Marine Biology (2002) 141: 581–589 DOI 10.1007/s00227-002-0843-4

J.-B. Charrassin Æ Y. Le Maho Æ C.-A. Bost Seasonal changes in the diving parameters of king (Aptenodytes patagonicus)

Received: 3 May 2001 / Accepted: 18 March 2002 / Published online: 5 July 2002 Springer-Verlag 2002

Abstract Contrastingconditions at-sea are likely to af- and with dive depth, but they were longer in spring fect the foraging behaviour of seabirds. However, the (2.3 min for dives over the 100–210 m layer) and sum- effect of season on the dive parameters of penguins is mer than in autumn and winter (1.6–1.8 min). The div- poorly known. We report here on an extensive study of ingefficiency decreased with increasingdive depth and the divingbehaviour of kingpenguins( Aptenodytes was higher in autumn and winter (0.22–0.29) than in patagonicus) over the ’s complete annual cycle at the summer and spring(0.15–0.18). The largeincrease Crozet Islands. Time-depth recorders were used to re- in bottom and dive duration from springto winter is in cord dive duration, bottom duration, post-dive interval, agreement with the seasonal drop in prey density, with ascent rate and descent rate in breedingadults during penguins spending more time searching for prey. In different seasons in 1995 and 1996. Seasons included contrast, the consistency of the vertical velocity during summer (n=6, incubation; n=6, chick brooding), au- contrasting conditions at-sea suggests that the transit tumn and winter (n=5 and n=3, respectively, chick at time to depth is an important component of the foraging the cre` che stage), and spring (n=4, at the post- behaviour (scanningof the water column) that is inde- moult stage). In all seasons dive duration increased with pendent of the prey availability. The time budget of the dive depth, but, for a given depth, dives were longer in penguins during diving in a fluctuating environment winter (6.8 min when averaged over the 100–210 m appears to vary primarily duringthe bottom phase of depth layer) than in spring(4.6 min) and summer the dives, with bottom duration increasingwith dimin- (4.4 min). The time spent at the bottom of the dives, ishingprey supplies, while post-dive intervals shorten in which probably represents a substantial part of the the same time. feedingtime, was much longerin winter (2.5 min per dive for dives over the 100–210 m layer) than during other seasons (1.0–1.4 min), i.e. there was a 2.5-fold Introduction augmentation for similar diving depths. Ascent and de- scent rates increased with increasingdive depth, but no In air-breathingdivers such as seabirds and pinnipeds, a difference in the relationships between rates of ascent dive typically consists of a descent phase to depth, a period and descent and dive depth was found amongseasons. spent at the bottom at maximum depth, and an ascent Furthermore, for all dive depths, ascent and descent phase of return to the surface (Schreer et al. 2001). This rates were independent of the bottom duration. In all basic pattern has been shown by numerous studies in seasons post-dive intervals increased with dive duration which the hydrostatic pressure encountered by the diver was recorded as a function of time by a miniaturised time- Communicated by S.A. Poulet, Roscoff depth recorder (TDR) carried by the (e.g. Kooy- man et al. 1992; Chappell et al. 1993; Wilson et al. 1996; J.-B. Charrassin (&) Æ Y. Le Maho Æ C.-A. Bost Kirkwood and Robertson 1997a; Georges et al. 2000; Centre d’Ecologie et Physiologie Energe´ tiques, Rodary et al. 2000). This large body of data has allowed Centre National de la Recherche Scientifique, major inter-specific comparisons of dive performance. 3 rue Becquerel, 67087 StrasbourgCedex, France Briefly, body size is the main source of variation between e-mail: [email protected] divingspecies, with largespecies divingdeeper and longer Tel.: +33-1-40793164 Fax: +33-1-40793163 than small . Secondly, for most species investigated so far, dive duration and vertical velocity duringdiving Present address: J.-B. Charrassin strongly increase with increasing dive depth. These pat- Laboratoire d’Oce´ anographie Physique, Muse´ um National d’Histoire Naturelle, terns have been described in several reviews (e.g. Wilson 43 rue Cuvier, 75231 Paris Cedex 05, France 1995; Schreer et al. 2001). 582 However, most studies investigating dive patterns were post-dive interval, and vertical ascent and descent rates conducted duringsummer, which corresponds to the pe- across seasons. The results are discussed in terms of riod of the year when most species breed. At that time, the seasonal prey availability and behavioural adjustments. are readily accessible at the colony, their foraging trips are short, and they regularly return to the colony to feed their young. Consequently, the probability of re- Materials and methods coveringthe equipment is highestin summer. In contrast, few studies have attempted to compare intraspecifically Animals and instrumentation the divingbehaviour over different seasons. This is Field work was conducted in 1995 and 1996 at ‘‘La Grande probably related to logistic problems and to the limited Manchotie` re’’ colony, Possession Island, Crozet Archipelago memory size of instruments that does not allow coverage (4625¢S; 5145¢E). About 40,000 pairs of kingpenguins( Apteno- of the entire duration of the much longer winter trips. dytes patagonicus) breed in this colony, with the whole Crozet However, the marine environment shows strongseasonal Archipelago (1 million pairs) representing 50% of the world population (Guinet et al. 1995). The divingbehaviour of 21 birds changes in biological production (Foxton 1956; Clarke was monitored so that birds over the complete annual cycle of the 1988), which can have a profound impact on the feeding kingpenguinwere represented, as described in Charrassin and Bost ecology of predators. The winter season is characterised (2001). Birds studied in summer (1995) were either incubatingor by a drastic drop in the marine primary production. Pre- broodinga 1- to 3-week-old chick. Birds studied in autumn (1995) and in winter (1995–1996) were caringfor an emancipated chick at dators breedingon the Antarctic continent may respond the cre` che stage (i.e. older than 6 weeks). Birds studied in spring by a shift of their diet, e.g. from fish (summer) to krill (1996) were at the post-moult stage; they were unsuccessful (winter) in emperor penguins (Aptenodytes forsteri)orby breeders from 1995 that were randomly captured just after moult in long-distance migrations in Ade´ lie ( adeliae) the earliest cohort of moulting penguins. All foraging trips were and chinstrap penguins (Pygoscelis ) (Davis performed by different individuals, except for three birds for which we studied the foraging activity at the incubating and brooding et al. 1996; Kirkwood and Robertson 1997b; Wilson et al. stages consecutively. TDRs deployed in summer, autumn and 1998). In the permanently open-ocean zone, zooplankton winter were Mk5 3.0 (95·38·15 mm, 70 g) (Wildlife Computers, (e.g. copepods) migrate to greater depth in winter, fol- USA), and those used at the post-moult stage were Mk5 3.3 lowed by some zooplankton consumers (e.g. mesopelagic (110·38·15 mm, 90 g). The TDRs recorded hydrostatic pressure with a 2 m depth resolution over a range of 0–500 m and had a fish) which, in turn, are potential prey of divingpredators 512 kb memory. The samplinginterval was 5 s in summer, and 10 s (Smith and Schnack-Schiel 1990; Koslov et al. 1991; duringthe autumn, winter and post-moultingperiods. Depth Ridoux 1994; Woehler 1995). Most species desert their measurements were made every second day in the winter cre` che breedingcolony in winter, and probably travel to more group to allow for complete coverage of long trips. To reduce the hydrodynamic drag(Bannasch et al. 1994), the TDRs were fitted to productive regions, but comprehensive behavioural data the lower back of the birds. Cable-ties were used to fix the TDRs to are missing. Inter-seasonal dive studies can elucidate the a small metal grid, which was glued to the feathers of the back with short-scale responses of divingpredators to these drasti- fast epoxy. Birds were flipper marked with coloured tape. After cally contrastingat-sea conditions. A few studies exist for equipment, the birds were released close to the edge of the colony. seals (e.g. Georges et al. 2000), but, with exception of a All birds were freed of their devices upon their return. Maximum care was taken to reduce the stress to the birds, e.g., by covering the study on gentoo penguins (Pygoscelis papua) at South birds’ eyes duringhandlingand by movingcarefully. Georgia (Williams et al. 1992b) and one on king penguins at Heard Island (Moore et al. 1999), virtually nothingis known about penguins. Analysis of divingbehaviour As part of a long-term study, we examined the for- A dive-per-dive analysis was conducted on depth data that were aging behaviour of the king (Aptenodytes pat- corrected for surface drift (range: ±2 to 10 m) using custom-made agonicus) at the Crozet Islands over its annual cycle software (Jensen Software Systems, Laboe, Germany). Only dives (Bost et al. 1997; Charrassin et al. 1998, 1999; ‡4 m were analysed, because dives <4 m could not be reliably re- solved by the instruments. Maximum dive depth, dive duration, post- Charrassin and Bost 2001). The kingpenguinis an dive interval, time spent at the bottom and vertical dive velocities oceanic, deep diver that feeds on mesopelagic fish dis- were obtained. Post-dive intervals were calculated between consec- tributed at depths of 100–500 m. An interestingaspect utive deep dives (i.e. dives to ‡50 m, Ropert-Coudert et al. 2000a). of its biology is that the breeding cycle lasts for >1 year, Post-dive intervals lasting>1,000 s represented <90% in number, with chicks hatchingin early summer and reared and were excluded from the analysis. Bottom duration was defined as the interval of time spent at a depth >90% of the maximum depth of throughout the winter until they fledge in spring. By the dive. This depth threshold was found by comparingthe ‘‘real’’ studyinganimals over the annual cycle with satellite- bottom duration, which was determined visually from the dive pro- trackingand TDRs, we observed a drastic increase of files, with calculated bottom durations usingdifferent thresholds. the foraging distances over the seasons. In summer birds For 900 dives to >50 m, recorded for three individuals, a 90% depth threshold resulted in a bottom duration equal to 95% of the ‘‘real’’ foraged 400 km south of the colony, while in autumn bottom duration. Furthermore, the average threshold given by the and winter they travelled down to the pack-ice area ‘‘real’’ bottom duration was 92% of the maximum dive depth. Ver- 1,600 km away. Dive depth simultaneously increased tical velocities were obtained duringthe descent phase (descent rate from 160 m in summer to 200 m in winter calculated from the surface to the beginning of the bottom phase) and duringthe ascent phase (ascent rate calculated from the end of the (Charrassin and Bost 2001). Here, we present detailed bottom phase to the surface). The divingefficiency was calculated data on the divingbehaviour of kingpenguinsover the followingthe equation of Ydenbergand Clark (1989): divingeffi- annual cycle, comparingdive duration, bottom duration, ciency=bottom time/(dive duration+post-dive interval). The data 583 were first analysed over each entire foraging trip. To highlight the autumn cre` che stages [Student–Newman–Keuls (SNK) divingbehaviour specifically for the main foragingareas,a separate post hoc tests]. Duringthe central phase, averagesof analysis was conducted for the central phase of each trip, i.e. when the birds were at the most southerly position. To achieve this, we studied dive durations over the 100–210 m layer increased from the 2 days from each trip when sea surface temperatures (recorded 4.4 min in summer to 6.8 min in winter (Table 1). simultaneously with dive data) were the coldest (Charrassin and Bost 2001). We verified the degree of auto-correlation within the temporal Bottom duration in relation to dive depth series of dives to ‡50 m by conductinga partial auto-correlation analysis. Maximum depth and dive duration of any particular dive were correlated to those of the consecutive dive, but lost correlation In all groups, the bottom duration increased in relation after two dives. We thus removed one dive (to ‡50 m) out of two to depth between 0 and 100 m, before becomingvir- from the data set. The resultingnumber of dives performed per tually stable at greater depths (Fig. 1b). Bottom dura- animal varied from 230 to 1,800, dependingon the season. To give an equal statistical weight to each penguin, we randomly selected tion for a given depth increased dramatically over the the same number of dives per bird (530 dives to <50 m and 230 season (nested two-way ANOVA accountingfor varia- dives to ‡50 m) for statistical analyses. tion due to dive depth, Fig. 1b). While bottom periods Statistics were performed on individual data, averaged per 10 m had a similar duration duringincubation in summer and depth interval, while data are presented in the figures as means (±SE) per group (season) of data, averaged per 10 m depth in- in spring, and were shortest at these stages, they were terval. Dive parameters in relation to depth were compared among longest in autumn and winter (SNK post hoc tests). groups using nested two-way ANOVAs (individuals nested in Duringthe central phase of the trip, averagesof bottom groups, and individuals and groups crossed with depth). To allow durations for dives over the 100–210 m layer ranged for the nested ANOVAs, comparisons were performed on groups from 1 min per dive duringsummer (incubation) to of four individuals, and thus did not include winter data. For this, individuals were taken as beingindependent (for incubatingand 2.5 min per dive in winter, i.e. a 2.5-fold increase for broodinggroups).The two-way ANOVAs were conducted over the similar divingdepths (Table 1). Figure2 illustrates the 80–270 m depth layer (270 m was the maximum dive depth found large increase in dive duration and bottom duration in all bird groups). between summer and winter for kingpenguinsdivingto 200 m. Results

Trip durations Vertical velocity in relation to dive depth

The trip durations (mean±SE) of kingpenguins Over all seasons the vertical velocity (ascent and descent (Aptenodytes patagonicus) at the different stages of rates) increased rapidly with increasingdive depth up to the annual cycle were 10.7±1.3 days (brooding, sum- 150 m, before increasingmuch more slowly or pla- mer; dates of departure at sea 12–28 February), teauingat deeper depths (Fig.1c, d). The season had no 12.8±1.1 days (incubation, summer; departures 22 influence on the relationship between rates of ascent or January–5 February), 56.6±11.7 days (autumn cre` che; descent and dive depth (nested two-way ANOVA ac- departures 15–21 March), 78±9.5 days (winter cre` che; countingfor variation due to dive depth, see Fig.1c, d). departures 11 June–11 August), and 25.3±1.1 days Duringthe central phase, the averageof vertical velocity –1 (spring; departures 5–8 October). Trip duration of birds over the 100–210 m layer was about 1.4 m s (Table 1). of the broodingand winter cre ` che group were longer Figure 2 shows that ascent and descent rates for a given than those of device-free conspecifics (6.8±0.7 and dive depth (200 m) were similar in summer and winter, 50.1±7.7 days, respectively, see Charrassin and Bost as opposed to dive durations and bottom durations. 2001). However, since we compared animals fitted with the same instruments over the year, the behavioural differences observed amonggroupsare due to the season Post-dive interval in relation to dive duration rather than to the equipment itself. and dive depth In all groups, the post-dive interval increased in relation Dive duration in relation to dive depth to dive duration and dive depth (Fig. 1e, f). Post-dive interval for a given dive duration or for a given depth In all groups the dive duration increased with dive depth decreased over the season (nested two-way ANOVA (Fig. 1a). The relationship between dive duration and accountingfor variation due to dive duration or due to dive depth calculated over the whole trip differed sig- dive depth, Fig. 1e, f). Post-dive intervals according to nificantly amongseasons (nested two-way ANOVA ac- dive durations (Fig. 1e) were similar during summer and countingfor variation due to dive depth, see Fig.1a), spring, and showed the highest values; they were the with dive duration for a given depth increasing with the shortest duringautumn and winter (SNK post hoc advancingseason. Dive durations were similar during tests). The same pattern was found when considering incubation (summer) and in spring, and were shortest at post-dive intervals accordingto dive depth (Fig.1f), these stages; dives were longest in winter, while they with values of each group being significantly different were of intermediate duration duringthe broodingand from the others, except for summer (incubation) and 584

Fig. 1a–g. Aptenodytes patago- nicus. Dive duration, bottom duration, descent rate, ascent rate, post-dive interval, and divingefficiency as a function of dive depth (a–d, f, g), and post- dive interval as a function of dive duration (e) for king penguins foraging in summer (n=6, incubatingstage; n=6, broodingstage),autumn and winter (n=5 and n=3, respec- tively, chick at the cre` che stage), and spring( n=4, post-moult stage). For each individual, 530 dives to <50 m and 230 dives to ‡50 m were analysed. For each parameter, values (means±SE) were averaged per group and per 10 m depth interval

springwhen post-dive intervals similar (SNK post hoc cre` che stages (SNK post hoc tests). During the central test). Duringthe central phase, averagesof the post- phase of the trip, the average diving efficiency increased dive intervals over the 100–210 layer decreased from from 0.15 to 0.18 in springand summer to 0.3 in winter 2.3 min in springto 1.6–1.8 min in autumn and winter (Table 1). (Table 1).

Vertical velocity in relation to bottom duration Divingefficiency in relation to dive depth For a given dive depth, the vertical velocity (averages of In all groups the diving efficiency (calculated for dives ascent and descent rates) was independent of the bottom to ‡50 m) decreased with increasingdive depth. Diving duration (P<0.05, Fig. 3a–c). For example, for dives to efficiency for a given dive depth increased over the sea- 200 m, both bottom durations of 1 min (summer) and son (nested two-way ANOVA accountingfor variation 3 min (winter) were associated with vertical velocities of due to dive depth, Fig. 1g). Diving efficiencies were 1.4 m s–1. The vertical velocity increased when dive similar, and lowest duringincubation (summer) and in depth increased from 100 to 280 m, and this occurred spring; they were highest in winter, while they were of irrespective of the time spent at the bottom of the dives intermediate value duringthe broodingand autumn (Fig. 3a–c). 585

Table 1. Aptenodytes patagonicus. Characteristics of the foraging methods’’). The effect of seasons on the dive parameters was tested behaviour of kingpenguinsin relation to dive depth duringdif- with a one-way ANOVA. Three birds did not dive deeper than ferent seasons of the annual cycle. Data (mean±SE) are shown for 210 m duringthe period considered; consequently, values are the central phase of the trips. For each individual, 57 dives to means per group over the 100–210 m layer [superscript a significant ‡50 m were selected at random over the period considered, after difference with all other groups; superscripts b and c paired signif- controllingfor temporal auto-correlation (see ‘‘Materials and icant differences (Student–Newman–Keuls post hoc test, P<0.05)] Foraging Summer Autumn Winter Spring Statistics parameters Cre` che stage Cre` che stage Post-moult IncubatingBrooding ( n=5) (n=3) (n=4) df F P (n=6) (n=6)

Foraging 161±13 192±9 167±12 204±16 183±11 Not tested depth (>70 m) (m)1,2 Dive duration 4.4±0.1 5.4±0.1 5.0±0.1 6.8±0.1a 5.0±0.1 4, 15 6.71 0.003 (min) Bottom duration 1.0±0.1b 1.3±0.1 1.4±0.1b 2.5±0.1a 1.0±0.1 4, 15 22.05 <0.001 (min) Descent rate (m s–1) 1.50±0.01 1.48±0.01 1.46±0.02 1.38±0.02 1.37±0.02 4, 15 0.32 0.86 Ascent rate (m s–1) 1.39±0.02 1.39±0.02 1.30±0.02 1.27±0.03 1.37±0.02 4, 15 0.52 0.73 Post-dive interval 1.8±0.1 1.9±0.1 1.6±0.1 1.8±0.1 2.3±0.17 4, 15 1.84 0.17 (min) Divingefficiency 0.17±0.01 b 1.18±0.01c 0.22±0.01b,c 0.29±0.01a 0.15±0.01 4, 15 18.78 <0.001

1From Charrassin and Bost (2001) 2For the winter cre` che stage, n=2 because one TDR had a 0–250 m range (so not tested for ANOVA)

Prey availability Discussion If we then consider that penguins feed substantially We show previously undescribed behavioural adjust- duringthe bottom phase of their dive, divingpenguins ments for individual dives accordingto seasons. We can be regarded as central-place foragers from the sur- found that kingpenguins( Aptenodytes patagonicus) face. In this case dive depth will represent the distance to varied the time spent at the bottom phase of their dives the food patch and bottom duration will reflect the time (at a given depth) over a two- to threefold range ac- spent feedingin the patch (Houston and Carbone 1992; cordingto season. Such a largevariation has not been Wilson et al. 1996). Central-place-foraging theory pre- reported for penguins or for mammalian marine pre- dicts that for a given distance to the food patch, birds dators. This is possibly because ours is one of the rare should remain longer in a patch of poor quality than in a detailed studies of a divingpredator under contrasting patch of good quality (Orians and Pearson 1979). Even at-sea conditions (Boyd et al. 1994; Kirkwood and if this model does not account for the fact that oxygen Robertson 1997b; McCafferty et al. 1998; Georges et al. stores limit the time available for a patch duringdiving 2000). It is generally hypothesised that penguins, like (see Houston and Carbone 1992), over our year-round most divingpredators, feed mostly at the bottom of study, extended bottom durations for dives to a given their dives. Although this might seem evident, it has depth are compatible with a reduced prey density in rarely been demonstrated directly (Schreer et al. 2001, autumn and winter. Conversely, short bottom durations but see Wilson and Wilson 1995). Recent studies based for dives to the same depth are compatible with a higher on the accurate detection of prey ingestions by prey density. Alternative to a lower prey density within a oesophagus temperature have shown that at least 40% given patch in winter, food may be rarer, more ephem- and 70% of feedingevents occur duringthe bottom eral, or more difficult to locate (Boyd 1996). These phase of the dives in kingand Ade ´ lie penguins, respec- patterns are in good agreement with the seasonal decline tively, with other ingestions mostly occurring during the of southern marine resources duringthe course of a year. ascent (Charrassin et al. 2001; Ropert-Coudert et al. King penguins feed mainly on the mesopelagic myc- 2001). A similar pattern was found by Wilson et al. tophid fish, a family of small-sized, schoolinglanternfish (2002) usingmandibular sensors to record beak opening (Cherel and Ridoux 1992). It is known that the main during diving in Magellanic penguins (Spheniscus summer prey of kingpenguin( Electrona carlsbergi at magellanicus). Therefore, even if bottom duration does Crozet Islands) migrate to great depth at the end of the not reflect the absolute value of the dive duration summer, probably to follow their own zooplankton prey devoted to feeding, it accounts for an important part which overwinter in deep waters (Koslov et al. 1991). In and is related to the total feedingtime. What is the addition, patch density seems reduced duringwinter, significance of our results in terms of prey availability when compared to the summer (Koslov et al. 1991). and behavioural adjustment? Consequently, even if birds also feed on other prey such 586 (Ropert-Coudert et al. 2001), longbottom durations in winter may reflect a longsearch time between consecu- tive captures in a patch of low density. Duringsummer, in contrast, dense prey patches may reduce the time necessary to feed (Orians and Pearson 1979; Boyd 1996). It is also possible that in winter the penguins might start to explore the water layer horizontally (duringthe bottom phase), even if a patch is not located during descent. The longbottom duration may thus reflect a longsearch for a prey patch in the horizontal direction, with or without prey capture. To our knowledge, no other penguin study has compared bottom durations duringdifferent seasons. In kingpenguins(present study) bottom periods of 80 s accounted for about 20–30% of the dive durations for dives to 100–200 m. In Ade´ lie penguins, bottom dura- tions of 24 s were found in summer for dives to 90 m (20% of the dive time, Chappell et al. 1993). In emperor penguins, summer values range from 1.5 min for dives to 150 m to 2.5 min for dives to 500 m (30% of the dive duration, Kooyman and Kooyman 1995), while gentoo penguins spend 48 s at the bottom during dives to 100 m (25% of the dive duration, Wilson et al. 1996). Hence, the proportion of time spent in the bottom phase of dives in summer is very similar across all four penguin species (20–30%), which probably reflects an optimisa- tion of time spent at the prey patch (Schreer et al. 2001). In light of our results on A. patagonicus, this suggests that penguins in general have a significant time margin for increasingtheir bottom duration when prey avail- ability is reduced in winter. Increasingbottom durations were not accompanied by increasingvertical velocities, which could have Fig. 2. Aptenodytes patagonicus. Typical profiles of dives to 200 m maintained short dive durations by reducingtransit time executed in summer and in winter by two kingpenguins,showing to and from the targeted depth. Accordingly, the dive the drastic increase of bottom duration and dive duration over the durations were longer in autumn and winter than in seasons (upper panel). This increase in winter occurs over the whole springand summer. Only two other penguinstudies range of dive depths, as shown by the bottom duration plotted against dive depth for the same birds (lower panel ) have compared dive durations at different seasons, in- cludingone for kingpenguins.Moore et al. (1999) found that dive duration of kingpenguinsat Heard Island also as squids in winter (Cherel et al. 1996), prey availability increases from summer to winter (but no details were for kingpenguinsis much reduced after summer. This is given on bottom duration or vertical velocities). Dive further confirmed by changes in other foraging param- durations of gentoo penguins in winter at South Georgia eters such as long-distance migrations to the pack-ice are also longer than in summer, with values for dives to area (Pu¨ tz et al. 1999; Charrassin and Bost 2001), ex- 100 m increasingfrom 2.5–2.8 min in summer to 2.9– tended foraging trips and deeper dives in autumn and 3.1 min in winter (Williams et al. 1992a,b). winter (Charrassin and Bost 2001). Thus, extended Our values of vertical velocity and the patterns of bottom durations in winter in contrast to short bottom change with depth during summer were similar to pre- durations in springand summer most likely reflect the vious studies of kingpenguins conducted in summer winter decrease in prey density. (Kooyman et al. 1992; Pu¨ tz et al. 1998), and did not vary accordingto season. Penguinsshow an optimum swimmingspeed at which the cost of transport is mini- Behavioural significance mal (2.2 m s–1 for kingpenguins, Culik et al. 1996), and, in nature, swim most of the time at this particular Our inter-seasonal comparative study presumes that speed (Kooyman et al. 1992; Ropert-Coudert et al. for different prey availabilities, the lower the prey 2000b). The vertical velocity depends on swimming density (myctophids in winter), the longer the bottom speed and dive angle by simple trigonometry (Wilson duration. If we consider that the bottom phase of a dive et al. 1996). Consequently, if swimmingspeed is kept starts once the penguin has located a prey patch constant, vertical velocity depends solely on the dive 587

Fig. 3. Aptenodytes patagonicus. Vertical velocities of kingpen- combined with increased bottom durations and constant guins (average of ascent and descent rates) in relation to bottom transit times, resulted in divingefficiencies that were duration for three categories of dive depth (100, 200 and 280 m). Each point corresponds to data for a bird at one of the different much higher in winter (0.26 at 200 m) than in spring and stages of the annual cycle. Data were taken during two periods of summer (0.11). In winter, kingpenguins minimised the the foraging trips so as to increase the sample size time surfacing, as expected if the proportion of time spent huntingand the divingefficiency have to be angle. Following this, our vertical velocity data suggest maximised. As for the prolonged bottom duration, this that penguins dive at a steeper angle with increasing dive is in agreement with the paucity of prey in winter. This depth. Steepeningdescent and ascent angleswith in- further confirms that the divingeffort in summer is not creasingdive depth will thus result in shorter transit at a maximum, due to the high prey availability at that times, as expected if dive durations are to be maintained time (Charrassin et al. 1999). Interestingly, shorter PDIs as short as possible. Since dive duration increases with for a given dive duration in winter also indicate that bottom duration, one may expect birds to reduce transit seasonal variations in physiological adaptations to div- ing might occur over the annual cycle. Short PDIs sug- time in winter so as to spare O2 for the prolonged bot- tom period. However, we found that the ascent rates gest that penguins may consume less oxygen during were kept constant over the different seasons, even if divingin winter, thereby prolongingdiveduration and prey were also found duringthe ascent (Charrassin et al. reducingrecovery time at the surface. The colder sea 2001; Ropert-Coudert et al. 2001). Why did penguins temperature in the winter foraging areas (1–2C) as not increase their vertical velocity when bottom duration compared to summer (4C, Charrassin and Bost 2001) increased? As suggested by Wilson et al. (1996) and seems to agree with reduced oxygen consumption Peters et al. (1998) scanningof the water layer duringthe through a deepest hypothermia during diving (see ascent and descent phases is probably essential for op- Handrich et al. 1997). However, the poor food condi- timal exploration of the water column, with the hori- tions in winter, probably forcingthe penguinsto dive at zontal scanningof the habitat beingfavoured by low a high rate, are certainly more important in explaining dive angles. For a penguin facing a decreasing prey the high diving efficiency than are the changes in thermal density, maintaininga low dive angleand thereby in- environment. Further work is required to examine the creasingthe probability of findingprey duringtransit physiological limits of king penguins (such as aerobic largely outweighs the potential cost of a prolonged dive dive limits) accordingto seasonal differences. duration. In conclusion, the time budget of king penguins on The post-dive interval (PDI) is considered the time the scale of a dive cycle appears most flexible duringthe necessary to physiologically recover from a dive and to bottom phase of dives and duringthe post-dive inter- prepare for the followingone (Chappell et al. 1993). vals. The ability to modulate the time spent at feeding Accordingly, for a given season, the PDI increases with depths is likely crucial for predators facingfluctuating increasingdive duration and dive depth, as reported in prey availability over their annual cycle. previous penguin studies (Chappell et al. 1993; Kooyman Acknowledgements The Institut Polaire Franc¸ ais (IPEV), The and Kooyman 1995; Cherel et al. 1999). However, our Terres Australes et Antarctiques Franc¸ aises (TAAF) and the Pro- study clearly shows that in kingpenguinsPDIs for a gramme Environnement du CNRS provided financial and logistical given dive duration are highly variable according to support. The Centre National de la Recherche Scientifique (CNRS) season, with PDIs beingshorter in winter and autumn and GDRE 1069 also helped fund this work. The Penguin Fund (Tokyo, Japan) has kindly provided J.B.C. with a grant. We thank (1.3 min for dives of 4–6 min) than in summer and Y. Ropert-Coudert, Y. Clerquin and N. Lambert for data collec- spring(2–3 min). PDIs for a givendepth were also tion in the field. We also thank the staffs of the 31st to 34th missions shorter in winter. The notable winter reduction of PDIs, to Crozet Islands. T. Zorn and G. Wittersheim offered statistical 588 advice. J. Lage and Y. Handrich aided extensively by writing cus- penguin population increase at Ile aux Cochons, Crozet tomised software for data analysis. We are also indebted to the Archipelago. Polar Biol 15:511–515 latter for constructive discussions on penguin energetics. M. En- Handrich Y, Bevan RM, Charrassin J-B, Butler PJ, Pu¨ tz K, stipp is thanked for helpingwith the English.We are particularly Woakes AJ, Lage J, Le Maho Y (1997) Hypothermia in for- grateful to R. Wilson for useful suggestions on an earlier version of aging king penguins. Nature 388:64–67 the manuscript. Two anonymous reviewers provided valuable Houston AI, Carbone C (1992) The optimal allocation of time comments on the manuscript. The experimental procedures com- duringthe divingcycle. Behav Ecol 3:255–265 plied with the current laws of French authorities (Ethics Committee Kirkwood R, Robertson G (1997a) The foraging ecology of female of the French Polar Institute). emperor penguins in winter. Ecol Monogr 67:155–176 Kirkwood R, Robertson G (1997b) Seasonal change in the foraging ecology of emperor penguins on the Mawson Coast, Antarctica. Mar Ecol ProgSer 156:205–223 References Kooyman GL, Kooyman TG (1995) Divingbehaviour of emperor penguins nurturing chicks at Coulman Island, Antarctica. Bannasch R, Wilson RP, Culik B (1994) Hydrodynamic aspects of Condor 97:536–549 design and attachment of a back-mounted device in penguins. Kooyman GL, Cherel Y, Le Maho Y, Croxall JP, Thorson PH, J Exp Biol 194:83–96 Ridoux V, Kooyman CA (1992) Divingbehaviour and ener- Bost CA, Georges JY, Guinet C, Cherel Y, Pu¨ tz K, Charrassin JB, getics during foraging cycle in king penguins. Ecol Monogr Handrich Y, Zorn T, Lage J, Le Maho Y (1997) Foraging 62:143–163 habitat and food intake of satellite-tracked kingpenguins Koslov AN, Shust KV, Zemsky AV (1991) Seasonal and inter- duringthe austral summer at Crozet Archipelago.Mar Ecol annual variability in the distribution of Electrona carlsbergi in ProgSer 150:21–33 the Southern Polar Front area. In: Selected scientific papers, Boyd IL (1996) Temporal scales of foraging in a marine predator. 1990. Committee Conservation Antarctic LivingResources Ecology 77:426–434 (CCAMLR), Hobart, pp 320–337 Boyd IL, Arnould JPY, Barton T, Croxall JP (1994) Foraging McCafferty DJ, Boyd IL, Walker TR, Taylor RI (1998) Foraging behaviour of Antarctic fur seals duringperiods of contrasting responses of Antarctic fur seals to changes in the marine envi- prey abundance. J Anim Ecol 63:703–713 ronment. Mar Ecol ProgSer 166:285–299 Chappell MA, Shoemaker VH, Janes DN, Bucher TL, Maloney Moore GJ, Wienecke B, Robertson G (1999) Seasonal change in SK (1993) Divingbehaviour duringforagingin breedingAde ´ lie foraging area and dive depths of breeding king penguins at penguins. Ecology 74:1204–1215 Heard Island. Polar Biol 21:376–384 Charrassin JB, Bost CA (2001) Utilization of the oceanic habitat by Orians GH, Pearson NE (1979) On the theory of the central place kingpenguinsover the annual cycle. Mar Ecol ProgSer foraging. In: Horn JD, Mitchell R, Stair GR (eds) Analysis of 221:285–297 ecological systems. Ohio State University Press, Columbus, Charrassin JB, Bost CA, Pu¨ tz K, Lage L, Dahier T, Zorn T, Le pp 154–177 Maho Y (1998) Foraging strategies of incubating and brooding Peters G, Wilson RP, Scolaro JA, Laurenti S, Upton J, Galleli H kingpenguins Aptenodytes patagonicus. Oecologia 114:194–201 (1998) The diving behaviour of magellanic penguins at Punta Charrassin JB, Bost CA, Pu¨ tz K, Lage J, Dahier T, Le Maho Y Norte, Peninsula Valde´ s, Argentina. Colon Waterbirds 21:1–10 (1999) Changes in depth utilization in relation to the breeding Pu¨ tz K, Wilson RP, Charrassin J-B, Raclot T, Lage J, Le Maho Y, stage: a case study with the Aptenodytes patago- Kierspel MAM, Culik BM, AdelungD (1998) Foragingstrat- nicus. In: Proceedings of the 3rd international penguin confer- egy of king penguins (Aptenodytes patagonicus) duringsummer ence. Mar Ornithol 27:37–41 at the Crozet Islands. Ecology 79:1905–1921 Charrassin JB, Kato A, Handrich Y, Sato K, Naito Y, Ancel A, Pu¨ tz K, Ropert-Coudert Y, Charrassin J-B, Wilson RP (1999) Bost CA, Gauthier-Clerc M, Ropert-Coudert Y, Le Maho Y Foraging areas of king penguins (Aptenodytes patagonicus) (2001) Feeding behaviour of free-ranging penguins determined breedingat Possession Island in the southern Indian Ocean. by oesophageal temperature. Proc R Soc Lond Ser B Biol Sci Mar Ornithol 27:77–84 268:151–157 Ridoux V (1994) The diets and dietary segregation of seabirds at Cherel Y, Ridoux V (1992) Prey species and nutritive value of food the Crozet Islands. Mar Ornithol 22:1–192 fed duringsummer to kingpenguin Aptenodytes patagonica Rodary D, Bonneau W, Le Maho Y, Bost CA (2000) Benthic chicks at Possession Island, Crozet Archipelago. Ibis 134:118– divingin male emperor penguins Aptenodytes forsteri foraging 127 in winter. Mar Ecol ProgSer 207:171–181 Cherel Y, Ridoux V, Rodhouse PG (1996) Fish and squid in the Ropert-Coudert Y, Bost C-A, Bevan RM, Handrich Y, Le Maho diet of kingpenguinchicks, Aptenodytes patagonicus, during Y, Woakes AJ, Butler PJ (2000a) Impact of externally-attached winter at sub-antarctic Crozet Islands. Mar Biol 126:559–570 logger on the diving behaviour of the king penguin (Apteno- Cherel Y, Tremblay Y, Guinard E, Georges JY (1999) Diving be- dytes patagonicus). Physiol Biochem Zool 74:438–444 haviour of female northern rockhopper penguins, Eudyptes Ropert-Coudert Y, Sato K, Kato A, Charrassin JB, Bost CA, Le chrysocome moseleyi, duringthe broodingperiod at Amsterdam Maho Y, Naito Y (2000b) Prey pursuit and capturingbeha- Island (southern Indian Ocean). Mar Biol 134:375–385 viour in deep divers estimated by multiple data recorders. Polar Clarke A (1988) Seasonality in the Antarctic marine environment. Biosci 13:101–112 Comp Biochem Physiol A 90:461–473 Ropert-Coudert Y, Kato A, Baudat J, Bost CA, Le Maho Y, Naito Culik BM, Pu¨ tz K, Wilson RP, Allers D, Lage J, Bost CA, Le Y (2001) Feeding strategies of free-ranging Ade´ lie penguins Maho Y (1996) Diving energetics in king penguins (Aptenodytes Pygoscelis adeliae analysed by multiple data recording. Polar patagonicus). J Exp Biol 199:973–983 Biol 24:460–466 Davis LS, Boersma PD, Court GS (1996) Satellite telemetry of the Schreer JF, Kovacs KM, O’Hara Hines RJ (2001) Comparative winter migration of Ade´ lie penguins (Pygoscelis adeliae). Polar divingpatterns of pinnipeds and seabirds. Ecol Monogr71:137– Biol 16:221–225 162 Foxton P (1956) The distribution of the standingcrop of zoop- Smith SL, Schnack-Schiel SB (1990) Polar zooplankton. In: Smith lancton in the Southern Ocean. Discov Rep 28:191–236 Jr WO (ed) Polar oceanography, part B: chemistry, biology, Georges JY, Tremblay Y, Guinet C (2000) Seasonal diving beha- and geology. Academic, San Diego, pp 527–598 viour in lactatingsubantarctic fur seals on Amsterdam Island. Williams TD, Briggs DR, Croxall JP, Naito Y, Kato A (1992a) Polar Biol 23:59–69 Diving pattern and performance in relation to foraging ecology Guinet C, Jouventin P, Malacamp J (1995) Satellite remote sensing in the , Pygoscelis papua. J Zool Lond 227:211– in monitoring change of seabirds: use of Spot Image in king 230 589

Williams TD, Kato A, Croxall JP, Naito Y, Briggs DR, Rodwell S, Wilson RP, Culik BM, Kosiorek P, AdelungD (1998) The over- Barton TR (1992b) Divingpattern and performance in non- winter movements of a (Pygoscelis antarcti- breeding gentoo penguins (Pygoscelis papua) duringwinter. ca). Polar Rec 34:111–116 Auk 109:223–234 Wilson RP, Steinfurth A, Ropert-Coudert Y, Kato A, Kurita M Wilson RP (1995) Foraging ecology. In: Williams TD (ed) The (2002) Lip-readingin remote subjects: an attempt to quantify penguins Spheniscidae. Oxford University Press, Oxford, and separate ingestion, breathing and vocalisation in free-living pp 81–106 animals usingpenguinsas a model. Mar Biol 140:17–27 Wilson RP, Wilson MPT (1995) The foraging behaviour of the Woehler EJ (1995) Consumption of Southern Ocean marine re- Spheniscus demersus. In: Dann P, Norman I, sources by penguins. In: Dann P, Norman I, Reilly P (eds) The Reilly P (eds) The penguins, ecology and management. Surrey penguins, ecology and management. Surrey Beatty, Chipping Beatty, ChippingNorton, pp 244–265 Norton, pp 267–291 Wilson RP, Culik BM, Peters G, Bannasch R (1996) Divingbe- YdenbergRC, Clark CW (1989) Aerobiosis and anaerobiosis haviour of gentoo penguins, Pygoscelis papua; factors keeping during diving by western grebes: an optimal foraging approach. dive profiles in shape. Mar Biol 126:153–162 J Theor Biol 139:437–449