A sum greater than its parts: merging multi-predator tracking studies to increase ecological understanding

Transcription

1 A sum greater than its parts: merging multi-predator tracking studies to increase ecological understanding A. D. LOWTHER, C. LYDERSEN, AND K. M. KOVACS Norwegian Polar Institute, Fram Centre, N-9296, Tromsø, Norway Citation: Lowther, A. D., C. Lydersen, and K. M. Kovacs A sum greater than its parts: merging multi-predator tracking studies to increase ecological understanding. Ecosphere 6(12): Abstract. Understanding how animals find prey in heterogeneous environments is a central goal of ecology. Placing this process in an environmental context requires a lot of information regarding the characteristics of both the habitat selected by the animal and its surroundings. In high-latitude marine systems, information about subsurface habitats of marine predators is often very limited. Animal-borne oceanographic instruments have added a new modality to improve our understanding of marine predators and their habitats. While these instruments do not collect environmental information beyond that experienced by the animals carrying them, our study makes use of an oceanographic dataset collected by southern elephant seals (Mirounga leonina; N ¼ 15), to provide environmental context for two sympatrically foraging penguin species in the waters close to the subantarctic island of Bouvetøya. The seals collected 154 CTD profiles during the study period, averaging 4.9 (63.67) profiles per day, documenting the stratification of the upper water layer in terms of both seawater density and temperature. Using these data, we quantitatively describe the relationship between the diving behavior of the penguins (N ¼ 3,745 dives) and the hydrographic properties of the three-dimensional area in which they were foraging. Both penguin species appeared to favor water characterized by a shallow mixed layer. The chinstrap penguins (Pygoscelis antarctica) dove within a shallow, unstable body of water close to the colony, whereas macaroni penguins (Eudyptes chrysolophus) exploited the bottom of the surface mixed layer further offshore. The hydrographic properties preferred by the penguins match closely those that describe the highest densities of their preferred prey, krill (Euphausia superba), identified during a temporally and spatially concurrent study. We demonstrate how merging multiple telemetric data streams from animals can shed new light on aspects of foraging behavior beyond simply relating movements to two-dimensional, remotely sensed measurements of the environment. Key words: biotelemetry; Bouvetøya; diving; Eudyptes chrysolophus; habitat modelling; Miroungia leonina; oceanography; penguins; Pygoscelis antarctica; Southern Ocean. Received 19 May 2015; revised 2 July 2015; accepted 6 July 2015; published 8 December Corresponding Editor: D. P. C. Peters. Copyright: Ó 2015 Lowther et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Andrew.Lowther@npolar.no INTRODUCTION One overarching goal of ecology is to characterize the relationships between organisms and the environments in which they exist. In artificial situations where prey are isolated from predation pressure and given a selection of habitat types, they will select those that provide the greatest energetic return (Hugie and Dill 1994). When predation pressure is introduced, prey are forced to optimize between energetic return and safety, such that the selected habitat will reflect a balance between these two forces, resulting in an evolutionarily stable state (Maynard Smith v 1 December 2015 v Volume 6(12) v Article 251

2 1974). An equivalent situation exists for predators, which must optimize between maximizing energetic returns through capturing adequate prey and selecting habitat that is optimal for reproduction (Hugie and Dill 1994). Thus, the distribution of predators depends on the distribution of prey, with both occupying habitats that reflect a process of optimization. In light of this the importance of environmental context is being recognized, as habitat selection may be based not only on the characteristics of the site itself, but also on the environmental characteristics of the area that surround it (De Knegt et al. 2011). Placing the habitat exploited by animals into a broader environmental context has the potential to offer substantial insight into their ecology. Quantifying the complexities actually involved in habitat selection typically involves relating when and where an animal moves to a range of biotic and abiotic parameters measured at varying degrees of spatial resolution such as land cover type (Foody 2002), forest structure (Hyde et al. 2006) and climatology (Daly et al. 1994) in some terrestrial environments. This has led to the field of landscape ecology, which relates geospatial information to the movement and behavior of animals (Wu 2013). Information on the physical and biological characteristics of the environment can be acquired through multiple streams of data collection ranging from aerial photography, remotely sensed satellite data and even physical inspection and cataloguing. Integrating these disparate datasets provides a mechanism to characterize the environmental context of animal movements. This endeavor is well advanced in many terrestrial systems, but in high-latitude marine environments contextualizing subsurface habitats used by predators is very challenging. Remotely sensed satellite data can provide information on characteristics of the ocean surface such as sea surface temperatures and height anomalies from which mesoscale (100s of km) and sub-mesoscale (10s of km) features can be derived (d Ovidio et al. 2010). In turn, these oceanographic features can be used to estimate regions of high primary productivity and these hot-spots can be related to the movement patterns of predators such as seabirds (Pakhomov and McQuaid 1996), penguins (Scheffer et al. 2012) and seals (Nordstrom et al. 2012). However, given the three dimensional nature of marine habitats these measurements are of limited use for characterizing the environment of diving predators that exploit the vertical dimension. Typically, vertically aligned oceanographic data are collected during vessel cruises or by floating drifter buoys such as ARGO floats (Roemmich et al. 2009). Both methods lack the ability to generate time-series data around a fixed point; cruises provide transect data and float movements are dictated by ocean currents. Electronic tags that can be carried by animals have provided a wealth of information on the subsurface environmental conditions experienced by marine species large enough to carry such instrumentation. Deploying a combination of temperature and light geolocating tags on tiger sharks Galeocerdo cuvier and Galapagos sharks Carcharhinus galapagensis highlighted that these two shark species remain primarily within the surface mixed layer (Meyer et al. 2010). Similarly, vertical profiles of environmental characteristics recorded by instruments on southern elephant seals Mirounga leonina revealed changes in dive behavior that were closely related to temperature (Biuw et al. 2010) and ambient light levels at depth (Jaud et al. 2012). However, animals such as flying and flightless seabirds are too small to carry large multi-sensor packages and thus data that they are able to collect is usually restricted to simplistic information on three-dimensional movement coupled with basic measurements of temperature and light. Regardless of the complexity of the environmental data collected by animal-borne tags, the instruments record only the environmental parameters in the immediate vicinity of the animal travels, which limits their utility in determining the environmental context of their movements. Consequently, most descriptions of marine predator habitat selection rely on descriptions of the conditions they experience coupled with relationships to remotely sensed surface data such as oceanographic fronts (Sabarros et al. 2013). Bouvetøya in the Southern Ocean (Fig. 1A) provides suitable breeding habitat for various central-place foraging marine predators including chinstrap penguins Pygoscelis antarcticus and macaroni penguins Eudyptes chrysolophus. Information on chinstrap penguin diet during the breeding season suggests a heavy reliance on Antarctic krill Euphausia superba, with mesopev 2 December 2015 v Volume 6(12) v Article 251

3 Fig. 1. (A) Location of Bouvetøya in the southeastern Atlantic Ocean (right panel), with the locations of the CTD casts from 15 southern elephant seals fitted with conductivity-temperature-depth satellite relay data loggers (CTD-SRDLs) during February 2008 (left panel). The red box highlights the geographical area over which vertical temperature and salinity profiles were collected to provide coverage of the three-dimensional area utilized by chinstrap and macaroni penguins. (B) Mixed layer depth derived from a mg m 3 change in seawater potential density anomaly from the surface value. Tracks of individual penguins are color-coded. v 3 December 2015 v Volume 6(12) v Article 251

4 lagic fish only occurring in their diet during periods of low krill abundance (Strass et al. 2002), whereas macaroni penguins are known to be generalist predators that consume several species of krill as well as myctophid fish Myctophidae spp. (Green et al. 2005). Thus, habitat selection by krill and myctophids likely determines the dive behavior of both penguin species. Both of these penguin species have been tracked from Bouvetøya in the recent past. Most of their diving was conducted at depths ranging between 10 and 100 m, in water that ranged between 0.58 and 1.58C. Most of their foraging trips were directed to the west of the colony within km of the island (Blanchet et al. 2013). The resolution of remotely sensed data from this area is typically at a scale of km, which makes it impossible to infer fine-scale relationships between surface conditions and the foraging behavior of birds. However, the presence of several oceanographic moorings that provided data across the threedimensional range in which these birds foraged provide a unique opportunity to look at the interactions between the subsurface environment and diving behavior of the birds. These moorings were 15 southern elephant seals instrumented with satellite-linked hydrographic sensors in the same austral summer season that the penguin tracking took place (Biuw et al. 2010). In the present study we use the hydrographic data collected by the instruments on the elephant seals to explore the late-breeding season dive behavior of the chinstrap and macaroni penguins in an oceanographic context, using the hydrographic data recorded by these southern elephant seals. Specifically, we predict that individuals from each penguin species target water masses with specific hydrographic properties that likely reflect the preferred conditions of their prey. MATERIALS AND METHODS This study utilizes data collected from elephant seals and penguins at Bouvetøya in the Southern Ocean ( S, E), as part of the Norwegian Antarctic Research Expedition (NARE). The island lies within the Antarctic Circumpolar Current (ACC), south of the Antarctic Polar Front (Fig. 1A). Oceanographic data presented in this study were collected using conductivity-temperature-depth satellite relay data loggers (CTD-SRDL; Sea Mammal Research Unit, University of St Andrews, UK) deployed on southern elephant seals (see Biuw et al for the details regarding instrument deployments on the seals). The oceanographic data reported herein are from 15 seals tagged on 20 January 2008 which remained in the vicinity of Bouvetøya until at least 28 February. The CTD- SRDLs deployed on these seals were calibrated after assembly (Boehme et al. 2009) and CTD profiles collected during dives were transmitted via the Argos satellite network. Detailed overviews of the instruments, onboard data handling, data compression and transmission, location and dive depth estimation are available in Boehme et al. (2009). The elephant seals collected CTD profiles throughout the water column around the western shelf of Bouvetøya, ensuring the entire vertical range of the penguins was covered. Southern elephant seal derived hydrography Error correction of Argos satellite telemetry data followed an iterative process: firstly, raw locations were pre-processed to remove extreme outliers by removing those with unclassified error estimates (LC- Z) and secondly, a swimspeed filter with a moving average of 2 m s 1 was applied (McConnell et al. 1992). Subsequently, locations were estimated using a Kalman filter under a state-space framework in the R package crawl (Johnson et al. 2008). This continuoustime correlated random walk model makes inferences on pre-processed data by considering time as a continuous variable, modeling the irregular temporal spacing of location estimates as a series of discrete time samples. The output of this analytical framework is a continuous movement path model from which location predictions can be made for any point in time. We estimated a location for each CTD profile that was likely accurate to within 65 km of the true location (Kuhn et al. 2009). For each hydrographic variable (temperature and salinity), each CTD profile was interpolated vertically at 1-m depth increments between the surface and 300 m. Then a two-dimensional isosurface was interpolated for each depth bin by inverse path distance weighting using the R package ipdw (Suominen et al. 2010) to create rasters of temperature and salinity at 1-m depth v 4 December 2015 v Volume 6(12) v Article 251

5 increments. We calculated seawater potential density anomalies (SWPDA; r h, mg m 3 ) at each point in the three-dimensional (3-D) hydrographic dataset using the Thermodynamic Equation of State of seawater (McDougall et al. 2009). A cross-sectional oceanographic profile running southwest from the Nyrøysa beach out to a distance of 55 km was generated for temperature (8C) and SWPDA in February (Fig. 2B), covering the general penguin foraging area described in Blanchet et al. (2013). Surface mixed layer depth (MLD) can be defined in terms of a critical SWPDA gradient or by a finite change in SWPDA from surface values (Brainerd and Gregg 1995). Given the coarse nature of depth sampling provided by compressed temperaturesalinity (T-S) profiles we employed the latter (Nilsen and Falck 2006). To avoid aliasing issues with the diurnal heating-cooling cycle and to acknowledge the effect of surface turbulence, we use a conservative SWPDA value of mg m 3. We also define the surface r h as the value calculated for the bottom of the first depth bin (4 m) to determine MLD at each interpolated location within the 3-D hydrographic dataset (Brainerd and Gregg 1995, Rudnick and Ferrari 1999, Huyer et al. 2005). Although quantitative analyses were conducted using this 3-D dataset, iso-surfaces and vertical cross-sections of the water column were visualized using the datainterpolating variational analysis method (DIVA) implemented in the software package ODV (Schlitzer 2011). This algorithm is analogous to optimal interpolation methods; however, it accounts for physical barriers such as coastlines and bathymetric features, and thus tends to produce realistic interpolations (Troupin et al. 2010). It also allows for creation of an easily interpreted visual representation of the hydrography, which is not readily achieved using the complex 3-D raster set used in the quantitative analysis. Chinstrap and macaroni penguin foraging behavior During the same field season in which the elephant seals carried CTD-SRDLs, 21 chinstrap and 16 macaroni penguins were also instrumented with satellite tracking devices (Kiwisat 101, Sirtrack, Havelock, New Zealand) and archival time-depth recorders (TDRs; Mk9, Wildlife Computers, Redmond, Washington, USA). In the current study, we selected three chinstrap penguins and two macaroni penguins for which dive and location data temporally spanned the CTD profiles collected by the elephant seals, when the birds were in the chick-feeding phase (N ¼ 3,745 dives). Penguin dive data were taken from the tags using the R package divemove, with a dive depth threshold of 6 m to reduce the influence of shallow transit dives on subsequent analyses. Penguin telemetry data were filtered using the R package crawl as outlined above, and a location for each penguin dive was estimated. Using the derived 3-D hydrographic dataset described above, a spatially referenced SWPDA value was extracted for each penguin dive. We investigated how each species exploited the water column by fitting density distributions to the maximum dive depths and the associated water temperature and SWPDA values for each species. We also derived metrics of maximum dive depth (m) and duration (s) from the penguin TDR data, and use these to construct a divebased residual as described in Bestley et al. (2015). As the depth to which an individual dives increases, so must the amount of time taken to descend and ascend. Thus, the magnitude and direction of residuals of a modeled relationship between dive depth and duration represent individual dives that were either shorter or longer than expected for a given depth. Intuitively, a longer dive may reflect an individual expending more effort in trying to forage. We constructed our dive residuals by fitting a linear mixed effects model with a log-log relationship, allowing the slope and intercept to vary between penguins with respect to dive depth (Bestley et al. 2015). Two models (one for each species) were fitted using restricted maximum likelihood (REML) estimation in the R package nlme, and Pearson residuals extracted from the fitted model were used as our dive residuals. Using these dive residuals, we quantified the environmental conditions in which penguins conducted longer-than-expected dives using species-specific generalized additive mixed models (GAMM), implemented in the R package gamm4. We chose to separate the species in our modeling approach to account for speciesspecific foraging behavior and prey preference. Each model was fitted with dive residual as the dependent variable, hydrographic variables v 5 December 2015 v Volume 6(12) v Article 251

6 Fig. 2. Vertical dive profiles for (A) chinstrap and (C) macaroni penguins tracked during February 2008 at Bouvetøya. Vertical dark bars represent darkness (defined as the onset and end of nautical twilight). (B) Interpolated vertical temperature (top) and seawater potential density anomaly (bottom) profiles derived from CTD-SRDL measurements collected by 15 instrumented southern elephant seals over the same time period in the bounding box outlined in Fig. 1. Thermoclines and isopycnals are clearly visible, with the previous years winter water forming a cold layer at 100 m. (SWPDA, temperature and salinity) as a smooth term fixed effect and individual penguins as random effects; a Gaussian error distribution with identity link was included. Given that the dive behavior of each individual was correlated, we included a first-order autocorrelation term to each model and fitted each using REML. Each GAMM model is represented as a curvilinear relationship (62 SE) depicting how the dive residuals differ from mean values in relation to changes in SWPDA, temperature and salinity. Results are presented as significant at p, 0.05 v 6 December 2015 v Volume 6(12) v Article 251

7 and mean 6 SD unless otherwise stated. RESULTS Southern elephant seal derived hydrography Telemetry data from 15 southern elephant seals were used to provide oceanographic data for the shelf region of western Bouvetøya between 1 and 28 February 2008, from 154 CTD profiles, collected at a rate of approximately per day (Fig. 1A). Profiles were taken over a mean depth of 337 m (65 m) with a maximum depth of 860 m; 31% of all oceanographic data profiles were from water depths greater than 400 m. The depth of the mixed layer on the west side of the island varied spatially, ranging between 60 and 160 m during February (Fig. 1B). The CTD data within the section outlined in Fig. 1A clearly show the vertical properties of the water column in terms of temperature and water density along the western shelf of Bouvetøya (Fig. 2B, top panel). During February the upper 300 m of water was thermally stratified, with a cold layer of water ( C) between 100 and 200 m overlaid by surface waters that were up to 1.58C. The cold layer represents the remnants of the cold winter water (WW) of the Antarctic Surface Water (ASW) from the previous year (Fig. 2B). At the same time, SWPDA data also described two isopycnals separating the two upper layers from the denser upper Circumpolar Deep Water (ucdw; ;27.7 mg m 3 ; Fig. 2B). Penguins dive behavior and environmental covariates The five penguins performed a total of 3,745 dives between 1 and 11 February 2008, with foraging trip durations ranging from 7.3 to 12.9 hours (Table 1). Within the period of this study, all individuals performed complete foraging trips, that is, they left the colony to go to sea and subsequently returned to land. On average, the macaroni penguins dove deeper ( m cf m; Welch s two sample t-test, t ¼ 21.6, p, 0.001) and longer ( s cf s; Welch s two sample t-test t ¼ 24.3, p, 0.001) than the chinstrap penguins (Table 1). The macaroni penguins also traveled approximately four times farther away from the Nyrøysa colony during foraging trips (Fig. 1B; Table 1). From both vertically and horizontally interpolated CTD data, it appears that both penguin species utilized areas characterized by a shallow MLD and avoided regions where the MLD was deeper than 100 m (Figs. 1B and 2). Chinstrap penguin maximum dive depth was best approximated by an exponential distribution, whereas the SWPDA values associated with these dives followed a normal distribution (mean SWPDA: mg m 3 ; Fig. 3A), with mean maximum dive depths occurring in the middle of the mixed layer, in a variable region of water within 10 km of the breeding colony (Figs. 1B and 2A, B). Conversely, macaroni penguins exhibited a bimodal normal distribution of maximum dive depths and SWPDA values associated with dives (mean maximum dives: 7.66 m and m, respectively; mean SWPDA: 27.2 and 27.4 kg m 3 ; Fig. 3B), with the deepest diving taking individuals to the bottom of the mixed layer almost 50 km west of the colony (Figs. 1B and 2B, C). Interestingly, both species focused their diving efforts in waters with a relatively narrow temperature range that fit a normal distribution (mean Chinstrap penguin maximum dive temperature: 0.948C; macaroni penguin maximum dive temperature: 0.948C; Fig. 3). Positive dive residuals for both penguin species were associated with larger salinity and SWPDA values, and a relatively narrow temperature range between 0.88 and 0.98C (GAMM: edf. 7.9, F. 97.5, p, in all cases; Fig. 4; Appendix B). Chinstrap penguins appeared to conduct longer dives in areas characterized by water with a SWPDA value ;0.1 kg m 3 lower than macaroni penguins and responded more strongly to water temperature (Fig. 4; Appendix B). DISCUSSION Typically, the data received from animal-borne sensors reflects the conditions that individuals experience in pursuit of their goals. For example, taken in isolation the data collected by the penguins at Bouvetøya describe the dive depths they traveled to in order to forage. However, sensor information simply stops at the point where the penguins achieved their goal, or had to abort the dive, to return to the surface for air. To place their behavior into a broader context v 7 December 2015 v Volume 6(12) v Article 251

8 Table 1. Summary trip and dive statistics for three Chinstrap (Chin) and two Macaroni (Mac) penguins tracked between 1st and 11th February 2008 at Bouvetøya. Values are given as mean 6 SD. Trip characteristics Chin 1 Chin 2 Chin 3 Mac 1 Mac 2 Trips (N) Duration (h) Dives (N) Depth (m), mean Depth (m), maximum Duration (s), mean Duration (s), maximum requires describing the conditions that the subject animals did not experience, so that the environmental thresholds or gradients that dictate the limits of animal movement can be identified. The tagging of deep-diving elephant seals with satellite-linked sensors is becoming an increasingly common means to observe the physical oceanographic properties of the southern ocean (Guinet et al. 2013, Roquet et al. 2014). The seals utilized in this study remained resident in the shelf waters around Bouvetøya during the time that other, shallower-diving species were also being tracked at sea, providing us with a unique dataset. The environmental parameters collected by the seals allowed us to place the diving and movement behavior of two species of penguins in context within a subset of the water column exploited by the elephant seals. Furthermore, comparing two species of penguins with divergent strategies during the same time period gives an important insight into the ecology of penguins. To our knowledge this study represents the first of its kind to place environmental context on the 3-D movement of air-breathing marine predators, including characterizing the environmental characteristics nearby the habitats that were not used by the predators. Both Blanchet et al. (2013) and Lowther et al. (2014) reported that during the breeding period (January February), chinstrap and macaroni penguins consistently traveled in a westerly direction on foraging trips out from Bouvetøya, though to different distances from the colony (chinstrap penguins, ;10 km; macaroni penguins, ;50 km). Chinstrap penguins appear to have a strong preference for krill (Rombolá et al. 2010), and the comparatively shallow diving patterns of this species at Bouvetøya seems to reflect a species-level trend as opposed to location or population-level trends (Bengtson et al. 1993, Takahashi et al. 2004, Kokubun et al. 2010). Antarctic krill can aggregate in swarms that stretch from the seafloor to the surface, though their growth rates appear to be linked to water temperature and food availability (Atkinson et al. 2008). Maximal growth rates are recorded at water temperatures between 0.58 and 1.08C, and living in the productive habitats offered by shelf regions likely represents a tradeoff between increased growth and risk of predation by land-based marine predators such as penguins (Atkinson et al. 2008). Indeed, a concurrent study, independent of ours, detected several areas of high krill abundance in the region of Bouvetøya (Krafft et al. 2010). Using cluster analysis, Krafft et al. (2010) documented that the densest swarms of krill were found in waters with mean SWPDA and water temperature values of 27.2 and kg m 3 and 0.28 and 1.58C, respectively (see Appendix A; Krafft et al. 2010). These findings match very closely the mean temperature and SWPDA distribution values at the maximum dive depths of penguins in our study, and are consistent with the water conditions in which penguins prolonged their dives. Thus, our data support the hypothesis that the near-shore foraging trips and shallow dive depths of chinstrap penguins in the shallow mixed layer closer to Bouvetøya represent these individuals specifically targeting water properties favored by krill. However, if krill are available this close to the colony, why did macaroni penguins in our study travel further, dive much deeper and appear to favor two different sets of water conditions compared to chinstrap penguins? Under optimal foraging theory one would predict that macaroni penguins traveling greater distances from the colony would need to be rewarded by richer or more plentiful food supplies (Charnow 1976). v 8 December 2015 v Volume 6(12) v Article 251

9 Fig. 3. Frequency and fitted distribution plots for (A) chinstrap penguins and (B) macaroni penguins electronically tracked at Bouvetøya in February Vertical dotted lines represent mean values of fitted distributions in terms of maximum depth (top), and the respective seawater potential density anomaly (SWPDA; middle) and water temperature (bottom) values for each dive. Chinstrap penguin maximum dive depth was best fitted with an exponential distribution. Mean distribution values for SWPDA and temperature were concordant with the environmental characteristics described by Krafft et al. (2010) for krill abundance in the Bouvetøya region. Note that macaroni penguins exhibited bimodal normal distributions of dive depth and SWPDA. Macaroni penguins are mixed-prey predators, eating myctophids such as the lanternfish (Krefftichthys anderssoni) and carnivorous crustacea (Amphipoda: hyperiidae spp; Deagle et al. 2007) in addition to several species of krill. The energetic content of myctophid fishes and carnivorous macro-zooplankton is much higher than that of krill (Williams and Robins 1979, Doidge and Croxall 1985). Studies of mesopelagic fish assemblages off another subantarctic island, Kerguelen, showed a dominance of hypderiid amphipods, associated with lanternfish, which tended to be km offshore in the upper layers of shelf and slope waters; these assemblages were rarely shallower than 50 m during daylight (Duhamel et al. 2000). Considering this, it is reasonable to assume that the increased energetic requirements of v 9 December 2015 v Volume 6(12) v Article 251

10 Fig. 4. Relationship between the effect of seawater potential density anomaly (SWPDA) on dive residuals characterized from generalized additive mixed modeling for (A) chinstrap penguins and (B) macaroni penguins. Mean SWPDA values (solid red lines) and their upper and lower estimates (dashed red lines) of highest krill densities in the Southern Ocean near Bouvetøya identified by Krafft et al. (2010; Appendix: Fig. A1). Positive and negative Y-axis values reflect longer and shorter dives than would be predicted, respectively. Both species dove for longer than predicted in oceanographic conditions where krill were likely to be found. growing macaroni penguin chicks may be more readily met by parents through foraging on these energetically more profitable prey items. We suggest that the increased horizontal and vertical travel by macaroni penguins compared to chinstraps, coupled with the bimodal distribution of maximum dive depths and water characteristics, may reflect prey switching at this critical period of the breeding cycle. Our study did not include a dietary component, however prey switching in favor of more energetically valuable prey during the late breeding period has been recorded previously for macaroni penguins (Croxall et al. 1988). CONCLUSIONS The collection of oceanographic data from deep-diving southern elephant seals has allowed us to characterize the physical environment around a remote subantarctic island that supports a few shallower diving central place foragers. In particular, this dataset allowed us to explore the environmental context of the foraging movements of individuals from two species of penguins with different foraging strategies, and provide further insight into what aspects of the marine environment might underpin these differences. Our study highlights that integrating multi-species studies in this way v 10 December 2015 v Volume 6(12) v Article 251

11 would greatly improve our understanding of why individuals do what they do by providing much more realistic and finer-scale environmental data for modeling exercises. In turn, scaling up these individual-level behaviors would enable population level processes to be summarized more appropriately. Thus, in circumstances in which guilds of centrally foraging predators reside, it is possible to enhance our understanding of why individuals select certain habitats by attaching multi-sensor tags to the species that dives deepest or ranges farthest. ACKNOWLEDGMENTS This study is part of MEOP (Marine Mammals Exploring the Oceans Pole to Pole), Norway, which was an International Polar Year (IPY) programme. Fieldwork was funded and carried out as part of the Norwegian Antarctic Research Expedition to Bouvetøya. We are greatly indebted to Martin Biuw, Greg Hofmeyr, Nico debruyn, Petrus Kritzinger and Aline Arriola for their efforts during the fieldwork. LITERATURE CITED Atkinson, A., V. Siegel, E. Pakhomov, P. Rothery, V. Loeb, R. Ross, L. Quetin, K. Schmidt, P. Fretwell, and E. Murphy Oceanic circumpolar habitats of Antarctic krill. Marine Ecology Progress Series 362:1 23. Bengtson, J. L., D. A. Croll, and M. E. Goebel Diving behaviour of chinstrap penguins at Seal Island. Antarctic Science 5:9 15. Bestley, S., I. D. Jonsen, M. A. Hindell, R. G. Harcourt, and N. J. Gales Taking animal tracking to new depths: synthesizing horizontal-vertical movement relationships for four marine predators. Ecology 96: Biuw, M., O. A. Nøst, A. Stien, Q. Zhou, C. Lydersen, and K. M. Kovacs Effects of hydrographic variability on the spatial, seasonal and diel diving patterns of southern elephant seals in the eastern Weddel Sea. PLoS ONE 5(11):e Blanchet, M.-A., M. Biuw, G. J. G. Hofmeyr, P. J. N. De Bruyn, C. Lydersen, and K. M. Kovacs At-sea behaviour of three krill predators breeding at Bouvetøya: Antarctic fur seals, macaroni penguins and chinstrap penguins. Marine Ecology Progress Series 477: Boehme, L., P. Lovell, M. Biuw, F. Roquet, J. Nicholson, S. E. Thorpe, M. P. Meredith, and M. Fedak Technical note: animal-borne CTD-Satellite Relay Data Loggers for real-time oceanographic data collection. Ocean Science 5: Brainerd, K. E., and M. C. Gregg Surface mixed and mixing layer depths. Deep-Sea Research, Part I, Oceanographic Research Papers 42: Charnow, E. L Optimal foraging, the marginal value theorem. Theoretical Population Biology 9: Croxall, J. P., R. W. Davis, and M. J. O Connell Diving patterns in relation to diet of gentoo and macaroni penguins at South Georgia. Condor 90: Daly, C., R. P. Neilson, and D. L. Phillips A statistical-topographic model for mapping climatological precipitation over mountainous terrain. Journal of Applied Meteorology 33: Deagle, B. E., N. J. Gales, K. Evans, S. N. Jarman, S. Robinson, R. Trebilco, and M. A. Hindell Studying seabird diet through genetic analysis of faeces: a case study on macaroni penguins (Eudyptes chrysolophus). PLoS ONE 2:e831. De Knegt, H. J., F. Van Langevelde, A. K. Skidmore, A. Delsink, R. Slotow, S. Henley, G. Bucini, W. F. De Boer, M. B. Coughenour, and C. C. Grant The spatial scaling of habitat selection by African elephants. Journal of Animal Ecology 80: Doidge, D. W., and J. P. Croxall Diet and energy budget of the Antarctic fur seal, Arctocephalus gazella, at South Georgia. Pages in W. R. Siegfried, P. R. Condy, and R. M. Laws, editors. Antarctic nutrient cycles and food webs. Springer Verlag, Heidelberg, Germany. d Ovidio, F., S. De Monte, S. Alvain, Y. Dandonneau, and M. Lévy Fluid dynamical niches of phytoplankton types. Proceedings of the National Academy of Sciences USA 107: Duhamel, G., P. Koubbi, and C. Ravier Day and night mesopelagic fish assemblages off the Kerguelen Islands (Southern Ocean). Polar Biology 23: Foody, G. M Status of land cover classification accuracy assessment. Remote Sensing of Environment 80: Green, J., I. Boyd, A. Woakes, N. Warren, and P. Butler Behavioural flexibility during year-round foraging in macaroni penguins. Marine Ecology Progress Series 296: Guinet, C., X. Xing, E. Walker, P. Monestiez, S. Marchand, B. Picard, T. Jaud, M. Authier, C. Cotté, and A. Dragon Calibration procedures and first dataset of Southern Ocean chlorophyll a profiles collected by elephant seals equipped with a newly developed CTD-fluorescence tags. Earth System Science Data 5: Hugie, D. M., and L. M. Dill Fish and game: a game theoretic approach to habitat selection by predators and prey. Journal of Fish Biology 45: v 11 December 2015 v Volume 6(12) v Article 251

12 Huyer, A., J. H. Fleischbein, J. Keister, P. M. Kosro, N. Perlin, R. L. Smith, and P. A. Wheeler Two coastal upwelling domains in the northern California Current system. Journal of Marine Research 63: Hyde, P., R. Dubayah, W. Walker, J. B. Blair, M. Hofton, and C. Hunsaker Mapping forest structure for wildlife habitat analysis using multi-sensor (LiDAR, SAR/InSAR, ETMþ, Quickbird) synergy. Remote Sensing of Environment 102: Jaud, T., A.-C. Dragon, J. V. Garcia, and C. Guinet Relationship between chlorophyll a concentration, light attenuation and diving depth of the southern elephant seal Mirounga leonina. PLoS ONE 7:e Johnson, D. S., J. M. London, M. A. Lea, and J. W. Durban Continuous-time correlated random walk model for animal telemetry data. Ecology 89: Kokubun, N., A. Takahashi, Y. Mori, S. Watanabe, and H.-C. Shin Comparison of diving behavior and foraging habitat use between chinstrap and gentoo penguins breeding in the South Shetland Islands, Antarctica. Marine Biology 157: Krafft, B., W. Melle, T. Knutsen, E. Bagøien, C. Broms, B. Ellertsen, and V. Siegel Distribution and demography of Antarctic krill in the Southeast Atlantic sector of the Southern Ocean during the austral summer Polar Biology 33: Kuhn, C. E., D. S. Johnson, R. R. Ream, and T. S. Gelatt Advances in the tracking of marine species: using GPS locations to evaluate satellite track data and a continuous-time movement model. Marine Ecology Progress Series 393: Lowther, A. D., C. Lydersen, M. Biuw, P. J. N. De Bruyn, G. J. G. Hofmeyr, and K. M. Kovacs Post-breeding at-sea movements of three centralplace foragers in relation to submesoscale fronts in the Southern Ocean around Bouvetøya. Antarctic Science 6: Maynard Smith, J The theory of games and the evolution of animal conflicts. Journal of Theoretical Biology 47: McConnell, B. J., C. Chambers, and M. A. Fedak Foraging ecology of southern elephant seals in relation to the bathymetry and productivity of the Southern Ocean. Antarctic Science 4: McDougall, T., R. Feistel, F. Millero, D. Jackett, D. Wright, B. King, G. Marion, C. Chen, P. Spitzer, and S. Seitz The international thermodynamic equation of seawater 2010 (TEOS-10): calculation and use of thermodynamic properties. Global shipbased repeat hydrography manual. Intergovernmental Oceanographic Commission Report. UNESCO., Paris, France. Meyer, C. G., Y. P. Papastamatiou, and K. N. Holland A multiple instrument approach to quantifying the movement patterns and habitat use of tiger (Galeocerdo cuvier) and Galapagos sharks (Carcharhinus galapagensis) at French Frigate Shoals, Hawaii. Marine Biology 157: Nilsen, J., and E. Falck Variations of mixed layer properties in the Norwegian sea for the period Progress in Oceanography 70: Nordstrom, C. A., B. C. Battaile, C. Cotté, and A. W. Trites Foraging habitats of lactating northern fur seals are structured by thermocline depths and submesoscale fronts in the eastern Bering Sea. Deep-Sea Research, Part II, Topical Studies in Oceanography 88-89: Pakhomov, E. A., and C. D. McQuaid Distribution of surface plankton and seabirds across the Southern Ocean. Polar Biology 16: Roemmich, D., G. C. Johnson, S. Riser, R. Davis, J. Gilson, W. B. Owens, S. L. Garzoli, C. Schmid, and M. Ignaszewski The Argo Program: observing the global ocean with profiling floats. Oceanography 22: Rombolá, E. F., E. Marschoff, and N. Coria Interannual variability in Chinstrap penguin diet at South Shetland and South Orkneys Islands. Polar Biology 33: Roquet, F., G. Williams, M. A. Hindell, R. Harcourt, C. McMahon, C. Guinet, J.-B. Charrassin, G. Reverdin, L. Boehme, and P. Lovell A Southern Indian Ocean database of hydrographic profiles obtained with instrumented elephant seals. Scientific Data 1: Rudnick, D. L., and R. Ferrari Compensation of horizontal temperature and salinity gradients in the ocean mixed layer. Science 283: Sabarros, P. S., D. Grémillet, H. Demarcq, C. Moseley, L. Pichegru, R. H. Mullers, N. C. Stenseth, and E. Machu Fine-scale recognition and use of mesoscale fronts by foraging Cape gannets in the Benguela upwelling region. Deep-Sea Research, Part II, Topical Studies in Oceanography. Scheffer, A., C.-A. Bost, and P. N. Trathan Frontal zones, temperature gradient and depth characterize the foraging habitat of king penguins at South Georgia. Marine Ecology Progress Series 465: Schlitzer, R Ocean Data View 4.5.3: users manual. Bremerhaven, Germany. Strass, V. H., A. C. Naveira Garabato, R. T. Pollard, H. I. Fischer, I. Hense, J. T. Allen, J. F. Read, H. Leach, and V. Smetacek Mesoscale frontal dynamics: shaping the environment of primary production in the Antarctic Circumpolar Current. Deep-Sea Research, Part II, Topical Studies in Oceanography 49: Suominen, T., H. Tolvanen, and R. Kalliola Surface layer salinity gradients and flow patterns in the archipelago coast of SW Finland, northern v 12 December 2015 v Volume 6(12) v Article 251

13 Baltic Sea. Marine Environmental Research 69: Takahashi, A., M. Dunn, P. Trathan, J. Croxall, R. P. Wilson, K. Sato, and Y. Naito Krill-feeding behaviour in a chinstrap penguin compared to fisheating in Magellanic penguins: a pilot study. Marine Ornithology 32: Troupin, C., F. Machin, M. Ouberdous, D. Sirjacobs, A. Barth, and J. M. Beckers High-resolution climatology of the northeast Atlantic using datainterpolating variational analysis (DIVA). Journal of Geophysical Research 115:1 20. Williams, R., and D. Robins Calorific, ash, carbon and nitrogen content in relation to length and dry weight of Parathemisto gaudichaudi (Amphipoda: Hyperiidea) in the North East Atlantic Ocean. Marine Biology 52: Wu, J. J Landscape ecology. Pages in R. Leemans, editor. Ecological systems. Springer, New York, New York, USA. SUPPLEMENTAL MATERIAL ECOLOGICAL ARCHIVES Appendices A and B are available online: v 13 December 2015 v Volume 6(12) v Article 251