George regardless of type of trip (Table 2). No colony differences were found in other dive parameters (P> 0.05; Table 2). Birds from St. George dove significantly more often during trips to the Slope than to the Shelf, and did not differ in dive depth (Model 2; Table 2). There were no differences in dive parameters between habitats except for faster descend and ascend rates during dives of trips to the Slope than to the Shelf 75

76 (Table 2). Further analysis of the dive locations of 2010 trips showed that on the Middle Shelf (100 m isobaths), birds from both islands dove significantly less at nocturnal hours (9%, n = 4203 dives) compared to daylight (61%) or crepuscular (30%; χ 2 1 = , P < ) hours. In deeper ocean domains (Outer Shelf =200 m, and Slope = > 1000 m), birds from St. George dove more often at nocturnal (46%, n = 5008) and crepuscular (34%) than daylight (20%; χ 2 1 = 441.1, P < ) hours (Figure 5). Dive depth of birds traveling overnight varied according to oceanographic domain (F 1, = , P < ); on average birds dove shallower in the Slope than in the Middle or Outer Shelf (post-hoc tests: P < ) with no differences between the Middle and Outer Shelf (P = 0.409). Birds from St. Paul dove to greater depths than those from St. George on the Middle Shelf during both daytime and overnight trips (Figure 6;Table 2). Diets Chicks Birds from St. Paul delivered predominately fish (St. Paul: %, n = 627 items) to their chicks; and birds from St. George delivered fish (67-77%, n = 508) and squid (29%) across years. Of the prey items that were identified to family and species level (n = 49%, n = 1,135), sandlance (Amnodytes hexapterus), pricklebacks (Stichaeidae spp.) and eelpouts (Lycodes spp.) were the most frequent prey at St. Paul; and squid (Gonatus spp., Gonatopsis borealis; Figure 6A) and pricklebacks at St. George. Other fish species such as age-0 pollock, flatfish (Pleuronectidae), myctophids (Myctophidae), capelin (Mallotus villosus), and eulachon (Thaleichthys pacificus) were much less frequent (Figure 7a). There were differences in the frequencies of prey size delivered to chicks among years on both St. Paul (χ 2 4 = 142.0, P < ) and St. George (χ 2 4 = 26.1, P < ) Islands. Chicks from St. Paul were fed more often with larger prey in 2009 than in other years; no annual differences in larger prey were found at St. George (Figure 7B). In 2008, chicks were fed more often with small (St. George) or medium (St. Paul) prey than 2009 and 2010 (Figure 7B). Adults Birds from St. Paul fed on a variety of fish species, whereas birds from St. George fed primarily on squid during the three study years (Figure 8). Birds from both colonies fed on age-0 pollock although only at a high frequency at St Paul in Across years, less than 3% of samples had age-1 pollock. Analysis of the diets of tracked birds indicated that squid were obtained both during trips to the Slope (n = 5) and to the Middle Shelf (n = 3). Eelpouts and euphasiids were obtained during trips to the Middle Shelf. Juvenile pollock were obtained on the Middle shelf by St. Paul birds (n = 17) and on the Outer shelf by St. George birds (n = 3). 76

77 Parental Time Budgets We did not find differences in chick feeding rates (log-transformed: F 1, 231 = 0.744, P = 0.389) of nontagged birds between colonies; but did between years (P > 0.009; Table 3). Pairs fed chicks more often in 2009 than in 2008 (post hoc tests: P < ) and 2010 (P = 0.001). The trip duration of non-tagged birds varied between colonies according to year (F 1, = 5.784, P = 0.003) and type of trip (F 1, = 7.518, P = 0.001). Birds from St. Paul had longer daytime trips than those from St. George (log-transformed data: F 1, = 7.58, P = 0.007) regardless of year (F 2, = 2.842, P = 0.061); and no colony differences were found overnight (P > 0.05; Table 3). Birds from both colonies had shorter overnight trips in 2009 than in 2008 (post-hoc tests: P = 0.010) and 2010 (P = 0.001; Table 3) Fledging success Fledging success did not differ between colonies (F 1, = 0.817, P = 0.371; Table 3) but did between years (F 2, = 8.976, P < ). Birds from both colonies had higher fledging success in 2009 and 2010 than in 2008 (post-hoc test: P < 0.008) with no difference between 2009 and 2010 (P = 0.602). Corticosterone levels We found an interaction between colony and year on corticosterone levels (F 2, 433 = , P = ); birds from St. Paul had significantly higher CORT levels than those from St. George in 2008 (post-hoc test: P < ) and 2009 (P = 0.024) but not in 2010 (P = 1.00). Within colonies, there were no interannual differences in CORT levels at St. George (P > 0.05); and only between and (post-hoc tests: P < ) at St. Paul (Figure 9) Instrument effect The majority of birds (85%, n = 189) with data loggers had chicks alive at the end of the deployment period (Table1). Most birds that lost a chick were from St. George Island (78%, n = 28). Most of the instruments deployed were retrieved (73%, n = 189), and more often from birds with TDRs (88%, n = 34) than from those with GPS and TDRs (68%, n = 120). Nest attendance differed between tagged and control pairs (F 1, 237 = , P < ) regardless of colony or year (P > 0.05). Pairs that had a tagged 77

78 bird spent less time at the nest site (61.7 ± 0.52 mins/hr -1, n = 23) than those in which both members were untagged (63 ± 0.35 mins/hr -1, n = 94) across years. There was an interaction effect between logger effect and year on chick feeding rates (F 1, 138 = 6.166, P = 0.014). In 2009, chicks from pairs with a tagged member (0.17 ± 0.04 feeds/h, n = 18) were fed less often than those from control pairs (0.28 ± 0.02 feeds/h, n = 67; F 1, 78 = 7.703, P = 0.007); however, no differences were found in 2008 (control: 0.15 ± 0.02 feeds/h, n = 46; tagged: 0.18 ± 0.04 feeds/h, n = 18; F 1, 60 = 0.657, P = 0.421). Overall trip duration did not differ between tagged and control birds (log-transformed: F 1, 526 = 0.192, P = 0.661) regardless of colony or year (P > 0.05). However, tagged birds (18.3 ± 1.0, n = 10) of both colonies had longer trips overnight than control birds (13.2 ± 0.33, n = 65; F 1, 67 = 14.07, P < ) with no differences during daytime trips (control: 2.2 ± 0.14, n = 416; tagged: 1.7 ± 0.71, n = 51; F 1, 459 = 0.017, P =0.896). Corticosterone levels were higher in tagged birds (St. Paul: 0.76 ± 0.04 ng/ml, n = 38; St. George: 0.86 ± 0.05 ng/ml, n = 28) than control birds (St. Paul: 0.72 ± 0.04 ng/ml, n= 61; St. George: 0.69 ± 0.03 ng/ml, n = 68; log-transformed: F 1, 191 = P = 0.023) with no differences between colonies (F 1, 191 = 0.055, P = 0.814) Discussion In this study we tested the hypothesis that oceanographic location of St. Paul (declining colony) farther from alternative oceanic habitats compared to St. George (stable colony) would restrict adults food supply resulting in increased foraging effort and nutritional stress during food shortages on the Continental Shelf. We found that birds from St. Paul foraged only on the Middle Shelf, and dove deeper and longer during extended daytime trips compared to their counterparts on St. George. Although there were no inter-colony differences in total distances traveled overnight, birds from St. George fed in the Outer Shelf and Slope regions while St. Paul birds foraged in the Middle Shelf. St. George birds spent more time foraging at night than St. Paul birds; suggesting that murres breeding on St. George have higher flexibility in their choice of foraging habitat compared to those breeding on St. Paul. At both colonies chicks were fed at similar rates; and fledging success also did not differ between the colonies. However, the increased foraging effort and perhaps lower quality prey of birds from St. Paul mirrored higher levels of nutritional stress incurred by breeders on St. Paul compared to St. George in two of the three study years. 78

79 Colony differences in the spatial and temporal use of marine habitats As expected by colony location, we found a consistent pattern in the spatial and temporal use of marine habitats, and also striking differences in diving activity relative to daylight periods across years. Birds from St. Paul foraged exclusively on the Middle Shelf both during day and overnight trips (leaving the colony late afternoon or evening and returning early the next morning). Like St. Paul, birds from St. George foraged on the Middle Shelf (day and overnight) but also used the Outer Shelf (200 m) and Slope- Pribilof Canyon (> 1000 m) overnight. We also found that birds from St. George actively foraged during nocturnal hours (32%; mean depth = 9 m) while birds from St. Paul rarely foraged at night (5%), concentrating feeding during daylight (47%; mean depth = 42 m). Diving activity during crepuscular hours (mean depth = 20 m) was comparable between the colonies. The observed colony differences in diurnal/nocturnal diving activity could possibly reflect different foraging strategies, and imply that birds from St. Paul may have obtained enough food during daytime to meet their needs. Nevertheless, the different levels of nutritional stress observed between the two colonies suggest (Kitaysky et al. 2007, 2010) that St. Paul birds were more food-limited compared to St. George birds. Given that birds from both colonies had rarely foraged during overnight trips in the Middle Shelf regions suggest that prey may not be not be as accessible there as it was in the Outer Shelf and Slope regions where foraging activity of murres peaked at darkness. Recent evidence indicates that foraging activity of predators at nocturnal hours not only coincide with the DVM of their prey, but also correlate positively with the bathymetry of the marine habitats (Dias et al. 2012). Basking sharks (Cetorhinus maximus) follow the DVM of their prey, which is reverse (descend at dusk and ascend at dawn) in shallow inshore shelf habitats, and normal (ascend at dusk and descend at dawn) in basin habitats (Sims et al. 2005). In the Bering Sea Middle Shelf, juvenile pollock vertically migrate to upper 20 m to feed on small prey (Swartzman et al. 1999; Benoit-Bird et al. 2011), however, feeding cessation appears to occur approx. at 0100 (Brodeur et al. 2000) suggesting that DVM may stop at darkness. Squid are also known to follow the DVM of zooplankton; and although very little information exists on neritic squid behavior, oceanic squid species (Gonatidae) that vertically migrate to the upper 0-7 m layer are found abundant throughout the night (Ropert and Young; 1975 Okutani et al. 1988). Additionally, differences in density of specific prey between habitats may also influence the nocturnal feeding of thick-billed murres. Hunting through random encounters at darkness could be a viable explanation of how murres can feed at darkness if prey were available in sufficiently high densities (Regular et al. 2010). Higher densities of euphasiids and 79

80 squid were found on the Outer Shelf and Slope-Pribilof Canyon zones respectively than in the Middle Shelf during a concurrent study (Benoit-Bird et al. 2011) Colony differences in foraging effort affect chick, adults or both? Although birds from both colonies foraged in the Middle Shelf during daytime, they differed in their spatial and vertical effort allocation; and did not during overnight trips despite differences in habitat use. Flying is one of the most costly activities for murres, followed by diving (Gaston et al. 1985; Pennycuick 1987; Elliott et al. 2013). Analysis of individual trips showed that birds from St. Paul had wider foraging ranges and longer distances traveled during daytime compared to St. George in all years. Overnight, the total distance traveled did not differ between the colonies probably because St. George birds made both long trips to the Slope and short trips to the Shelf. Likewise, in 2010 birds from St. Paul dove as often but at greater depths during daytime trips; however dive effort overnight appear to be balanced between colonies with St. Paul making fewer but deeper dives than St. George. We did not find differences in other dive parameters, except for St. George s faster ascent and descent rates, probably because overnight trips include some daylight periods with deep dives possibly masking colony differences in nocturnal/crepuscular diving. Thus, it seems that birds from St. Paul worked harder during daylight than their counterparts at St. George to meet both chick and adult demands probably because had reduced time to feed at nocturnal hours when prey is apparently more abundant shallower depth Our results support the proposed idea that parental effort for the young in murres is fixed (Kitaysky et al. 2000) and breeding performance not variable (Piatt et al. 2007); however it appears to come at different cost for the parents when food is limited. We found that despite differences in foraging effort, birds from both colonies fed chicks at similar rates, with equivalent quality and size of prey (pelagic and benthic fish and squid) resulting in similar fledging rates. In concordance with results in foraging effort, birds from St. Paul had higher levels of nutritional stress than those from St. George suggesting that adults may have had different quality or quantity of prey and/or work harder to self-feed. Adults and chicks overlapped in diets as has been found previously at the Pribilof Islands (Motohiro et al. 2010). This similarity in prey type is expected as only the last item of all prey captured during each trip (day and overnight) is allocated for chick feeding (Piatt et al. 2007). In addition, adults from both islands fed on euphausiids and amphipods, which were apparently underestimated due to timing of sampling (starting late morning) and 80

81 faster digestion rates than other prey (Neves et al. 2006). Actually, stomach contents of birds collected at sea near the Pribilofs Islands in 2008 and 2009 had 50-60% more crustaceans than in this study (Jones et al. unpubl. data). Diets of tracked murres and observations of meal deliveries showed that squid were obtained both on the Bering Sea slope and Middle Shelf mainly by birds from St. George. Analysis of the energy content of these prey items indicates that squid from the Bering Sea slope and euphausiids from the Outer Shelf had significantly higher energy content than those collected on the Middle Shelf (Whitman 2010). Therefore, birds from St. George appear to have access to higher and more abundant prey during overnight trips for self-feeding compared to those from St. Paul that appear to work harder and fed in lower quality prey to restore adult reserves on the Middle Shelf. Clearly, the colony differences in foraging and diving behavior found in this study may vary in warm years when food conditions are better on the Middle Shelf (Benowitz-Fredericks et al. 2008). For instance baseline CORT levels of thick-billed murres at both colonies was ~ 2 times lower in 2005 (warm year; Stabeno et al. 2012) than in 2009 (cold year; Kitaysky et al. unpubl. data) indicating better foraging conditions. Preliminary analysis of diving behavior in warm years indicates little nocturnal diving activity of birds from St. George (Kitaysky et al. unpubl. data). Given the high-energy flying costs of murres (Pennycuick 1986, Elliott et al. in Press) and the contrasting diving activity between habitats bathymetry, we would expect birds from St. George to forage mainly on the Middle Shelf in warm years or when food conditions are better. Elevated baseline CORT levels in seabirds are correlated with reduced prey availability, a greater probability of skipping breeding in subsequent years, and lowered survival rates (Kitaysky et al. 2007; 2010; Goutte et al. 2010). Preliminary analysis of adult survival ( ) of thick-billed murres at both colonies suggests that birds at St. Paul may have lower survival rates than at St. George although sample sizes are currently small (Renner et. al. unpubl. data). Therefore, higher stress levels in birds from St. Paul due to chronic reductions in food availability on the Middle Shelf may be a plausible mechanism for explaining the declining population of thick-billed murres at this colony Inter-annual differences and prey availability Our measures of chick-feeding rates, fledging success, and prey size delivered to chicks indicate that food availability was relatively better in 2009, intermediate in 2010 and worse in 2008 for both colonies. In 2009, chicks were fed more often, prey was larger and more chicks were fledged compared to The 81

82 results of a detailed study of prey abundance and distribution available for the study area in 2008 and 2009 (Benoit-Bird et al. 2011) are in parallel with changes in occurrence of age-0 pollock and squid in murre diets. Age-0 pollock did not differ in total biomass density between years but few patches were closer to St. Paul than St. George in In the same year, the numbers of squid captured at the Pribilof Canyon area were higher than in If seabird diets mirror changes in prey availability in marine habitats, then, the increased presence of amphipods in St. Paul s adult diets in 2010 may also reflect an increase of this prey on the Middle Shelf. Furthermore, one of the main crustacean preys consumed by murres is the amphipod Themisto libellula, an Arctic species, which is relatively high in energy content (Whitman et al. 2010). T. libellula has increased in abundance from 2008 to 2010 in the Southeastern Bering Sea Continental Shelf (Hunt et al. 2011; Pinchuk et al. unpubl. data). This increase in amphipod abundance might have boosted St. Paul adults reserves in 2010, which could partially explain their lower nutritional stress found in this year compared to 2008 and It is puzzling, however, why levels of nutritional stress in St. Paul murres were higher in 2009 if patches of age-0 pollock were apparently available. One possible explanation is that feeding on age-0 pollock schools may not have been as efficient as feeding on swarms of amphipods, which are higher in energy content than age-0 pollock (Whitman 2010), and apparently distributed at shallower depths at dusk (0 m vs. 20 m; Hiroki 1988; Benoit-Bird et. al 2011). Finally, although no statistical differences were found in dive depth between years when daylight period was controlled for, birds from St. George made deep dives during daytime in 2008 when food conditions were apparently poorer Instrument effect Our study is valuable in that it provides the first novel three-dimensional data on thick-billed murres foraging behavior but not without instrument effects. Undoubtedly, the poor foraging conditions of the study period probably aggravated effects of instrumentation but equally in both colonies. Although most instrumented birds had chicks when tags were retrieved, they had lower chick-feeding frequencies, shorter nest attendance, and higher stress levels than control birds. We also found that trip duration did not differ between GPS and control birds; and that, although with small sample sizes, dive depth did not differ between birds with double tags (GPS & TDR) and single tags (TDR). The leg-mounted TDRs used in this study had no effect on chick-feeding frequencies, which concurs with other studies (Elliot et al. 2007). Altogether these results indicate that birds deployed with GPS or GPS and TDRs worked harder to raise chicks than non-instrumented birds, their foraging patterns were probably less affected. We believe 82

83 that the back attachment, needed to acquire satellite locations, along with the increased drag and possible slight positive buoyancy of the GPS rather than the added weight of two tags (GPS and TDR; less than 3% of the an adult s weight) could explain the birds increased foraging effort. The hand-made GPScasing with shrink tubing at every deployment might have increased the chances of positive buoyancy if air bubbles were left inside the casing. We therefore recommend the use of pre-made aerodynamic small GPS, water and buoyancy proofed. We also recommend researchers make shorter deployments (< 24 h) to minimize logger effect and increase recovery rates using a protocol that minimizes handling time and chick disturbance (ours is available upon request) In conclusion, our study supports the hypothesis that during cold years thick-billed murres breeding on St. Paul (declining population) are both spatially and temporally restricted to feed in the food-poor Shelf regions. In contrast, St. George murres (stable population) are able to access abundant high-lipid prey resources at nocturnal hours in the Outer Shelf and Slope regions. As a result, St Paul murres worked harder than their counterparts at St. George to meet their energy demands. The increase in foraging effort of birds from St. Paul came with the costs of elevated adult nutritional stress. Given that nutritional stress appears to be a good predictor of adult survival (Kitaysky et al. 2010; Satterthwaite et al. 2012), we propose that chronic reduction in food availability on the Middle Shelf may in part explain the divergent population trends of thick-billed murres at the Pribilof Islands Acknowledgments This collaborative study was part of the Bering Sea Integrated Ecosystem Research Program funded by the North Pacific Research Board. We are grateful for the enthusiastic assistance and excellent work of crew members on St George: Brie Drummond, Dean Kildaw, Rob Massangale, Nathan Banfield, Caroline Poli, Vijay Patil, Rolanda Steenweg, Donald Lyons, Ram Papish, Chris Barger and Rob Marshall and on St Paul: John Warzybok, Ine Dorreteijn, Dan Cushing, Kerrith McKay, Ana Santos, Tom Harten, and Alexis Will. We thank Zhenya Kitaiskaia for conducting hormonal assays and Kathy Turco for expert diet analyses. We thank Karin Holser (St. George Island Institute), Sally and Chris Merculief (Traditional Tribal Council of St. George Island), Phil Zavadil and Debbie Lestenkof (Aleut Community of St. Paul Island) and Priscilla Wohl and Arina Purcella (Northern Forum) for logistical and financial assistance. Thanks to Karen Brenneman and Michelle St. Peters (USFWS-Anchorage) for invaluable expeditor 83

84 assistance. Bruce Robson, Steve Insley, Scott Shaffer and Carlos Zavalaga provided technical and academic assistance at early stages of telemetry project. This manuscript benefit from discussion with Kyle Elliot and Steve Insley. This study was funded by NPRB BSIERP project B63 and B77 to David Irons and Dan Roby, project B65 to Heather Renner, and project B77 to Alexander Kitaysky. This is contribution No. XXX of the North Pacific Research Board References 84

85 Anderson PJ, and Piatt JF (1999) Community reorganization in the Gulf of Alaska following ocean climate regime shift. Mar Ecol Progr Ser 189: Benoit-Bird, KJ, Kathy K, Heppell S, Jones N, and Hoover B (2011) Active acoustic examination of the diving behavior of murres foraging on patchy prey. Mar Ecol Progr Ser 443: Benowitz-Fredericks, ZM, Shultz MT, Kitaysky, AS (2008) Stress hormones suggest opposite trends of food availability for planktivorous and piscivorous seabirds in two years. Deep-Sea Res. II 55: Brodeur RD, Wilson MT, and Ciannelli L (2000) Spatial and temporal variability in feeding and condition of age-0 walleye pollock (Theragra chalcogramma) in frontal regions of the Bering Sea. ICES Journal of Marine Science 57: Brown CR, Brown MB, Raouf SA, Smith LC, Wingfield JC (2005) Effects of endogenous steroid hormone levels on annual survival in cliff swallows. Ecology 86: Burger AE, Piatt JF (1990) Flexible time budgets in breeding common murres: buffers against variable prey abundance. Stud Avian Biol 14: Byrd GV, Schmutz JA, Renner HM (2008). Contrasting population trends of piscivorous seabirds in the Pribilof Islands: A 30-year perspective. Deep-Sea Res. II 66: Chaurand T, Weimerskirch H (1994) Regular alternation of short and long foraging trips in the blue petrel Halobaena caerulea: a previously undescribed strategy of food provisioning in a pelagic seabird. J Anim Ecol 63: Conners ME, Hollowed AB, Brown E (2002) Retrospective analysis of Bering Sea bottom trawl surveys: regime shift and ecosystem reorganization. Prog Oceanogr 55: Coulson JC (2002) Why do adult kittiwakes survive so long but breed so poorly in the Pacific? J Avian Biol 33: Coyle, KO, Eisner LB, Mueter FJ, Pinchuk AI, Janout MA, Cieciel KD, Farley EV, and Andrews AG (2011) Climate change in the southeastern Bering Sea: impacts on pollock stocks and implications for the oscillating control hypothesis. Fish Oceanogr 20: Elliott KH, Davoren GK, Gaston AJ (2007) The influence of buoyancy and drag on the dive behaviour of 85

86 an arctic seabird, the thick-billed murre. Can J Zool 85: Elliott KH, Gaston AJ, Crump D (2010) Sex-specific behaviour by a monomorphic seabird represents risk partitioning. Behav Ecol 21: Elliott KH, LeVaillant ML, Kato A, Speakman JR, Ropert-Coudert Y (2013) Accelerometry predicts daily energy expenditure in a bird with high activity levels. Biol. Lett. 9 (1): Gaston, A.J., Energy invested in reproduction by thick-billed murres (Uria lomvia). Auk 102, Golet GH, Irons DB, Estes JA (1998) Survival costs of chick rearing in black-legged kittiwakes. J Anim Ecol 67: Goutte A, Angelier F, Welcker J, Moe B and others (2010) Long-term survival effect of corticosterone manipulation in black-legged kittiwakes. Gen Comp Endocrinol 167: Harding AMA, Piatt JF, Schmutz JA, Kettle AB, Shultz M, Van Pelt TI, Kettle AB, Speckman SG (2007) Prey density and the behavioural flexibility of a marine predator: the common murre (Uria aalge). Ecology 88: Hollowed, A.B., Barbeaux, S.J., Cokelet, E.D., Farley, A., Kotwicki, S., Ressler, P.H., Spital, C., and Wilson, C.D. (2012) Effects of climate variations on pelagic ocean habitats and their role in structuring forage fish distributions in the Bering Sea. Deep-Sea Res. II 65-70: Hanssen SA, Hasselquist D, Folstad I, Erikstad KE, (2005) Cost of reproduction in long-lived bird: incubation effort reduces immune function and future reproduction. Proc R Soc B 272: Hare SR, Mantua NJ (2000). Empirical evidence for North Pacific regime shifts in 1977 and Prog Oceanogr 47: Harding AMA, Paredes R, Suryan RM, Roby, DD, Irons, DB, Orben RA, Renner H, Young R, Barger R, Dorresteijn I, Kitaysky, A. Does location matter? An inter-colony comparison of thick-billed murres (Uria lomvia) breeding at varying distances from productive oceanographic features in the Bering Sea. Deep Sea Research II: In press. Harris MP, Birkhead TR (1985) Breeding ecology of North Atlantic auks. The Atlantic Alcidae (eds D.N. Nettleship and T.R. Birkhead), pp Academic Press, London 86

87 Hiroki M (1988) Relation between diel vertical migration and locomotor activity of a marine hyperiidean amphipod, Themisto japonica (Bovallius) J Crustacean Biol 1: Hunt, JL Jr, Coyle KO, Eisner LB, Farley AV, Heintz RA, Mueter F, Napp JM, Overland JE, Ressler PH, Salo S, Stabeno, PJ (2011) Climate impacts on eastern Bering Sea foodwebs: a synthesis of new data and an assessment of the Oscillating Control Hypothesis. ICES J Mar Sci 68: Kitaysky AS, Hunt GL, Flint EN, Rubega MA, Decker MB (2000) Resource allocation in breeding seabirds: responses to fluctuations in their food supply. Mar Ecol Prog Ser 283: Kitaysky AS, Kitaiskaia EV, Piatt JF Wingfield JC (2003) Benefits and costs of increased levels of corticosterone in seabird chicks. Horm Behav 43: Kitaysky,AS, Piatt JF, Wingfield JC (2007) Stress hormones link food availability and population processes in seabirds. Mar Ecol Prog Ser 352: Kitaysky AS, Piatt JF, Hatch SA, Kitaiskaia EV, Benowitz-Fredericks ZM, Shultz MT, Wingfield JC (2010) Food availability and population processes: severity of nutritional stress during reproduction predicts survival of long-lived seabirds. Funct Ecol 24: Lewis S, Grémillet D, Daunt F, Ryan PG, Crawford RJM, Wanless S (2006) Using behavioural and state variables to identify proximate causes of population change in a seabird. Popul Ecol 147: Litzow, MA, Bailey KM, Prahl FG, Heintz R (2006) Climate regime shifts and reorganization of fish communities: the essential fatty acid limitation hypothesis. Mar Ecol Progr Ser 315: 1-11 Monaghan P, Uttley JD, Burns MD, Thaine C, and Blackwood J (1989) The relationship between food supply, reproductive effort and breeding success in Arctic Terns. Sterna Paradisaea. J Anim Ecol 58: Motohiro I, Takahashi A, Kokubun N, Kitaysky AS and Watanuki Y Foraging behavior of incubating and chick-rearing thick-billed murres (Uria lomvia) Mar Ecol Prog Ser 8: Mueter FJ, Bond NA, Ianelli JN, Hollowed AB (2011) Expected declines in recruitment of walleye pollock (Theragra chalcogramma) in the eastern Bering Sea under future climate change. ICES J Mar Sci 68: Neves, VC, Bolton M, Monteiro LR (2006) Validation of the water offloading technique for diet 87

88 assessment: an experimental study with Cory s shearwaters (Calonectris diomedea). J. Ornithol. 147, Okutani T, Kubodera T, Jefferts K (1988) Diversity, distribution and ecology of gonatid squids in the subarctic Pacific : a review. Bull. Ocean Res. Inst. Univ. Tokyo 26: Orians GH, Pearson NE (1979) On the theory of central place foraging. Pages in D.J. Horn, R.D. Mitchell, and G.R. Stairs, editors. Analysis of ecological systems. Ohio State University Press, Columbus, Ohio, USA Paredes R, Jones IL, Boness DJ (2005) Reduced parental care, compensatory behaviour and reproductive costs experienced by female and male thick-billed murres (Uria lomvia) equipped with data loggers. Anim Behav 69: Paredes R, Jones IL, Boness, DJ, Tremblay Y, Renner M (2008). Sex-specific differences in diving behaviour of two sympatric Alcid species: thick-billed murres and razorbills. Can J Zool 86: , doi: /z Pennycuick CJ (1987) Flight of auks (alcidae) and other northern seabirds compared with southern procellariformes: Ornithodolite observations. J Exper Biol 128: Piatt JF, Harding AMA, Shultz M, Speckman SG, Van Pelt TI, Drew GS, Kettle AB (2007) Seabirds as indicators of marine food supplies: Cairns revisited. Mar Ecol Prog Ser 352, Regular PM, Hedd A, Montevecchi WA (2011) Fishing in the Dark: A Pursuit-Diving Seabird Modifies Foraging Behaviour in Response to Nocturnal Light Levels. PLoS ONE 6(10): e doi: /journal.pone Reid WV (1987) The cost of reproduction in the Glaucous-winged Gull. Oecologia 74: Romero LM, Wikelski M (2001) Corticosterone levels predict survival probabilities of Galapagos marine aguanas during El Niño events. P Nat Acad Sci Biol 98: Roper, C FE, Young RE (1975) Verlical distribution of pelagic cephalopods, Smith- son. Contrib. Zool. 209, 51 p. Saino N, Suffritti C, Martinelli R, Rubolini D, and Møller AP (2003) Immune response covaries with corticosterone plasma levels under experimentally stressful conditions in nestling barn swallows (Hirundo 88

89 rustica) Behav Ecol 14: Saraux C, Robinson-Laverick SM, Le Maho Y, Ropert-Coudert Y, Charadia A (2011) Plasticity in foraging strategies of inshore birds: how Little penguins maintain body reserves while feeding offspring. Ecology 92: 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: Satterthwaite WH, Kitaysky AS, and Mangel M (2012). Linking climate variability, productivity and stress to demography of a long-lived seabird. Mar Ecol Prog Ser 454: Shealer DA (2002) Foraging behavior and food of seabirds. In: Schreiber EA, Burger J, editors. Biology of marine birds. Boca Raton, Florida: CRC Press. pp Sims DW, Southall EJ, Tarling GA, Metcalfe JD (2005) Habitat-specific normal and reverse diel vertical migration in the plankton-feeding basking shark. J. Anim. Ecol. 74, 2005b doi: /j x Sinclair EH, Vlietstra LS, Johnson DS, Zeppelin TK, Byrd GV, Springer AM, Ream RR, Hunt GL Jr (2008). Patterns in prey use among furseals and seabirds in the Pribilof Islands. Deep-Sea Res. II 55: Springer AM, McRoy, CP, Flint MV (1996) The Bering Sea Green Belt: shelf-edge processes and ecosystem production. Fish Oceanogr 5, Stabeno, P.J., Kachel, N.B., Moore, S.E., Napp, J.M., Sigler, M., Yamaguchi, A., and Zerbini, A.N. (2012) Comparison of warm and cold years on the southeastern Bering Sea shelf and some implications for the ecosystem. Deep-Sea Res. II 65-70: Swartzman G, Brodeur R, Napp J, Hunt G, Demer D, Hewitt R (1999) Spatial proximity of age-0 walleye pollock (Theragra chalcogramma) to zooplankton near the Pribilof Islands, Bering Sea, Alaska. ICES Journal of Marine Science 56: Takahashi A, Matsumoto K, Hunt GL Jr, Shultz M.T, Kitaysky AS, Sato K, Lisa K, Watanuki Y (2008) Thick-billed murres use different diving behaviours in mixed and stratified waters. Deep-Sea Res. II 55:

90 Van Pelt TI, Piatt JF, Lance BK, Roby DD (1997) Proximate composition and energy density of some North Pacific forage fishes. Comp Biochem Physiol A 118: Weimerskirch, H., Are seabirds foraging for unpredictable resources? Deep Sea Res. Part II 54: Weimerskirch H, Zimmermann L, Prince PA (2001) Influence of environmental variability on breeding effort in a long-lived seabird, the yellow-nosed albatross. Behav Ecol 12 :22-30 Welcker J, Varpe O, Steen H (2012) Mass fluctuations suggest different functions of bimodal foraging trips in a central place forager. Behav. Ecol. 23: Whitman LD (2010) Variation in the density of forage fishes and invertebrates from the Southeastern Bering Sea. Msc. Thesis. University of Oregon. USA Zador SG, Piatt JF (1999) Time-budgets of common murres at a declining and increasing colony in Alaska. Condor 101: Zavalaga CB, Dell'Omo G, Becciu P, Yoda K (2011) Patterns of GPS Tracks Suggest Nocturnal Foraging by Incubating Peruvian Pelicans (Pelecanus thagus). PLoS ONE 6(5): e doi: /journal.pone

91 Table 1. Summary of data loggers deployed in Thick-billed murres at the Pribilof Islands during 2008, 2009 and St. Paul St. George Bird- tags recovered/deployed TDR 12/13 0/0 0/0 18/21 0/0 0/0 GPS 0/0 19/26 3/3 4/5 0/2 0/0 GPS & TDR 2/5 0/2 32/40 3/6 15/26 29/41 Total 15/17 19/28 35/43 25/32 15/28 29/41 Bird- tags with nest failed Bird- tags with data analyzed Diving Location Diving location # Trips w/distance attained # Trips complete # Trips w/ dive data # Dives

92 Table 2. Diving effort of thick- billed murres nesting at St. Paul and St. George Islands according to type of trip and foraging habitat in Mean ± SE. ST. PAUL ST. GEORGE P- value Foraging habitat Shelf Shelf Shelf Slope Total Model 1 Model 2 Type of trip Day Overnight Day Overnight Type Colony Type* Colony Shelf vs. Slope # Dives per trip 27 ± ± ± ± ± 39* 195 ± 29* Dive depth (m) 44± 4.7* 31 ± 5.9* 22 ± ± ± ± Dive Duration (s) 124 ± 9.7* 105 ± 12.3* 72 ± ± ± ± Bottom time (s) 50 ± 4.1* 49 ± 5.0* 32 ± ± ± ± Descend rate a (m/s) 0.93 ± ± ± ± ± 0.11* 0.95 ± Ascend rate a (m/s) 1.1± 0.07* 0.91 ± 0.09* 0.75 ± ± ± 0.11* 0.88 ± Post- dive interval (s) 32 ± ± ± ± ± ± Dive efficiency 0.33± ± ± ± ± ± # Wiggles bottom 6.7± ± ± ± ± ± Total trips Mixed linear models with colony and time of day (model 1) and foraging habitat overnight (model 2) as fixed factors and individual nested within trip as random factor. ( * ) Significant higher means a Log- transformed for statistical analysis.

93 Table 3. Foraging and breeding performance of non-tagged thick-billed murres from St. Paul and St. George islands in Values are mean ± SE, with sample size in parentheses. (*) indicate significance between islands Trip Duration (h) Day 3.2 ± 0.4* (55) Overnight 13.7 ± 0.8 (13) Chick feeding rate (feeds h - 1 ) 0.14 ± 0.03 (19) Fledging success 0.76± 0.03 (chicks fledged per nest (155) w/chicks) island year island x year St. Paul St. George St. Paul St. George St. Paul St. George p p p 2.4 ± 0.3 (161) 14.6 ± 0.6 (19) 0.16 ± 0.03 (27) 0.77± 0.03 (201) 1.5 ± 0.4 (88) 11.9 ± 0.6 (22) 0.29 ± 0.02 (40) 0.84± 0.03 (217) 2.3 ± 0.4 (112) 12.4 ± 0.8 (11) 0.27 ± 0.03 (27) 0.85 ± 0.02 (217) 3.5 ± 0.3* (151) 15.7 ± 0.6 (25) 0.19 ± 0.05 (12) 0.91± 0.02 (229) 2.1 ± 0.3 (90) 14.2 ± 0.9 (38) 0.13 ± 0.04 (21) 0.85 ± 0.03 (164) <08& >08& &10>

94 Figure 1. Geographic and bathymetric location of the study colonies, Pribilof Islands (St. Paul and St. George) in the southeastern Bering Sea Continental Shelf, Alaska

95 2385 a) 2386 Middle shelf Outer shelf 2390 Slope ( n =2) 2009 (n =41) 2010 (n = 64) Middle shelf St. George b) Outer shelf Slope 10 km 20 km St. Paul Middle shelf

96 Figure 2. GPS tracks of thick- billed murres at the Pribilofs Islands during 2008, 2009 and 2010 (a). Examples of daytime trips (yellow) and overnight trips (pink and purple). Vertical lines indicate dive locations within tracks

97 Trip Frequency a) % Total # trips St. Paul (n = 47) St. George ( n = 30) Middle Shelf Middle Shelf Outer shelf Slope 2420 Day Overnight 2421 b) 2422 Max. Distance (km) St. Paul St. George * Day Trip Distance * Overnight Figure 3. Foraging trip frequency (a) and maximum distance (b) of thick-billed murres nesting at the Pribilof Islands according to foraging habitat (Shelf and slope) and type of trip (day and overnight) during 2009 and (*) Denotes significance between colonies (P < 0.023). Number of trips inside bars. 97

98 " 80" Nocturnal" Day" Crepuscular" a) %"total"dives" 60" 40" 20" " St."Paul" St."George" St."Paul" St."George" St."Paul" St."George" 2008" 2009" 2010" Nocturnal" Day" Crepuscular" b) 2429 Maximum"depth"(m)" 60" 50" 40" 30" 20" 10" 0" St."Paul" St."George" St."Paul" St."George" St."Paul" St."George" 2008" 2009" 2010" Figure 4. Inter-annual differences in the frequency (A) and dive depth (B) of thick-billed murres nesting at St. Paul and St. George Islands according to daylight periods (based in sunset and sunrise) during 2008, 2009 and

99 Dive"depth"(m)" 0" 20" 40" 60" 80" 100" 3 km 17:30" 17:39" 17:47" 17:56" 18:04" 18:13" 18:22" 18:30" 18:38" 18:47" 18:55" 19:03" 19:10" 19:19" 19:28" 19:37" 19:44" 19:52" 20:02" Dive"depth"(m)" 0" 20" 40" 60" 80" 100" St."Paul"Shelf?Day" 23 km 19:44" 19:45" 19:47" 20:12" 20:22" 20:27" 20:32" 20:36" 20:41" 20:46" 20:47" 20:48" 20:50" 20:55" 20:56" 20:57" 20:59" 21:00" 15 km St."Paul"Shelf>Night" 34 km 0" 18:05" 18:42" 19:13" 19:42" 20:17" 20:49" 21:15" 21:47" 23:57" 0:35" 1:33" 2:17" 2:46" 3:08" 3:30" 4:19" 4:55" 5:24" 5:51" 6:34" 0" 18:42" 20:02" 0:40" 1:06" 1:30" 1:55" 2:21" 2:46" 3:11" 3:36" 4:01" 4:26" 4:50" 5:15" 5:40" 6:05" 6:30" 7:45" Dive"depth"(m)" 20" 40" 60" 80" Dive"depth"(m)"" 20" 40" 60" 80" " 20" 40" 60" 80" 100" 120" 100" St."George"Slope7Night" 97 km 19:26" 20:02" 20:45" 21:31" 0:39" 1:12" 2:45" 6:34" 7:24" 8:16" 9:06" 9:48" 10:26" 11:50" 12:51" 14:30" 100" Departure Arrival 2452 Figure 5. Examples of dive profiles of thick-billed murres from St. Paul and St. George Island during 2453 daytime and overnight trips to the Bering Sea Shelf and Slope. Maximum distance of each trip is shown 2454 in the top corner of each graph. Deep dives of the 99trip to Slope occurred on the Shelf during the in

100 2455 and outbound part of the trip N = 446 Mean = 47 ± 1.2 m N = 1590 Mean = 27 ± 0.5 m N = 97 Mean = 19 ± 2.2 m N = 579 Mean = 24 ± 0.9 m N = 1773 Mean = 27 ± 4.1m N = Mean = 14 ± 4.1 m

101 Figure 6. Hourly diving activity and mean depth of thick-billed murres nesting at St. Paul and St. George Islands during complete trips (day and overnight) to the Shelf (Middle and Outer) and Slope in The Outer Shelf was final destination of overnight trips or transit area from/to the Slope by birds from St. George. Color bars indicate daylight periods: black = nocturnal, grey = crepuscular and white = day

102 A 2494 Lipid content 2495 B St.#Paul# St.#George# 100%# n#=#450# n#=#202# n#=#97# 100%# n#=#102# n#=#321# n#=#222# %#Prey#items# 80%# 60%# 40%# 20%# Small# Medium# Large# %#Prey#items# 80%# 60%# 40%# 20%# Small# Medium# Large# %# 2008# 2009# 2010# 0%# 2008# 2009# 2010# Figure 7. Prey delivered to chicks by thick-billed murres at St. Paul (n = 278) and St. George Island (n = 264) between 2008 and A) Proportion of total prey species and schematic lipid content based on Van Pelt et al. (1997) and Whitman (2010). Osmeridae includes capelin, eulachon, and smelt. Gadids lipid content was assumed to be similar to juvenile pollock. B) Frequencies of prey size by year relative to adult bill-gape length. 102