The adaptive significance of crèches in the king penguin

Céline Le Bohec; Michel Gauthier-Clerc; Yvon Le Maho

The adaptive significance of crèches in the king penguin

Animal Behaviour, 2005

The adaptive signiﬁcance of cre `ches in the king penguin CE ´ LINE LE BOHEC*, MICHEL GAUTHIER-CLERC† & YVON LE MAHO* *Centre d’E ´ cologie et Physiologie E ´ nerge ´tiques-CNRS yStation Biologique de la Tour du Valat (Received 14 January 2004; initial acceptance 6 March 2004; ﬁnal acceptance 10 November 2004; published online 15 July 2005; MS. number: 7963R) Cre `ching behaviour in penguins is deﬁned as the rearing of chicks by their own parents in large ﬂocks called ‘cre `ches’. Although several hypotheses have been proposed to account for the behaviour, the factors inducing chicks to aggregate remain relatively poorly understood, in particular for colonial seabirds. We studied cre `ching behaviour in the king penguin, Aptenodytes patagonicus, by looking at the dynamics of cre `che formation and possible costs and beneﬁts associated with this strategy. Cre `ches increased in size but declined in number throughout the austral winter. They were located preferentially in the central parts of the colony.Lone chicks suﬀered the most aggression from unrelated adults, whereas chicks in a cre `che suﬀered the least. Chicks attacked by unrelated adults preferentially joined a cre `che. Adult aggression appeared to be a major factor inducing cre `ching behaviour. Chicks at the periphery of a cre `che were more vigilant while sleeping,as measured by eye openings. Cre `ches seemed to occasion intense competition among chicks for access to the centre. Chicks in poor condition were attacked and pushed to the periphery of the cre `che, where they were preyed on by giantpetrels.During harsh weather conditions,chicks amalgamated into larger cre `ches, tolerated lower interindividual distances and turned their backs to the wind and rain. Our results accord with the idea that cre `ching behaviour in king penguins is a strategy that protects chicks from adult aggression, predation and severe weather. Ó 2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. Cre `ching behaviour is a rearing strategy observed in several colonial species ( Gorman & Milne 1972). Coloniality, the aggregation of conspeciﬁc individuals on breeding territo- ries distinct from foraging sites (Kharitonov & Siegel- Causey 1988),appears to supply many beneﬁts such as optimal breeding habitatselection (Danchin & Wagner 1997) and reduction of predation (Wittenberger & Hunt 1985;Siegel-Causey & Kharitonov 1991). However,there are several obvious disadvantages of colonial breeding, such as having to forage away from the colony, needing a parent– oﬀspring recognition system and having to protect young against conspeciﬁc adult aggression (Wittenberger & Hunt 1985; Kharitonov & Siegel-Causey 1988). Cre `ching appears to be a partial substitute for continuous parental protection and care,permitting both parents to leave theiryoung temporarily and go to foraging areas (Evans 1984; Besnard 2001). Several adaptive advantages have been proposed for cre `ching behaviour, including reduced predation (Pettingill 1960; Davis 1982; Tourenq et al. 1995), increased thermo- regulation efﬁciency (Pettingill 1960;Yeates 1975; Davis 1982; Evans 1984; Carter & Hobson 1988; Tourenq et al. 1995) and improved social tolerance (Bildstein 1993). Some authors have suggested that intraspeciﬁc aggression could be the main proximate cause of cre `che formation (Seddon & van Heezik 1993; Tourenq et al. 1995; Besnard 2001). However,there is little agreementon why chicks form cre `ches, mainly because the behaviour is so variable between species. The term ‘cre `che’ was ﬁrst used to describe cases of chick amalgamation in the emperor penguin, Aptenodytes forsteri (Wilson 1907). Although this concept of cre `che applies speciﬁcally to birds (Brown & Root 1971; Gorman & Milne 1972), a few authors have used itto describe oﬀspring gatherings in mammals such as Mexican free-tailed bats, Tadarida brasiliensis mexicana (McCracken 1984), harbour seals, Phoca vitulina richardsi (Slater & Markowitz 1983), giraﬀes,Giraﬀa camelopardalis (Leuthold 1979), Nubian ibexes, Capra ibex nubiana (Levy & Bernadsky 1991) and sable antelopes, Hippotragus niger (Thompson 1998). The king penguin, Aptenodytes patagonicus, is a prime model for studying cre `ching behaviour. Its breeding cycle is unusual for seabirdsbecauseit exceedsa year and the chick ﬂedging period lasts about 11 months (Barrat 1976). During the austral winter, chicks left alone in the colony are subject to prolonged starvation (up to 5 months) Correspondence: Ce ´ line Le Bohec, CEPE-CNRS, 23 rue Becquerel, 67087 StrasbourgCedex 02, France(email: celine.lebohec@c-strasbourg. fr). MichelGauthier-Clerc is at the Station Biologique de la Tour du Valat, Le Sambuc, 13200 Arles, France. 527 0003–3472/04/$30.00/0 Ó 2005 The Association for the Study of Animal Behaviour.Published by Elsevier Ltd. All rights reserved. ANIMAL BEHAVIOUR, 2005, 70, 527–538 doi:10.1016/j.anbehav.2004.11.012

between the infrequentfeeding visits by their parents (Cherel et al. 1987; Descamps etal. 2002). Thus, large metabolic reserves and cre `ching behaviour are important for the chicks’ survival(Barrat 1976;Cherelet al. 1987). King penguin cre `ches are aggregations of unrelated chicks, in which the chicks continue to be fed only by their own parents.Cre `ching behaviour in king penguins has been mentioned by some authors (e.g. Barrat 1976;Descamps et al. 2002; Le Bohec et al. 2003), but has not been examined in detail. Our aim in this study was to describe cre `ching by king penguin chicks over the annual cycle for various habitats within a colony. Since king penguins are very aggressive to conspeciﬁcs(Le Maho et al. 1993; Challet et al. 1994;Coˆte ´ 2000),we tested the eﬀectof aggression on cre `ching. Finally, we investigated whether cre `ches protect against predation and inclement weather. METHODS Study Species The king penguin is a subantarctic seabird that breeds in large dense colonies. The pair incubates a single egg directly on their feet. During incubation and brooding periods, the birds vigorously defend a small territory (approximately 0.5–0.8 m 2 ) against breeding neighbours and intruders that approach within pecking distance (Co ˆte ´ 2000). The breeding cycle lasts 14–16 months. Two peaks of laying occur:a ﬁrst peak of early breeders (at the beginning of December) that failed their previous breeding season and a second peak of breeders (at the end of January) that bred late because they ﬂedged a chick during the previous breeding season (Barrat 1976). Chick rearing can be split into four periods: (1) brooding of young chicks by their parents;(2) formation of small prewintering cre `ches while chicks wait for their parents to return from feeding trips (Descampset al. 2002); (3) formation of large wintering cre `ches when feeding visits become infrequent (Descampset al. 2002) and the majority of the adults have deserted the colony; (4) at the end of winter, progressive breaking up ofcre `ches when adults return to the colony to moult and start breeding again and the chicks moult before going to sea. Data Collection We conducted the study in the breeding colony La Grande Manchotie `re (around 16 000 pairs) on Possession Island, Crozet Archipelago (46 25 0 S, 51 45 0 E) during a complete breeding cycle (2001–2002). For each aspect of the study we collected data during similar time slots and under similar weather conditions to minimize bias. Cre ` che formation and dynamics A cre `che has been deﬁned as two or more chicks in close association (Evans 1984; Carter & Hobson 1988),where individuals are less than two chick wing lengths apart (i.e. less than 60 cm in king penguins). From February 2001 to February 2002, we determined cre `che numbers and sizes (chicks per cre `che) in two designated parts of the colony (Fig. 1): Zone A (beach and river, 0.6 ha) and Zone B (side of the valley, 0.8 ha). The high frequency of observations conducted in Zone A (twice a week until May then once a week when cre `che sizes varied less) allowed us to follow in detail the progressive formation of cre `ches during the annual cycle. To investigate whether chicks gather prefer- entially in particular areas,we divided Zone B into 60 squares(each of about 10 ! 10 m) which we observed once a week until May and then twice a month. We deﬁned four categories of habitat in Zone B: Central zone: nonﬂoodable centralareas;Floodable zone:areas poten- tially ﬂoodable at the bottom of the slope; Peripheral zone: peripheral areas at the top of the slope; Rocky zone: rocky faced areas. The average number of chicks per cre `che per square and the average number of cre `ches per square were determined for each category. We made these observations for the three cre `ching phases of the annual cycle: prewintering (February to May), wintering (June to October) and postwintering (November to February). Intraspeciﬁc aggressiveness To record aggressivebehaviours,we used the focal animal and continuous sampling method (Altmann 1974; Martin & Bateson 1993).We quantiﬁed agonistic interactions (number of bill strokes, hits and misses) between an individualselected randomly in the colony and conspeciﬁcsduring 10-min observation periods.A total of 720 focal observations were conducted in Zone A from February to November 2001. Adult–adult aggressive interactions. From 3 to 13 February, we counted the acts of aggression on neighbours by adults at diﬀerent breeding stages: incubating (N Z 40),brood- ing a 1-week-old chick (N Z 40) and brooding a 3-week- old chick (N Z 40). We estimated chick size by the chick’s height relative to that of its parent to deﬁne the following two categories: (1) approximate height of 10%, 1-week-old grey chick (with no down, hence dependent on its parent) and (2) approximate heightof 30%, 3-week-old brown chick (covered by down). Adult–chick aggressive interactions. From about 20 days, when they can thermoregulate (Barre ´ 1984), chicks are left alone between feeding visits.At this age chicks either remain alone or approach other adults or chicks. From 15 February to 10 March 2001, we recorded the number of pecks that chicks received from breeding adults and other chicks (guarded by an adult and/or in a cre `che; N Z 120). Chicks guarded by parents were used as a control situation (N Z 120). We used the method of all occurrence behav- iour sampling ( Altmann 1974) to record lone chicks starting to move, (2) duration and number of pecks received from adults during the movement, and (3) the chicks’ destination. Chick–chick aggressive interactions. Agonistic interactions between chicks in good and poor condition were observed at the beginning of cre `che formation (March) and at the end of the wintering cre `ches (September). Chicks in good ANIMAL BEHAVIOUR, 70, 3 528

ANIMAL BEHAVIOUR, 2005, 70, 527–538 doi:10.1016/j.anbehav.2004.11.012 The adaptive signiﬁcance of crèches in the king penguin C ÉLINE LE BOHEC*, M IC HEL G AUTH I ER- CLERC † & Y VON LE MAH O* *Centre d’Écologie et Physiologie Énergétiques-CNRS yStation Biologique de la Tour du Valat (Received 14 January 2004; initial acceptance 6 March 2004; final acceptance 10 November 2004; published online 15 July 2005; MS. number: 7963R) Crèching behaviour in penguins is deﬁned as the rearing of chicks by their own parents in large ﬂocks called ‘crèches’. Although several hypotheses have been proposed to account for the behaviour, the factors inducing chicks to aggregate remain relatively poorly understood, in particular for colonial seabirds. We studied crèching behaviour in the king penguin, Aptenodytes patagonicus, by looking at the dynamics of crèche formation and possible costs and beneﬁts associated with this strategy. Crèches increased in size but declined in number throughout the austral winter. They were located preferentially in the central parts of the colony. Lone chicks suffered the most aggression from unrelated adults, whereas chicks in a crèche suffered the least. Chicks attacked by unrelated adults preferentially joined a crèche. Adult aggression appeared to be a major factor inducing crèching behaviour. Chicks at the periphery of a crèche were more vigilant while sleeping, as measured by eye openings. Crèches seemed to occasion intense competition among chicks for access to the centre. Chicks in poor condition were attacked and pushed to the periphery of the crèche, where they were preyed on by giant petrels. During harsh weather conditions, chicks amalgamated into larger crèches, tolerated lower interindividual distances and turned their backs to the wind and rain. Our results accord with the idea that crèching behaviour in king penguins is a strategy that protects chicks from adult aggression, predation and severe weather. Ó 2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. Crèching behaviour is a rearing strategy observed in several colonial species (Gorman & Milne 1972). Coloniality, the aggregation of conspeciﬁc individuals on breeding territories distinct from foraging sites (Kharitonov & SiegelCausey 1988), appears to supply many beneﬁts such as optimal breeding habitat selection (Danchin & Wagner 1997) and reduction of predation (Wittenberger & Hunt 1985; Siegel-Causey & Kharitonov 1991). However, there are several obvious disadvantages of colonial breeding, such as having to forage away from the colony, needing a parent– offspring recognition system and having to protect young against conspeciﬁc adult aggression (Wittenberger & Hunt 1985; Kharitonov & Siegel-Causey 1988). Crèching appears to be a partial substitute for continuous parental protection and care, permitting both parents to leave their young temporarily and go to foraging areas (Evans 1984; Besnard 2001). Several adaptive advantages have been proposed for crèching behaviour, including reduced predation (Pettingill 1960; Davis 1982; Tourenq et al. 1995), increased thermoregulation efﬁciency (Pettingill 1960; Yeates 1975; Davis Correspondence: Céline Le Bohec, CEPE-CNRS, 23 rue Becquerel, 67087 Strasbourg Cedex 02, France (email: celine.lebohec@c-strasbourg. fr). Michel Gauthier-Clerc is at the Station Biologique de la Tour du Valat, Le Sambuc, 13200 Arles, France. 0003–3472/04/$30.00/0 1982; Evans 1984; Carter & Hobson 1988; Tourenq et al. 1995) and improved social tolerance (Bildstein 1993). Some authors have suggested that intraspeciﬁc aggression could be the main proximate cause of crèche formation (Seddon & van Heezik 1993; Tourenq et al. 1995; Besnard 2001). However, there is little agreement on why chicks form crèches, mainly because the behaviour is so variable between species. The term ‘crèche’ was ﬁrst used to describe cases of chick amalgamation in the emperor penguin, Aptenodytes forsteri (Wilson 1907). Although this concept of crèche applies speciﬁcally to birds (Brown & Root 1971; Gorman & Milne 1972), a few authors have used it to describe offspring gatherings in mammals such as Mexican free-tailed bats, Tadarida brasiliensis mexicana (McCracken 1984), harbour seals, Phoca vitulina richardsi (Slater & Markowitz 1983), giraffes, Giraffa camelopardalis (Leuthold 1979), Nubian ibexes, Capra ibex nubiana (Levy & Bernadsky 1991) and sable antelopes, Hippotragus niger (Thompson 1998). The king penguin, Aptenodytes patagonicus, is a prime model for studying crèching behaviour. Its breeding cycle is unusual for seabirds because it exceeds a year and the chick ﬂedging period lasts about 11 months (Barrat 1976). During the austral winter, chicks left alone in the colony are subject to prolonged starvation (up to 5 months) 527 Ó 2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. 528 ANIMAL BEHAVIOUR, 70, 3 between the infrequent feeding visits by their parents (Cherel et al. 1987; Descamps et al. 2002). Thus, large metabolic reserves and crèching behaviour are important for the chicks’ survival (Barrat 1976; Cherel et al. 1987). King penguin crèches are aggregations of unrelated chicks, in which the chicks continue to be fed only by their own parents. Crèching behaviour in king penguins has been mentioned by some authors (e.g. Barrat 1976; Descamps et al. 2002; Le Bohec et al. 2003), but has not been examined in detail. Our aim in this study was to describe crèching by king penguin chicks over the annual cycle for various habitats within a colony. Since king penguins are very aggressive to conspeciﬁcs (Le Maho et al. 1993; Challet et al. 1994; Côté 2000), we tested the effect of aggression on crèching. Finally, we investigated whether crèches protect against predation and inclement weather. METHODS (Fig. 1): Zone A (beach and river, 0.6 ha) and Zone B (side of the valley, 0.8 ha). The high frequency of observations conducted in Zone A (twice a week until May then once a week when crèche sizes varied less) allowed us to follow in detail the progressive formation of crèches during the annual cycle. To investigate whether chicks gather preferentially in particular areas, we divided Zone B into 60 squares (each of about 10 ! 10 m) which we observed once a week until May and then twice a month. We deﬁned four categories of habitat in Zone B: Central zone: nonﬂoodable central areas; Floodable zone: areas potentially ﬂoodable at the bottom of the slope; Peripheral zone: peripheral areas at the top of the slope; Rocky zone: rocky faced areas. The average number of chicks per crèche per square and the average number of crèches per square were determined for each category. We made these observations for the three crèching phases of the annual cycle: prewintering (February to May), wintering (June to October) and postwintering (November to February). Study Species The king penguin is a subantarctic seabird that breeds in large dense colonies. The pair incubates a single egg directly on their feet. During incubation and brooding periods, the birds vigorously defend a small territory (approximately 0.5–0.8 m2) against breeding neighbours and intruders that approach within pecking distance (Côté 2000). The breeding cycle lasts 14–16 months. Two peaks of laying occur: a ﬁrst peak of early breeders (at the beginning of December) that failed their previous breeding season and a second peak of breeders (at the end of January) that bred late because they ﬂedged a chick during the previous breeding season (Barrat 1976). Chick rearing can be split into four periods: (1) brooding of young chicks by their parents; (2) formation of small prewintering crèches while chicks wait for their parents to return from feeding trips (Descamps et al. 2002); (3) formation of large wintering crèches when feeding visits become infrequent (Descamps et al. 2002) and the majority of the adults have deserted the colony; (4) at the end of winter, progressive breaking up of crèches when adults return to the colony to moult and start breeding again and the chicks moult before going to sea. Data Collection We conducted the study in the breeding colony La Grande Manchotière (around 16 000 pairs) on Possession Island, Crozet Archipelago (46 250 S, 51 450 E) during a complete breeding cycle (2001–2002). For each aspect of the study we collected data during similar time slots and under similar weather conditions to minimize bias. Crèche formation and dynamics A crèche has been deﬁned as two or more chicks in close association (Evans 1984; Carter & Hobson 1988), where individuals are less than two chick wing lengths apart (i.e. less than 60 cm in king penguins). From February 2001 to February 2002, we determined crèche numbers and sizes (chicks per crèche) in two designated parts of the colony Intraspecific aggressiveness To record aggressive behaviours, we used the focal animal and continuous sampling method (Altmann 1974; Martin & Bateson 1993). We quantiﬁed agonistic interactions (number of bill strokes, hits and misses) between an individual selected randomly in the colony and conspeciﬁcs during 10-min observation periods. A total of 720 focal observations were conducted in Zone A from February to November 2001. Adult–adult aggressive interactions. From 3 to 13 February, we counted the acts of aggression on neighbours by adults at different breeding stages: incubating (N Z 40), brooding a 1-week-old chick (N Z 40) and brooding a 3-weekold chick (N Z 40). We estimated chick size by the chick’s height relative to that of its parent to deﬁne the following two categories: (1) approximate height of 10%, 1-week-old grey chick (with no down, hence dependent on its parent) and (2) approximate height of 30%, 3-week-old brown chick (covered by down). Adult–chick aggressive interactions. From about 20 days, when they can thermoregulate (Barré 1984), chicks are left alone between feeding visits. At this age chicks either remain alone or approach other adults or chicks. From 15 February to 10 March 2001, we recorded the number of pecks that chicks received from breeding adults and other chicks (guarded by an adult and/or in a crèche; N Z 120). Chicks guarded by parents were used as a control situation (N Z 120). We used the method of all occurrence behaviour sampling (Altmann 1974) to record lone chicks starting to move, (2) duration and number of pecks received from adults during the movement, and (3) the chicks’ destination. Chick–chick aggressive interactions. Agonistic interactions between chicks in good and poor condition were observed at the beginning of crèche formation (March) and at the end of the wintering crèches (September). Chicks in good LE BOHEC ET AL.: CRÈCHING IN KING PENGUINS Scale 0 10 20 30 m Central zone Floodable zone Peripheral zone Zone B La Baie du Marin Rocky zone d roa Technical area Zone A Zone C Camp River Figure 1. Schematic map of the breeding colony La Baie du Marin showing the various study areas. Dashes: boundary of the breeding colony La Grande Manchotière. Zone A (beach and river, blue outline), for study of crèche formation dynamics during the annual cycle; Zone B (side of the valley, red outline), for study of the habitat; Zone C (beach and river, green outline), for study of the effect of weather conditions. condition were the tallest and heaviest individuals in the crèche: 70–80% (March, N Z 20) or 90% (September, N Z 20) of an adult’s height, with an invisible breastbone and protruding abdomen. Chicks in poor condition were the smallest and thinnest individuals in the crèche: 30– 50% (March, N Z 20) or 60% (September, N Z 20) of an adult’s height, with a prominent breastbone and abdomen not protruding. We also took focal samples of chicks fed by an adult and chicks in a crèche from March to November to compare their aggressiveness (N Z 260). between eye openings for randomly selected individuals (Gauthier-Clerc et al. 1998, 2000). These data were collected for birds on the periphery (N Z 30) and in the centre (N Z 30) of crèches adopting the typical sleep posture, i.e. standing up, head lying on back, with the visible eye closed and bill tucked underneath a ﬂipper (Challet et al. 1994; Dewasmes et al. 1989). This part of the study was conducted in Zone A in May (beginning of wintering crèches), at the end of June (middle of wintering crèches) and at the end of August (end of wintering crèches). Body mass of chicks and position in the crèche We deﬁned the crèche periphery as that part formed by the ﬁrst two rows of chicks on the outside of the group. Chicks on the periphery (N Z 25) and in the centre (N Z 25) of crèches in Zone A were caught by hand and weighed with an electronic balance (G10g) monthly from March to October to establish a correlation between chick position in the crèche and body condition. Vigilance and sleep We collected vigilance and sleep data using the focal animal and continuous sampling technique (Altmann 1974; Martin & Bateson 1993). Observations lasted 2 min. Using a 40! telescope, a tape recorder and a stopwatch, we recorded the frequency of eye openings, the duration of consecutive eye openings and intervals Predation We recorded instances of predation (N Z 157) according to the method of all occurrence behaviour sampling (Altmann 1974). In all zones we recorded the type of predator (giant petrels Macronectes halli and M. giganteus, brown skua, Catharacta lonnbergi, kelp gull, Larus dominicanus, or lesser sheathbill, Chionis minor), chick size (small, medium or tall, depending on its height relative to the adult), chick body condition (see above), and predation outcome: success or failure. Weather conditions We investigated crèching in relation to weather in Zone C (beach and river, 0.3 ha, Fig. 1): (1) winter, Cold)Wind)Rain (N Z 8 days, %5 C, wind R 28 knots, 529 530 ANIMAL BEHAVIOUR, 70, 3 with rain); (2) winter, Cold)Wind)No rain (N Z 12 days, %5 C, wind R 28 knots, without rain); (3) winter, Cold)No wind)No rain (N Z 10 days, %5 C, no wind, no rain); and (4) summer, Warm)No wind)No rain (N Z 8 days, R 10 C, wind 0–6 knots, without rain). Variables recorded were number and size of crèches, interindividual distance (N Z 100) estimated in units of chick ﬂipper length, and position of chicks in good condition (N Z 20) and poor condition (N Z 20) within the crèche. Data Analysis Results are reported as means G SE. We assume that our random samplings did not include signiﬁcant replication because there were more than 20 000 king penguin chicks in the colony, and the same bird was unlikely to be observed more than once. When the data were normally distributed and homoscedasticity of data was conﬁrmed, we compared samples using one-way or two-way ANOVAs followed by a parametric post hoc test adjusted by Tukey (parametric tests, Scherrer 1984). When application conditions for ANOVA were not satisﬁed (even after transformation), we used nonparametric tests (Sheirer– Ray–Hare test (two-way analysis of variance by ranks) or Kruskal–Wallis test (one-way analysis of variance by ranks), Siegel & Castellan 1988) to compare more than two samples. These tests were followed by a nonparametric post hoc test adjusted by Dunn (samples of different size). To compare means of two independent groups and to analyse frequencies we used the Mann–Whitney U test and the chi-square test, respectively. For correlations between variables we used Spearman rank and Kendall rank partial correlation coefﬁcients. For the statistical analyses we used SYSTAT 9.0 and Sigmastat 2.0 (SPSS Inc., Chicago, IL, U.S.A.). All tests were two tailed with signiﬁcance level set at a Z 0.05. chicks increased progressively until the end of June (to about 6500 chicks in Zone A). At the same time, the number of crèches with up to 20 chicks increased until 6 April, when it reached a maximum (357 in Zone A; Fig. 2). The number of crèches was highly correlated with the number of chicks left unattended by their parents (Spearman rank correlation: rS Z 0.90, N Z 17, P ! 0.005). Crèche size was negatively related to the number of crèches (Kendall rank partial correlation coefﬁcient: rDE Z ÿ0.25, N Z 60, P ! 0.005) and size increased progressively from February to September (Fig. 2). The number of crèches started to decrease at the beginning of April. This decrease resulted from the grouping of several small crèches (less than 50 chicks) into larger ones. The minimum number of crèches (8) and the maximum size (about 500 chicks) were observed between 13 August and 24 October. At the end of October, crèches split up and became more numerous but smaller in size. Location in the colony and period of the year had a signiﬁcant effect on the number of chicks in crèches per square (two-way ANOVA: zone: F3,141 Z 15.66, P ! 0.005; period: F2,142 Z 22.11, P ! 0.005; interaction: F6,133 Z 3.51, P Z 0.001; Fig. 3a) and the number of crèches per square (two-way ANOVA: zone: F3,141 Z 16.77, P ! 0.005; period: F2,142 Z 81.75, P ! 0.005; interaction: F6,133 Z 4.73, P ! 0.005; Fig. 3b). During the prewintering and wintering periods, the number of chicks in crèches and the number of crèches were signiﬁcantly higher in the central zone than in the other three locations (Tukey tests: P ! 0.005). There were no signiﬁcant differences between the four locations during the postwintering period (Tukey tests: NS). These were signiﬁcantly more prewintering than wintering and postwintering crèches in the ﬂoodable and peripheral zones (Tukey tests: P ! 0.05) and more postwintering than wintering crèches in the ﬂoodable and rocky zones (Tukey tests: P ! 0.05). Intraspecific Aggression Ethical Note Observations were made 10–100 m from outside the colony. At short distances, we observed the birds from behind a low wall or from a blind. This type of observation did not cause any disturbance nor did it expose birds to predators. During monthly weighings we minimized disturbance by moving slowly towards the crèches and releasing captured chicks close to the crèche from where they came. We veriﬁed that the crèches reformed immediately after the capture. No predation was observed during that time. Observed and weighed birds were selected randomly in the colony and were not marked. The study received the consent of the Ethics Committee of the Institut Polaire Français – Paul-Emile Victor. RESULTS Number and Size of Crèches The ﬁrst chicks abandoned temporarily were seen during the week 9–15 February. The number of such The level of adult aggression was on average very high whatever their reproductive status (23 G 2 agonistic interactions per 10 min, N Z 120). Adult aggressive behaviour varied signiﬁcantly with reproductive status (ANOVA: F2,117 Z 3.59, P Z 0.031) and was highest for adults brooding a 3-week-old chick (30 G 3 interactions, N Z 40; Tukey tests: P ! 0.05). However, there was no signiﬁcant difference between adults incubating and adults brooding a 1-week-old chick (19 G 2 and 20 G 2 interactions, respectively, N Z 40 for both cases, Tukey test: NS). Aggression towards a chick by adults and by other chicks (guarded by an adult and/or in a crèche) differed in the four scenarios (Kruskal–Wallis tests: from adults: H3 Z 139.26, N Z 240, P ! 0.001; from chicks: H3 Z 91.76, N Z 240, P ! 0.001; Fig. 4). A chick with one of its parents experienced the fewest pecks in both cases (0.14 G 0.06 and 0.21 G 0.07, respectively; Dunn tests: P ! 0.05). Attacks by adults on a lone chick (20.98 G 2.99) were about four times higher than aggression towards a chick with an unrelated adult (4.90 G 1.30; Dunn test: P ! 0.05) LE BOHEC ET AL.: CRÈCHING IN KING PENGUINS 750 400 600 450 200 300 Crèche size Number of crèches 300 100 150 0 Feb Mar 0 Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb 2001 2002 Figure 2. Change in number (histogram) and size (curve) of crèches in Zone A. Crèche sizes (number of chicks per crèche) are reported as means C SE. or towards a chick in a crèche (3.23 G 0.54; Dunn test: P ! 0.05). Conversely, other chicks pecked a lone chick less (1.30 G 0.72) than a chick with an unrelated adult (3.25 G 1.43; Dunn test: P ! 0.05) or a chick in a crèche (3.63 G 0.46; Dunn test: P ! 0.05). Lone unguarded chicks mainly moved away in response to adult aggression (72%). During the displacement (which lasted on average 33 G 5 s), a chick received on average 1 peck/s. Chicks usually moved to crèches and stayed in them (61%). Interindividual Relations within the Crèche Within a crèche, chicks in good condition were significantly more aggressive than chicks in poor condition at the beginning of crèche formation in March and at the end of winter in September (Mann–Whitney U tests: March: U Z 342, N1 Z N2 Z 20, P ! 0.005; September: U Z 328, N1 Z N2 Z 20, P ! 0.005; Fig. 5). Chicks in poor condition received signiﬁcantly more pecks than chicks in good condition, in March and in September (March: U Z 73, N1 Z N2 Z 20, P Z 0.001; September: U Z 66.50, N1 Z N2 Z 20, P ! 0.005). Chicks in crèches became signiﬁcantly less aggressive between March and September (Fig. 5). The presence of a feeder adult and period of the year had a signiﬁcant effect on the agonistic interactions between chicks (two-way Sheirer–Ray–Hare tests: effect on the number of pecks given: adult present: F1,258 Z 119.71, P ! 0.005; period: F4,255 Z 15.234, P ! 0.005; interaction: F4,250 Z 19.70, P ! 0.005; effect on the number of pecks received: adult present: F1,258 Z 52.19, P ! 0.005; period: F4,255 Z 17.782, P ! 0.005; interaction: F4,250 Z 13.76, P Z 0.005). A chick fed by an adult was more aggressive than a lone chick in a crèche (Dunn tests: P ! 0.05; Fig. 6) except in November (Dunn test: P ! 0.05). Lone chicks in crèches were least aggressive from May to August. Predation Chick body mass was signiﬁcantly different according to position within the crèche and month (two-way ANOVA: position: F1,389 Z 51.13, P ! 0.005; month: F7,383 Z 30.40, P ! 0.005; interaction: F7,375 Z 4,82, P ! 0.005; Fig. 7). Throughout the wintering period, chicks situated on the periphery of crèches weighed less than chicks in a central position. The time spent with an eye open in a typical sleep posture corresponds to both the vigilance level and the time spent awake. Position within the crèche and month had a signiﬁcant effect on the proportion of time spent with one eye open (two-way Sheirer–Ray–Hare test: position: F1,177 Z 368.48, P ! 0.001; month: F2,176 Z 9.51, P ! 0.001; interaction: F1,177 Z 11.33, P Z 0.001), the frequency of eye openings (two-way Sheirer–Ray–Hare test: position: F1,177 Z 368.48, P ! 0.001; month: F2,176 Z 9.51, P ! 0.001; interaction: F1,177 Z 11.33, P Z 0.001) and the duration of eye closure (two-way Sheirer– Ray–Hare test: position: F1,177 Z 372.68, P ! 0.001; month: F2,176 Z 9.52, P ! 0.001; interaction: F1,177 Z 11.95, P ! 0.001). The proportion of time spent with one eye open and frequency of eye openings were significantly higher among peripheral chicks than central chicks whatever the month (Dunn tests: P ! 0.05; Table 1). The duration of eye closure was signiﬁcantly higher among central chicks than peripheral chicks whatever the month (Dunn tests: P ! 0.05). Time spent with one eye open and frequency of eye openings increased from May to August in the periphery, but were minimal in June in the central position. The duration of eye closure decreased from May 531 180 Prewintering Wintering Postwintering (a) 160 140 120 100 80 60 40 20 Aggressive behaviour per 10 min (number of pecks) Number of chicks in crèches per square ANIMAL BEHAVIOUR, 70, 3 10 8 6 4 2 0 Poor Good (N = 20) (N = 20) Poor Good (N = 20) (N = 20) March September 8 Figure 5. Aggressive behaviour between chicks in a crèche according to body condition. Values are means C SE. Mann–Whitney U tests comparing March and September data: number of pecks received by chicks in good condition: U Z 338, N1 Z N2 Z 20, P ! 0.001; chicks in poor condition: U Z 343, N1 Z N2 Z 20, P ! 0.001; number of pecks by chicks in good condition: U Z 335, N1 Z N2 Z 20, P ! 0.001; chicks in poor condition: U Z 287, N1 Z N2 Z 20, P Z 0.004. 6 4 2 Central Floodable Peripheral Area of Zone B Rocky Figure 3. (a) Number of chicks in crèches per square and (b) number of crèches per square by zone in the colony (Central zone: nonfloodable central area; Floodable zone: areas potentially floodable at the bottom of the slope; Peripheral zone: peripheral areas at the top of the slope; Rocky zone: rocky areas) and period (prewintering crèches: February to May; wintering crèches: June to October; postwintering crèches: November to February). Values are means C SE. 100 Received from adults Received from chicks 10 b c e b e f 1 a d 0.1 0.01 With parent (N = 120) With unrelated adult (N = 40) to August in the periphery, but was maximal in June in the central position. Out of 157 instances of predation observed on chicks, giant petrels were the most frequent predators with 95% of the attacks. Brown skuas, kelp gulls and lesser sheathbills committed only 1% of the attacks and were generally scavengers and detritivores, feeding notably on carcasses abandoned by giant petrels. The remaining 4% corresponded to attacks by mixed groups of these four predators. Small chicks were the object of 77% of the attacks (Table 2). Attacks on chicks in poor condition and/or already weakened by previous injuries represented 43% of the instances of predation and generally ended with predator success (97%). In contrast, attacks on chicks in Alone (N = 40) In crèche (N = 40) Figure 4. Aggressive behaviour shown by adults and by other chicks to a chick with one of its parents, a chick with an unrelated adult, a lone chick and a chick in a crèche. Values are means C SE. Values not assigned the same letter (aggression from adult: a,b,c; aggression from chicks: d,e,f) are significantly different (post hoc test adjusted by Dunn: P ! 0.05). Aggressive behaviour per 10 min (number of pecks) Number of crèches per square Received from chicks By chicks 12 (b) Aggressive behaviour per 10 min log (number of pecks) 532 c 20 Chick alone in crèche Chick fed by an adult * 15 * d 10 * * * d d a a e 5 b 0 b b Mar May Jun Aug Nov (N = 60) (N = 50) (N = 50) (N = 50) (N = 50) Figure 6. Aggression shown by a chick fed by an adult or by a lone chick in a crèche during the austral winter. Values are means C SE. Values not assigned the same letter (lone chick: a,b; fed chick: c,d,e) are significantly different (post hoc test adjusted by Dunn). Comparison of lone and fed chicks: *P ! 0.05, post hoc test adjusted by Dunn. LE BOHEC ET AL.: CRÈCHING IN KING PENGUINS 10 9 Body mass (kg) up to half a ﬂipper apart) than when it was Warm)No wind)No rain (25% of chicks; Tukey tests: P ! 0.005). On the other hand, crèches were clearly looser when weather conditions were better (30% of chicks three ﬂippers or more apart versus 6% during Cold weather conditions, Tukey tests: P ! 0.001). During harsh weather conditions (Cold)Wind)Rain and Cold)Wind)No rain), the distribution within the crèche of chicks in good and poor condition was not uniform (ANOVAs: good: F3,28 Z 15.84, P ! 0.001; poor: F3,40 Z 50.07, P ! 0.001). Fewer chicks were in poor condition in the central position than on the periphery (4.68 G 0.76 versus 15.32 G 0.76 chicks; Tukey tests: P ! 0.001) and more chicks were in good condition in the centre (13.37 G 0.51 chicks; Tukey test: P ! 0.001). Fewer chicks were in good condition on the periphery (6.63 G 0.51 chicks) than in the central position (Tukey test: P ! 0.005). In contrast, when it was Cold)No wind)No rain and Warm)No wind)No rain, chicks had a uniform distribution within the crèche for both body conditions (10 G 0.53 chicks for each category; ANOVAs: good: F3,36 Z 2.35, P Z 0.088; poor: F3,28 Z 0.07, P Z 0.975). Peripheral chicks Central chicks * 8 * * * 7 6 5 (N M = ar 40 ) (N Ap = r 50 ) (N M = ay 50 (N Ju ) = n 51 ) (N Ju = l 50 ) (N Au = g 50 (N Se ) = p 50 ) (N O = ct 50 ) 4 Figure 7. Chick body mass and distribution in crèches (peripheral or central position) during the austral winter. Values are means G SE. Comparison of central and peripheral chicks: *P ! 0.05, post hoc test adjusted by Tukey. good condition (57% of attacks) succeeded only 24% of the time. DISCUSSION Weather Conditions Dynamics of Crèche Formation Weather conditions had a signiﬁcant effect on number of crèches (ANOVA: F3,34 Z 11.85, P ! 0.001) and size of crèches (ANOVA: F3,34 Z 6.33, P Z 0.002; Fig. 8). The number of crèches was signiﬁcantly lower in Cold)Wind) Rain and Cold)Wind)No rain conditions than in Warm)No wind)No rain (Tukey tests: P % 0.001). There were also fewer crèches during Cold)Wind)Rain than in Cold)No wind)No rain (Tukey test: P Z 0.005). Crèche size was markedly higher in Cold)Wind)Rain than in Warm)No wind)No rain or Cold)No wind)No rain (Tukey tests: P Z 0.009 and P Z 0.003, respectively). Weather conditions had a signiﬁcant effect on distance between chicks (ANOVAs: chicks half a ﬂipper apart: F3,34 Z 12.42, P ! 0.001; chicks one ﬂipper apart: F3,34 Z 2.94, P Z 0.047; chicks two ﬂippers apart: F3,34 Z 8.69, P Z 0.007; chicks three ﬂippers or more apart: F3,34 Z 16.93, P ! 0.001). Interindividual distance was lower during the Cold weather conditions (nearly 70% of chicks Crèche size gradually increased from February to September, whereas the number of crèches dropped from the beginning of April as a result of small crèches grouping together into larger ones. The process of gradual amalgamation ﬁrst into numerous small prewintering crèches can be explained as the consequence of hatching asynchrony, which generates an asynchrony of temporary chick desertions and the formation of crèches constituted of chicks of all ages. At the end of the summer, the colony is deserted by adults, freeing space for chicks to gather into fewer and larger crèches. White ibis, Eudocimus albus, and greater ﬂamingo, Phoenicopterus ruber, chicks also gather ﬁrst in small then in larger groups and ﬁnally into a single crèche (Dinep 1988; Tourenq et al. 1995). Evans (1984) considered that the temporary desertion of young by parents is the single most important factor triggering the onset of crèching, whereas the return of parents to the Table 1. Proportion of time spent with one eye open, frequency of eye openings and duration of eye closure by position in the crèche and month Periphery % Time spent with one eye open Frequency of eye openings (per 2 min) Duration of eye closure (s) Values are means G SE. Centre May June August May Sample size 30 30 29 14.7G2.9 10.2G1.8 31.6G7.7 33.5G4.9 18.8G2.2 8.3G2.1 38.6G4.3 26.0G2.4 3.7G0.5 June August 30 30 30 0.7G0.3 1.0G0.3 89.4G7.6 0.5G0.2 0.6G0.2 96.4G6.4 0.8G0.3 0.9G0.3 92.0G7.1 533 ANIMAL BEHAVIOUR, 70, 3 Table 2. Success of predation by giant petrels on chicks according to chick height (estimated in relation to adult height) and chick condition (estimated from breastbone and abdomen prominence) Chick size/condition Attacks (%) Predation success (%) Small Poor Good 77 51 49 97 34 Medium Poor Good 17 22 78 100 5 Tall Poor Good 6 0 100 d 10 N Z 149 predation events. colony stimulates young to leave a crèche. Davis (1982) showed that crèching behaviour is closely associated with the number of adults present in the colony and suggested that the crèche may serve as an alternative means of defence against predators when too few adults are present in the colony to deter predators effectively. This hypothesis does not appear to be valid for king penguins, however, because, ﬁrst, they do not cooperate to deter predators and, second, they are aggressive towards each other during the breeding period. Crèches are consequently more likely to be a partial substitute for continuous parental care (Besnard 2001) and dependent on how much space is available for their formation. Habitat Quality We found that chicks congregated mostly in the central areas of the colony, as do pelicans and ﬂamingos (Brown & Root 1971; Tourenq et al. 1995). In other species such as terns and gulls, crèches are set up on the edge of the colony (Buckley & Buckley 1976). In king penguins, central parts of the colony are initially occupied by early 30 25 600 20 400 15 10 Crèche size Number of crèches 200 5 0 0 C ol d * (N Ra* W = in ind 8) C ol * d (N No * W = ra in C 12) in d ol d * * (N No No = rai wi W 10) n nd ar m *N * (N o No = rain wi 8) nd 534 Figure 8. Number (histogram) and size (curve) of crèches for four categories of weather conditions (Zone C, see Methods for description of categories). Values are means C SE. breeders (Côté 2000). Individuals that lay ﬁrst also leave their chick ﬁrst once it can thermoregulate efﬁciently. Spaces free of adults are then created in those central areas of the colony, allowing chicks to gather together and form numerous small prewintering crèches. As the summer wears on, chicks born later in peripheral and potentially ﬂoodable zones are temporarily abandoned by their parents and the number of small crèches increases in these areas. From June to October, the great majority of adults desert the colony. The whole area of the colony then becomes accessible to chicks. However, chick distribution was not uniform on the space available, and rocky areas, peripheral areas at the top of the slope and potentially ﬂoodable areas at the bottom of the slope remained unoccupied. This suggests that wintering crèche locations of king penguins depend on habitat quality. Central areas are considered high-quality territories in colonial species (Kharitonov & Siegel-Causey 1988; Vinuela et al. 1995). Carter & Hobson (1988) suggested that the location of crèches of chicks of Brandt’s cormorant, Phalacrocorax penicillatus, depends on habitat quality. Levy & Bernadsky (1991) noted that Nubian ibex, Capra ibex nubiana, formed crèches on shady even terrain. In our study, crèches broke up progressively in the spring when adults returned for moulting and courting. As for prewintering crèches, available space again became a restrictive factor because adults occupied most of the colony area. Chicks were pushed towards peripheral areas of the colony, probably of lower quality, since they had avoided these areas before the adults returned. These peripheral areas are known to be infested by Ixodes uriae ticks, which is a parasite of the king penguin (Gauthier-Clerc et al. 1999; Mangin et al. 2003). The parasitic constraint hypothesis could explain the nonoccupation of peripheral areas by the chicks when central areas were still available. Adult Aggression Breeders were aggressive to alien chicks, especially lone unguarded chicks. Attacked chicks preferentially joined crèches where they experienced less aggression and where the risk of injury was lower compared with pecks and ﬂipper blows from adults. Other studies have also shown a high level of aggression between breeders (Le Maho et al. 1993; Challet et al. 1994; Côté 2000). Since ﬁghts entail high energetic costs (Högstad 1987), the beneﬁts of defending a territory should therefore be high in terms of reproductive success (Montgomerie & Weatherhead 1988). Intraspeciﬁc aggression in colonial species varies during the reproductive cycle (Burger & Gochfeld 1990; Lamey 1993; Challet et al. 1994). The increase in aggression between adult king penguins from incubation to brooding may be explained by the higher ﬁtness value of a chick, as proposed by parental investment theory (Williams 1966; Trivers 1972; Burger 1981; Siegel-Causey & Hunt 1981). Adelie penguins, Pygoscelis adeliae (Spurr 1974) and chinstrap penguins, Pygoscelis antarctica (Vinuela et al. 1995; Amat et al. 1996) similarly defend chicks more strongly than eggs. LE BOHEC ET AL.: CRÈCHING IN KING PENGUINS Chicks temporarily abandoned by their parents experienced the highest level of aggression because of the strong territoriality of incubating and brooding adults. Intraspeciﬁc aggression towards such chicks has been frequently reported in colonial species (Wittenberger & Hunt 1985) and attack by unrelated adults is generally recognized as one of the major causes of chick mortality in gulls (Pierotti 1988; Brown & Morris 1995). In our study, aggression by adults resulted in chicks leaving their natal territory in 72% of cases and joining a crèche in 61% of cases. Seddon & van Heezik (1993), who obtained similar results for the jackass penguin, Spheniscus demersus (moving towards a crèche: 74%), suggested that intraspeciﬁc aggression is the main proximate cause of crèche formation. In March, up to 10% of adult king penguins guard two or three chicks and will adopt chicks, at least temporarily. A lone unguarded chick heads for an unrelated adult in 24% of movements from the natal territory, tries to take the place of the legitimate chick and attempts to chase it away by pecking. Then the parent attacks both chicks without distinction until its own chick vocalizes. If the parent is unable to recognize its own chick quickly, the risk of injuring it is high. The conﬂict of interest between an adult and an alien chick may result in a forced adoption that would then reduce the care for the legitimate chick and would increase its risk of rejection and injury. This may explain why a parent defends its territory against alien chicks. A crèche might then offer the most advantages to unguarded chicks in terms of less aggression and more safety than close proximity to an unrelated adult. Indeed, our data show that the level of aggression between chicks in a crèche is low. In contrast, when a chick is joined by a parent, it leaves the centre of a crèche and becomes more aggressive towards other chicks. This behaviour can manifest itself as pushing away foreign chicks that might steal meals (Boersma & Davis 1997). Protection from Predators The central position within a crèche allows a chick to reduce its vigilance and increase time spent sleeping. One of the functions of gregarious behaviour is to reduce predation risk (Pulliam 1973; Powell 1974; Caraco et al. 1980). Individuals in a group can reduce the time spent in vigilance against predators by taking advantage of vigilance by other group members, without reducing the probability of detecting the predator or increasing an individual’s risk of capture by a predator (Elgar & Catterall 1981). A reduction in individual vigilance with an increase in group size has been reported for numerous birds, mammals and ﬁsh (Lendrem 1984; Martella et al. 1995; Gauthier-Clerc et al. 1998). This group size effect has been explained by the ‘many eyes hypothesis’, i.e. a collective detection that increases with group size (Dimond & Lazarus 1974). King penguin chicks should thus have a two-fold advantage of being in large crèches: an increased likelihood of predator detection (detection effect) allowing an individual to devote less time to vigilance and hence more time to sleep, and a reduction in risk for any given individual (dilution effect, Dehn 1990). Larger crèches suffer less predation, probably because the proportion of chicks at the periphery of the crèche declines quickly as the crèche gets larger (Hamilton 1971). This may be an inﬂuential factor in the tendency for crèche size to increase over the wintering crèche period. Sleep and vigilance are mutually exclusive. The trade-off between these two activities may vary according to an individual’s position in the group (Elgar 1989). Individuals at the periphery are more vigilant than those at the centre because they are at greater predation risk. They are the ones that will be encountered ﬁrst by an approaching ground predator (Hamilton 1971; Jennings & Evans 1980; Petit & Bildstein 1987). Our results corroborate this hypothesis. When a predator approached a crèche, peripheral chicks extended themselves up to their full height and bumped into central chicks which were alerted to danger and then all the chicks gathered in a denser ﬂock. We also noticed changes in sleep and vigilance during the winter. Vigilance of central chicks was slightly lower in the middle of the winter. June corresponded to the longest average durations of eye closure, the shortest proportion of time spent with one eye open, and the lowest frequency of eye openings. Conversely, sleep decreased and vigilance increased throughout the wintering period for peripheral chicks. In particular, chicks undergo a long period of fasting during the austral winter (the coldest period of the year, around ÿ5 C). One of the adaptations allowing birds to tolerate the fasting and cold lies in their ability to reduce energy expenditure. Several authors have suggested that sleep is important for energy conservation because it decreases body temperature and thermoregulatory costs (Stahel et al. 1984; Berger & Philipps 1993; Criscuolo et al. 2001). Chicks in the centre of crèches (in a safe microenvironment and probably thermally more stable) can drop their vigilance in favour of sleep in midwinter unlike those on the periphery, which are exposed to predation as well as the cold, and sleep less. Studies on pigeons, Columbia livia, green-winged teals, Anas crecca, and little penguins, Eudyptula minor, also reported a decrease in the time spent sleeping when birds were subjected to cold (Stahel et al. 1984; Graf et al. 1987; Tamisier & Dehorter 1999). According to Pulliam et al. (1974) both group size and vigilance behaviour are inﬂuenced by ambient temperature. We found that chicks in poor condition experienced the most aggression and were pushed towards the periphery of a crèche. These chicks suffered the highest predation by giant petrels. Southern and northern giant petrels are considered the main king penguin predators, in particular for chicks during the winter (Hunter & Brooke 1992; Le Bohec et al. 2003). Our results show that 77% of attacks were directed towards the smallest chicks and 43% towards chicks in poor condition and/or already weakened by previous injuries, among which were 51% of small chicks. Predation attempts on small chicks and chicks in poor condition were generally successful (66% and 97%, respectively), in contrast to attacks on mediumsized and large chicks in good condition (6% success). Thus, a better body condition increases the chances of chick survival, not only from the standpoint of resistance 535 536 ANIMAL BEHAVIOUR, 70, 3 to starvation, but also considering protection from predators. Our data reveal an effect of chick body condition on intraspeciﬁc relation in the crèche. Chicks in good condition were more aggressive than chicks in poor condition. This dominance status seems to give them access to the protected area of the crèche centre. Chicks in poor condition, and consequently less able to defend themselves, were pushed to the periphery. This rejection of chicks in poor condition to the periphery of crèches was most evident during harsh weather. Protection from Harsh Weather During inclement weather (cold and rain and/or wind), there were fewer crèches, but these were larger and chicks were closer together. Choosing a thermally favourable environment is an integral part of a chick’s thermoregulatory ability (Whittow 1976). The crèche probably works as a hygrometrical and thermally stable microenvironment (Pettingill 1960). This environment may allow chicks to decrease their energy expenditure and therefore to increase survival probability. Our results corroborate this hypothesis. Indeed, as winter progresses, weather conditions become less favourable and king penguin chicks amalgamated into bigger but fewer crèches. In the rockhopper penguin, Eudyptes chrysocome, variations in crèche size were related to ﬂuctuations in air temperature (Pettingill 1960). According to Yeates (1975), harsh weather may result in crèche formation in Eudyptidae and Pygoscelidae chicks. In contrast, Davis (1982) noted that crèching in the Adélie penguin did not vary consistently with ﬂuctuations in climatic variables. Chick rearing in this species occurs during the summer and lasts only a month. Owing to the relatively mild conditions during the short period when crèching occurs, Adélie chicks can probably maintain a constant body temperature. On the other hand, contact behaviour may be most pronounced when ambient temperature is lowest and wind speed and relative humidity are highest. This contact behaviour apparently has a thermoregulatory function (Davis 1982; Evans 1984). The chill factor associated with high wind speed dramatically increases heat loss, and this effect is further accentuated by high relative humidity (Davis 1982). Tight elongated crèches could reduce the individual convection surfaces to a single collective surface, which would reduce individual heat loss and thereby energy expenditure linked to thermogenesis during extreme climatic conditions (Taylor 1962; Evans 1984). This probably implies that chicks are in competition for access to the most sheltered areas in the group. This life in compact groups, as illustrated by the huddles of emperor penguins, could be essential for survival and reproductive success (Ancel et al. 1997). At lower temperatures, dark-eyed juncos Junco hymelis (Caraco 1979), and willow tits, Parus montanus (Högstad 1988), form larger ﬂocks of birds with little aggression. The drop in aggression observed among king penguin chicks during winter might be explained by the need to lower energy expenditure in harsh weather and to establish group cohesion against predation. This social tolerance strategy, by reducing the overall intraspeciﬁc aggression within the crèche, may therefore have emerged in response to environmental constraints. This study has allowed us to describe the genesis and functioning of king penguin crèches during the annual cycle and to stress the preferential occupation of the high-quality areas of the colony by crèches. Parental aggression towards unguarded alien chicks appears to be an important factor leading to chicks joining a crèche. We suggest that crèching behaviour has adaptive advantages such as protection against predation and severe weather. Food dispersion and lack of protection against predators and severe weather, induced by the open environment characteristic of subantarctic islands, may be selection pressures that promote the development of this chick-rearing strategy. Further research should look into the costs associated with this strategy, such as the increase in risk of disease and parasite transmission by the close contact between individuals and food theft (i.e. kleptoparasitism). Finally, to test the hypothesis that crèches confer important energy savings, as has already been shown in emperor penguin huddles (Ancel et al. 1997), the energy expenditure of chicks should be measured. Acknowledgments This work was supported by the Institut Polaire Français – Paul-Emile Victor (Programme 137) and by the project Zones Ateliers of the Programme Environnement Vie et Société of the CNRS. We are grateful to C. Gilbert, D. Grémillet, A. Lescroël, C. Salmon, S. Samtmann, A. Schmidt and C. Villemin for constructive comments and the anonymous referees for greatly improving the manuscript. We thank C. Salmon for his help in preparing software for analysis of the data sets. References Altmann, J. 1974. Observational study of behavior: sampling methods. Behaviour, 49, 227–267. Amat, J. A., Carrascal, L. M. & Moreno, J. 1996. Nest defence by chinstrap penguins Pygoscelis antarctica in relation to offspring number and age. Journal of Avian Biology, 27, 177–179. Ancel, A., Visser, H., Handrich, Y., Masman, D. & Le Maho, Y. 1997. Energy saving in huddling penguins. Nature, 385, 304–305. Barrat, A. 1976. Quelques aspects de la biologie et de l’écologie du manchot royal (Aptenodytes patagonicus) des ı̂les Crozet. Commission Nationale Française de Recherche Antarctique, 40, 9–51. Barré, H. 1984. Metabolic and insulative changes in winter and summer acclimatized king penguin chicks. Journal of Comparative Physiology, 154, 317–324. Berger, R. J. & Philipps, N. H. 1993. Sleep and energy conservation. Neural Information Processing Systems, 8, 276–281. Besnard, A. 2001. Evolution de l’élevage des poussins en crèche chez les Laridés. Ph.D. thesis, Université de Montpellier. Bildstein, K. L. 1993. White Ibis: Wetland Wanderer. Washington: Smithsonian Institute Press. Boersma, P. D. & Davis, L. S. 1997. Feeding chases and food allocation in Adélie penguins, Pygoscelis adeliae. Animal Behaviour, 54, 1047–1052. LE BOHEC ET AL.: CRÈCHING IN KING PENGUINS Brown, K. M. & Morris, R. D. 1995. Investigator disturbance, chick movement and aggressive behavior in ring-billed gulls. Wilson Bulletin, 107, 140–152. Brown, L. H. & Root, A. 1971. The breeding behaviour of the lesser flamingo Phoeniconaias minor. Ibis, 113, 147–172. Buckley, P. A. & Buckley, F. G. 1976. Late-blooming terns. Natural History, 84, 46–56. Burger, J. 1981. Aggressive behaviour of black skimmers (Rynchops niger). Behaviour, 76, 207–222. Burger, J. & Gochfeld, M. 1990. The Black Skimmer: Social Dynamics of a Colonial Species. New York: Columbia University Press. Caraco, T. 1979. Time budgeting and group size: a test of theory. Ecology, 60, 618–627. Caraco, T., Martindale, S. & Pulliam, H. R. 1980. Avian flocking in the presence of a predator. Nature, 285, 400–401. Carter, H. R. & Hobson, K. A. 1988. Creching behavior of Brandt’s cormorant chicks. Condor, 90, 395–400. Challet, E., Bost, C.-A., Handrich, Y., Gendner, J.-P. & Le Maho, Y. 1994. Behavioural time budget of breeding king penguins (Aptenodytes patagonicus). Journal of Zoology, 233, 669–681. Cherel, Y., Stahl, J. C. & Le Maho, Y. 1987. Ecology and physiology of fasting king penguin chicks. Auk, 104, 254–262. Côté, S. D. 2000. Aggressiveness in king penguins in relation to reproductive status and territory location. Animal Behaviour, 59, 813–821. Criscuolo, F., Gauthier-Clerc, M., Le Maho, Y., Zorn, T. & Gabrielsen, G. W. 2001. Sleep changes during long-term fasting of the incubating common eider Somateria mollissima. Ardea, 89, 15–22. Danchin, E. & Wagner, R. H. 1997. The evolution of coloniality: the emergence of new perspectives. Trends in Evolution and Ecology, 12, 342–347. Davis, L. S. 1982. Creching behaviour of Adélie penguin chicks (Pygoscelis adeliae). New Zealand Journal of Zoology, 9, 279–286. Dehn, M. M. 1990. Vigilance for predators: detection and dilution effects. Behavioral Ecology and Sociobiology, 26, 337–342. Descamps, S., Gauthier-Clerc, M., Gendner, J.-P. & Le Maho, Y. 2002. The annual breeding cycle of unbanded king penguins Aptenodytes patagonicus on Possession Island (Crozet). Avian Science, 2, 87–98. Dewasmes, G., Buchet, C., Geloen, A. & Le Maho, Y. 1989. Sleep changes in emperor penguins during fasting. American Journal of Physiology, 256, R476–R480. Dimond, S. & Lazarus, J. 1974. The problem of vigilance in animal life. Brain, Behavior and Evolution, 9, 60–79. Dinep, A. 1988. Social behavior, parental care, and creching: a chronology of development in white ibis chicks. Colonial Waterbird Society Newsletter, 12, 42. Elgar, M. A. 1989. Predator vigilance and group size in mammals and birds: a critical review of the empirical evidence. Biology Review, 64, 13–33. Elgar, M. A. & Catterall, C. 1981. Flocking and predator surveillance in house sparrows: test of an hypothesis. Animal Behaviour, 29, 868–872. Evans, R. M. 1984. Some causal and functional correlates of creching in young white pelicans. Canadian Journal of Zoology, 62, 814–819. Gauthier-Clerc, M., Tamisier, A. & Cézilly, F. 1998. Sleep-vigilance trade-off in green-winged teals (Anas crecca crecca). Canadian Journal of Zoology, 76, 2214–2218. Gauthier-Clerc, M., Jaulhac, B., Frenot, Y., Bachelard, C., Monteil, H., Le Maho, Y. & Handrich, Y. 1999. Prevalence of Borrelia burgdorferi (the Lyme disease agent) antibodies in king penguin Aptenodytes patagonicus in Crozet Archipelago. Polar Biology, 22, 141–143. Gauthier-Clerc, M., Tamisier, A. & Cézilly, F. 2000. Sleep-vigilance trade-off in gadwall during winter period. Condor, 102, 307–313. Gorman, M. L. & Milne, H. 1972. Creche behaviour in the common eider Somateria m. mollissima L. Ornis Scandinavica, 3, 21–25. Graf, R., Heller, H. C., Sakaguchi, S. & Krishna, S. 1987. Influence of spinal and hypothalamic warming on metabolism and sleep in pigeons. American Journal of Physiology, 252, R661–R667. Hamilton, W. J. 1971. Geometry for selfish herd. Journal of Theoretical Biology, 31, 295–311. Högstad, O. 1987. It is expensive to be dominant. Auk, 104, 333– 336. Högstad, O. 1988. Social rank and antipredator behaviour of willow tits Parus montanus in winter flocks. Ibis, 130, 45–56. Hunter, S. & Brooke, M. D. L. 1992. The diet of giant petrels, Macronectes spp. at Marion Island, Southern Indian Ocean. Colonial Waterbirds, 15, 56–65. Jennings, T. & Evans, S. M. 1980. Influence of position in the flock and flock size on vigilance in the starling, Sturnus vulgaris. Animal Behaviour, 28, 634–635. Kharitonov, S. P. & Siegel-Causey, D. 1988. Colony formation in seabirds. Current Ornithology, 5, 223–272. Lamey, T. C. 1993. Territorial aggression, timing of egg loss, and egg-size differences in rockhopper penguins, Eudyptes c. chrysocome, on New Island, Falkland Islands. Oikos, 66, 293–297. Le Bohec, C., Gauthier-Clerc, M., Gendner, J.-P., Chatelain, N. & Le Maho, Y. 2003. Nocturnal predation of king penguins by giant petrels on Crozet Islands. Polar Biology, 26, 587–590. Le Maho, Y., Gendner, J.-P., Challet, E., Bost, C.-A., Gilles, J., Verdon, C., Plumeré, C., Robin, J.-P. & Handrich, Y. 1993. Undisturbed breeding penguins as indicators of changes in marine resources. Marine Ecology Progress Series, 95, 1–6. Lendrem, D. W. 1984. Sleeping and vigilance in birds. II. An experimental study of the Barbary dove (Streptopelia risoria). Animal Behaviour, 32, 243–248. Leuthold, B. M. 1979. Social organization and behaviour of giraffe in Tsavo East National Park. African Journal of Ecology, 17, 19–34. Levy, N. & Bernadsky, G. 1991. Crèche behavior of nubian ibex Capra ibex nubiana in the Negev Desert Highlands, Israël. Israël Journal of Zoology, 37, 125–137. McCracken, G. F. 1984. Communal nursing in Mexican free-tailed bat maternity colonies. Science, 223, 1090–1091. Mangin, S., Gauthier-Clerc, M., Frenot, Y., Gendner, J.-P. & Le Maho, Y. 2003. Ticks Ixodes uriae and the breeding performance of a colonial seabird, king penguin Aptenodytes patagonicus. Journal of Avian Biology, 34, 30–34. Martella, M. B., Renison, D. & Navarro, J. L. 1995. Vigilance in the greater rhea: effects of vegetation height and group size. Journal of Field Ornithology, 66, 215–220. Martin, P. & Bateson, P. 1993. Measuring Behaviour: An Introductory Guide. Cambridge: Cambridge University Press. Montgomerie, R. D. & Weatherhead, P. J. 1988. Risks and rewards of nest defence by parent birds. Quarterly Review of Biology, 63, 167–187. Petit, D. R. & Bildstein, K. L. 1987. Effect of group size and location within the group on the foraging behavior of white ibises. Condor, 89, 602–609. Pettingill, O. S., Jr. 1960. Crèche behavior and individual recognition in the colony of rockhopper penguins. Wilson Bulletin, 72, 213–221. Pierotti, R. 1988. Intergenerational conflicts in species of birds with precocial offspring. Proceedings of the International Ornithological Congress, 19, 1265–1274. 537 538 ANIMAL BEHAVIOUR, 70, 3 Powell, G. V. N. 1974. Experimental analysis of the social value of flocking by starlings (Sturnus vulgaris) in relation to predation and foraging. Animal Behaviour, 22, 501–505. Pulliam, H. R. 1973. On the advantages of flocking. Journal of Theoretical Biology, 38, 419–422. Pulliam, H. R., Anderson, K. A., Misztal, A. & Moore, N. 1974. Temperature-dependent social behaviour in juncos. Ibis, 116, 360–364. Scherrer, B. 1984. Biostatistique. Québec: Gaëtan Morin Editeur. Seddon, P. J. & van Heezik, Y. 1993. Chick crèching and intraspecific aggression in the jackass penguin. Journal of Field Ornithology, 64, 90–95. Siegel, S. & Castellan, N. J., Jr. 1988. Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill. Siegel-Causey, D. & Hunt, G. L., Jr. 1981. Colonial defence behavior in double-crested and pelagic cormorants. Auk, 98, 522–531. Siegel-Causey, D. & Kharitonov, S. P. 1991. The evolution of coloniality. Current Ornithology, 7, 285–330. Slater, L. M. & Markowitz, H. 1983. Spring population trends in Phoca vitulina richardi in two central California coastal areas. California Fish and Game, 69, 217–226. Spurr, E. B. 1974. Individual differences in aggressiveness of Adélie penguins. Animal Behaviour, 22, 611–616. Stahel, C. D., Megirian, D. & Nicol, S. C. 1984. Sleep and metabolic rate in the little penguin, Eudyptula minor. Journal of Comparative Physiology, 154, 487–494. Taylor, R. H. 1962. The Adélie penguin Pygoscelis adeliae at Cape Royds. Ibis, 104, 176–204. Tamisier, A. & Dehorter, O. 1999. Camargue, Canards et Foulques. Nı̂mes: Centre Ornithologique du Gard. Thompson, K. V. 1998. Spatial integration in infant sable antelope, Hippotragus niger. Animal Behaviour, 56, 1005–1014. Tourenq, C., Johnson, A. R. & Gallo, A. 1995. Adult aggressiveness and crèching behavior in the greater flamingo, Phoenicopterus ruber roseus. Colonial Waterbirds, 18, 216–221. Trivers, R. L. 1972. Parental investment and sexual selection. In: Sexual Selection and the Descent of Man 1871–1971 (Ed. by B. Campbell), pp. 136–179. Chicago: Aldine. Vinuela, J., Amat, J. A. & Ferrer, M. 1995. Nest defence of nesting chinstrap penguins (Pygoscelis antarctica) against intruders. Ethology, 99, 323–331. Whittow, G. C. 1976. Regulation of body temperature. In: Avian Physiology (Ed. by P. D. Sturkie), pp. 146–173. Berlin: SpringerVerlag. Williams, G. C. 1966. Natural selection, costs of reproduction and a refinement of Lack’s principle. American Naturalist, 100, 687– 690. Wilson, E. A. 1907. Aves. British National Antarctic Expedition 1901–1904. Natural History, 2, 1–121. Wittenberger, J. F. & Hunt, G. L., Jr. 1985. The adaptive significance of coloniality in birds. In: Avian Biology (Ed. by D. S. Farner, J. R. King & K. C. Parkes), pp. 1–78. New York: Academic Press. Yeates, G. W. 1975. Microclimate, climate and breeding success in Antarctic penguins. In: The Biology of Penguins (Ed. by B. Stonehouse), pp. 397–409. London: University Park Press.

We developed a Hidden Markov mark–recapture model (R package marked) to examine sex‐specific demography in Magellanic Penguins (Spheniscus magellanicus). Our model was based on 33 yr of resightings at Punta Tombo, Argentina, where we banded ~44,000 chicks from 1983 to 2010. Because we sexed only 57% of individuals over their lifetime, we treated sex as an uncertain state in our model. Our goals were to provide insight into the population dynamics of this declining colony, to inform conservation of this species, and to highlight the importance of considering sex‐specific vital rates in demographic seabird studies. Like many other seabirds, Magellanic Penguins are long‐lived, serially monogamous, and exhibit obligate biparental care. We found that the non‐breeding‐season survival of females was lower than that of males and that the magnitude of this bias was highest for juveniles. Biases in survival accumulated as cohorts aged, leading to increasingly skewed sex ratios. The survival bias was greatest in years when overall survival was low, that is, females fared disproportionality worse when conditions were unfavorable. Our model‐estimated survival patterns are consistent with independent data on carcasses from the species’ non‐breeding grounds, showing that mortality is higher for juveniles than for adults and higher for females than for males. Juveniles may be less efficient foragers than adults are and, because of their smaller size, females may show less resilience to food scarcity than males. We used perturbation analysis of a population matrix model to determine the impact of sex‐biased survival on adult sex ratio and population growth rate at Punta Tombo. We found that adult sex ratio and population growth rate have the greatest proportional response, that is, elasticity, to female pre‐breeder and adult survival. Sex bias in juvenile survival (i.e., lower survival of females) made the greatest contribution to population declines from 1990 to 2009. Because starvation is a leading cause of morality in juveniles and adults, precautionary fisheries and spatial management in the region could help to slow population decline. Our data add to growing evidence that knowledge of sex‐specific demography and sex ratios are necessary for accurate assessment of seabird population trends.

"In the context of rapid global changes, understanding the foraging mechanisms of marine top predators is crucial to determine their ability to adapt to environmental variability, in order to predict the consequences of changes on population dynamics and to define appropriate conservation measures for threatened species and the trophic webs to which they belong. In the Benguela upwelling system, in South Africa, Cape gannets (Morus capensis) and African penguins (Spheniscus demersus) are vulnerable to extinction and are facing for 10 years onward a progressive south-eastward shift of their main prey, sardines (Sardinops Sagax) and anchovies (Engraulis encrasicolus), probably due to climate change. We performed a detailed study on Cape gannet foraging strategies, using bird-borne recorders, and analysing their diet and the distribution and abundance of their prey in the Benguela. We showed that Cape gannets are able to increase substantially their foraging effort facing poor natural prey availability on the west coast of South Africa. However they are limited in the extent of their foraging area by the need to regularly come back to the colony to feed their young. Thus they turn to feeding behind trawlers on hake fishery discards (Merluccius spp.) of poor nutritional content. These prey represent « Junk-food » for the gannet chicks and have a negative impact on their survival. Even if they allow adults to significantly decrease their foraging effort during winter, fishery discards can not compensate for a poor availability in sardines and anchovies during the breeding season. We have determined that below a density of 25 g m-2 of natural prey in their foraging area, breeding Cape gannets can not equilibrate their energy balance and fail massively in their breeding attempt. Faithful to their breeding site, Cape gannets come back to attempt breeding under sub-optimal conditions, and so are lead to an ecological trap, instead of emigrating. The social structure of a species can therefore lag its adaptation to environmental changes. Small pelagic fish, sardines and anchovies, are currently scarce off the west coast of South Africa, enhancing the competition between seabirds and purse-seine fisheries. Biotelemetry, and particularly GPS tracking on Cape gannets and African penguins at sea, allowed us to precisely define theses birds’ foraging zones. Basing our decision on these data and on purse-seine fisheries activities, we suggest conservation measures that take into account foraging areas of these top predators, while considering local economic needs. These zones are crucial for the maintenance of threatened endemic populations and our data would help in an ecosystemic approach to fisheries management, for a sustainable exploitation of the environment."

Log In

The adaptive significance of crèches in the king penguin