S20.2: Brood size determination in penguins

National Wildlife Research Center, National Commission for Wildlife Conservation and Development, PO Box 1086, Taif, Kingdom of Saudi Arabia, fax 966 2 7455 176, e-mail NWRC@compuserve.com

Penguins evolved during the Cretaceous, some 65+ million years ago (MYA), from a common ancestor to the Procellaridae, Gaviidae and Fregatidae (Williams 1995). The proto-penguin probably lived in a temperate climate and cool waters, but although restricted to the southern hemisphere, no clear centre of origin is known. From 25 MYA, possibly in parallel with the development of circumpolar ocean currents (Fordyce and Jones 1990), the spheniscidae underwent rapid speciation, expanding to fill niches ranging from continental Antarctica to islands north of the equator. Ancestral adaptations for the pursuit of prey underwater, sometimes to great depth, have resulted in a common body form across the 17 species; adaptation to the great variety of environmental conditions however, have resulted in a diversity of reproductive and life history strategies. The defining feature shaping penguin reproductive strategies is the need to come ashore to lay eggs and raise chicks, thus increasing the distance to feeding grounds. Penguin breeding patterns are thus strongly influenced by conditions on land, access to feeding grounds and food availability.

Two of the 17 species of penguin lay a single-egg clutch; the 15 species of penguin that lay two eggs can be divided into two groups, brood maximisers (sensu Edge 1996) and brood reducers (sensu Lamey 1990), depending on the potential for brood reduction that arises from three inter-linked factors: egg-size dimorphism, hatching asynchrony, and sibling size asymmetry.

Brood-reducers include penguins of the genera Spheniscus (4 species) and Pygoscelis (3 species), which show some degree of egg-size dimorphism, and generally commence incubation before the laying of the second egg, resulting in a mean hatching interval of 1-2 days (Williams 1995). Brood reduction theory (Lack 1947, 1954) suggests that in times of extreme food shortage, sibling size asymmetries established by hatching asynchrony may facilitate the rapid starvation of the smaller chick within a brood without endangering the survival of the larger chick, which will then, in the absence of competition for food, have a greater chance of fledging. A review by Timothy Lamey (1990), of the potential for adaptive brood reduction in penguins, lamented the lack of direct evidence available. His synthesis preceded a number of studies designed to examine brood reduction and sibling competition explicitly.

The six species in the genus Eudyptes, the crested penguins, exhibit varying degrees of obligate brood reduction (Lamey 1990), brought about through extreme egg-size dimorphism, large hatching intervals, reversed hatching asynchrony, and marked sibling-size asymmetry. The origins and adaptive significance of these patterns have been the subject of considerable speculation and study.

In this review I examine the inter-specific variation in penguin brood size and consider some of the principal factors which influence the size of penguin broods. I start with a taxonomic summary, dividing species into those that lay one versus two eggs, and further dividing the two-egg species according to their potential for brood reduction. I consider two groups in detail: those with the apparent potential for adaptive brood reduction, and those which exhibit obligate brood reduction. In doing so I aim to revise the Lamey (1990) review, to see how understanding of these two groups has progressed in the intervening eight years.

Of the 17 species of penguin, only two lay a single egg, the other 15 lay two eggs. A single egg clutch is laid by penguins of the genus Aptenodytes, the Emperor A. forsteri, and the King A. patagonicus. Three factors are associated with a single egg clutch in these two species: a large egg, a long chick growth period, and the need to remain mobile during incubation. With adult male pre-breeding body masses of around 40 kg and 16 kg, respectively, the aptly named Emperor and King are the largest of the extant penguin species, and both incubate their single egg on top of their feet (Williams 1995). Large body mass requires a long chick growth period, longer than the period of peak food availability for both species; an apparent obstacle to successful reproduction that is overcome in two ways, both of which impose constraints that would make a two egg clutch impossible.

The breeding cycle of the Emperor takes nine months, excluding pre- and post breeding moult (Robertson 1995). If breeding commenced in the austral spring, chick rearing would have to extend into the Antarctic winter, and chick survival would probably be impossible due to reduced food availability to parents (Kooyman and Kooyman 1995; Robertson 1995), and the extreme cold. The Emperor Penguin therefore must undertake incubation during the austral winter in order to complete chick rearing during the improved feeding conditions in summer. Unlike all other penguin species therefore, the Emperor comes into reproductive condition in response to decreasing daylight (Groscolas et al. 1986). Adults arrive on breeding grounds on level areas of fast sea-ice in late March. After laying of the single egg in mid-May, the females depart for the open sea leaving the males to perform winter incubation alone. This unusual breeding strategy of male-only incubation during winter is because extension of the sea ice would necessitate up to eight days travel one way between colony and open sea, and thus effectively prevents incubation changeovers (Kirkwood and Robertson 1997). The male therefore fasts for the pre-nuptial period and the entire 64 days of incubation, a total of four months (Williams 1995), and endures a loss of up to 40% in body mass (Prevost 1961).

Thermoregulation is a major concern for males, particularly during severe weather when temperatures may drop below -30^oC (Jouventin et al. 1995; Kirkwood and Robertson 1997). Incubating Emperor Penguins may form huddles of 500 to 1000 birds (Le Maho 1977, Le Maho et al. 1976). These huddles move slowly in the direction of the prevailing wind as males on the outer edges move to the leeward side (Robertson 1990). The single egg is held on the top of the feet, beneath an overhanging brood pouch, enabling the penguins to adopt a slow shuffle. Hatching occurs in late winter, in mid-July, at which time the female returns to feed the chick and relieve the male for a 20 day feeding trip (Robertson 1995). Chick growth is constrained by low rates of food delivery and poor thermoregulation during the first third of the five month rearing period, apparently because parents experience difficulty obtaining enough food, due to low food availability, reduced daylight, limited access to foraging waters and severe weather (Robertson 1995). Spring and summer conditions improve adult access to feeding grounds, while increased daylight and twilight allow increased frequency of feeding dives (Robertson 1995). Consequently chicks growth improves, allowing fledging by mid-December, when chicks depart for the ice edge. The Emperor Penguin is thus able to occupy its Antarctic niche and successfully rear a single large chick, to fledge in relatively mild conditions with good food availability, by beginning breeding in winter. The cost of winter breeding comes from the constraints imposed by reduced access to foraging waters, and are borne largely by the male, but may be additionally passed on to the chick in the form of delayed growth and chick mortality due to freezing after abandonment.

The King Penguin similarly incubates its single egg on top of the feet (Hazard 1894), but incubation takes place in the relatively kinder conditions of the sub-Antarctic spring. Access to offshore feeding grounds allows mates to share incubation duties. However, the long period of chick dependence and growth means that chicks are not yet fledged when winter conditions cause an apparent change in availability of myctophid fishes (Adams and Klages 1987; Cherel et al. 1987; Hindell 1988; Ridoux et al. 1988). Parents are unable to rear their single chick to sufficient size before the food supply declines (Stonehouse 1960), so between May and August/September, King Penguin parents largely stop feeding an almost fully grown chick (Stonehouse 1960; Barrat 1976; Cherel et al. 1987; van Heezik et al. 1993). The four-month fast may result in chick deaths due to starvation, or increased vulnerability to predation by Giant Petrels Macronectes giganteus (Hunter 1991; Weimerskirch et al. 1992). For those chicks that survive, feeding recommences in September-October with an apparent increase in food availability (Adams and Klages 1987) and chicks fledge from December, at 10-13 months of age (summary in Williams 1995). The breeding cycle of the King Penguin is therefore extended to 14 to 16 months, and pairs are unable to breed successfully in two consecutive years. Consequently, successful breeders cannot start a new attempt early in the season, and late attempts almost always fail; whereas failed breeders in one year can start a new attempt early and have increased likelihood of success (van Heezik et al. 1994). The ability to raise no more than one chick every two years results in a breeding strategy that has the appearance of being biennial (Barrat 1976). The King penguin is however, endocrinologically an annual breeder; the timing of annual cycles of LH and steroid hormones is probably in response to increasing photoperiod in spring (Mauget et al. 1994; Cherel et al. 1994; Jouventin and Mauget 1996). As a result there are two apparent paradoxes. The first is that successful breeders will start new breeding attempts late in the season, in the face of almost certain failure (van Heezik et al. 1994;), resulting in an extended laying period and the overlap of several stages of breeding within one colony. Increased disturbance in breeding colonies caused by the overlap of stages may be responsible for adults that are incubating eggs or brooding young chicks having to spend almost 8% of their time in aggressive territorial defence, day and night (Challet et al. 1994). The second paradox lies in the ability of adults to sustain chick rearing behaviour over extended absences from the chick. This is due to sustained levels of prolactin (Cherel et al. 1994), maintained possibly by an endogenous rhythm as an adaptation to long parental care in the face of extended absences from the chick (Garcia et al. 1996). Prolactin secretion is sustained for a lesser period in failed breeders, preventing relaying (Jouventin and Mauget 1996), but, presumably in the absence of the stimulus of a chick subsequent to the completion of the endogenous cycle, failed breeders are able to make new attempts early in the following season. Annual periodicity in King penguins is therefore an artefact of low individual breeding success, so that the King penguin reproductive cycle only coincides with the annual rhythm of the environmental cycle when breeding fails (Jouventin and Mauget 1996). In captivity, without environmental constraints on chick feeding, parents are able to raise a chick within nine months (Gillespie 1932). Presumably in captivity therefore, prolactin secretion is triggered by laying/hatching, with the levels sustained endogenously during the shortened period of chick rearing possible with ad lib food, but with early chick independence the stimulus for sustained prolactin would be lost just when the endogenous cycle is ending (when birds in the wild are recommencing chick care after the winter fast), and annual breeding in captivity (or with abundant food) would be possible.

Therefore, as in Emperor Penguins, in King Penguins the long developmental stages needed to raise a large chick to fledging have become constrained by a winter decrease in food availability. Unlike Emperor Penguins however, winter conditions for King Penguins are more mild and the burden of an enforced fast is borne by the chicks, not by the adult males. Breeding has remained entrained by photoperiod so that hatching takes place in spring, and peak food availability allows chicks to gain sufficient fat reserves to survive the winter fast (van Heezik et al. 1993), regain lost weight at the end of winter, and fledge into the good feeding conditions of the following spring.

It has been suggested that the King Penguin represents an intermediate evolutionary step towards cold adaptation (Le Maho 1977), with the Emperor being the furthest evolutionary stage (Jouventin 1971). However, it is likely that the divergent breeding strategies of Aptenodytes have arisen from two different selection pressures. Evolutionarily, a declining availability of food during winter could have selected for King Penguin adults that had the endogenously sustained prolactin that allows extended chick rearing, but with a cut off point that required the active presence of a chick in order for prolactin to continue to be sustained, otherwise failed breeders would be unable to start new attempts early in the following season. For Emperor Penguins, increasing cold, and consequential decreasing accessibility to foraging waters, may have selected for birds that commenced breeding earlier in the season. To a large extent timing of breeding would be dictated proximately by the set and break-out of fast ice. Sustained selection for early breeding would have pushed the start of laying back to early winter, reversing the more usual response to decreasing photoperiod, since birds that could start laying then and which could complete a winter incubation successfully, would raise chicks to fledging in time for relatively mild austral summer conditions. In both species, long chick growth periods cannot be sustained in the face of seasonal declines in food availability; environmental constraints on the possible responses to this dilemma have given rise to two markedly different breeding strategies within the one genus. For Aptenodytes, as for the other penguin species, it should be remembered that observed breeding strategies are workable solutions to problems posed by environmental and physiological constraints, but are not necessarily the most parsimonious, elegant, or even the best.

Adjustment of individual brood size may be related to a number of factors, including age (young breeders are more likely to lay or retain only a single egg), and nest site (nests on the periphery of large colonies, or in exposed settings, are most likely to lose eggs or chicks to disturbance, inclement weather, or predators). Apart from this brief mention, I will not discuss individual brood size adjustment further, but will consider in detail only generic patterns of brood size determination. Throughout the following discussions I use the designation A-(egg or chick) and B-(egg or chick) to refer to the first- and second-laid egg and their chick, respectively.

The Yellow-eyed Penguin Megadyptes antipodes lays two similar-sized eggs (Richdale 1957), and delays the start of incubation (Seddon 1989), so that within a brood the hatching interval is often less than one day (Darby and Seddon 1990), and two same-sized siblings are produced (van Heezik and Davis 1990). Intra-brood aggression is rare or absent, and there is little sibling competition for the food regurgitated by either parent (Seddon 1990). Parents do not appear to favour one chick over its sibling, and as a consequence growth rates of siblings are similar (van Heezik 1990) and in most years a high percentage of breeding attempts will result in the fledging of two chicks (van Heezik and Davis 1990).

In a recent study Keri-Anne Edge (1996) simulated a moderate degree of hatching asynchrony by the exchange of newly hatched Yellow-eyed Penguin chicks between nests. The resultant sibling asymmetries were unstable however, and no differences were recorded in subsequent chick survival, or age or weight at fledging, and there was no apparent effect on parental condition. It was therefore postulated that Yellow-eyed Penguins did not experience selection pressure for a certain degree of hatching asynchrony, and that hatching synchrony was the norm because neither are there selection pressures for early start of incubation. The low degree of hatching asynchrony may possibly be a derived trait due to environmental conditions which favour delayed incubation (Edge 1996), for example, a temperate climate, low risk of predation, and good food availability in most years.

The remaining 13 penguin species exhibit varying degrees of brood reduction. This ranges from the obligatory loss of one egg of the clutch, the obligatory loss of one chick before fledging, to the pre-established potential for loss of one chick only under certain conditions. A key question is whether this facultative brood reduction is of adaptive significance.

Asynchronous hatching and subsequent disparities in chick size and competitive ability have been viewed as a means to facilitate brood reduction (reviewed in: Clark and Wilson 1981; Magrath 1990). The brood reduction hypothesis (Lack 1947, 1954, 1968; Ricklefs 1965; O’Connor 1978) postulates that sibling asymmetries set up by hatching asynchrony are an adaptive mechanism whereby parents can adjust brood size to match feeding conditions in an unpredictable environment. When food is plentiful, all chicks are able to receive enough food and can fledge. When food supplies fail during the breeding season, the starvation of the smaller chick is facilitated through the competitive advantage accrued to the larger sibling, thus avoiding wasteful competition between equally matched siblings to the detriment of both. Broods adjusted by early loss of the weaker chick may be able to salvage at least one chick, rather than experience total failure.

Two genera of penguin, Pygoscelids and Spheniscids, appear to be suitable candidates for testing these ideas, to see whether the establishment of sibling asymmetries has an adaptive significance.

Egg-size dimorphism in the three species of Pygoscelis is slight, ranging from 4.4% in Chinstraps P. antarctica (Moreno et al. 1994) to 6.3% in Adelie P. adeliae (Lamey 1990), and is not believed to be the principal factor leading to the establishment of sibling size differences (Moreno et al. 1994). Hatching asynchrony is more marked, but also more variable, ranging from 1-4 days (mean 1 day between eggs) in Chinstraps (Moreno et al. 1994) to a mean of 1.4 days in Adelies and 1.6 days in Gentoos P. papua (Lamey 1990). The temporal advantage gained by the first-hatched (A) chick means that when the second (B) chick hatches there is a sibling mass asymmetry of 22.4% in Chinstraps (Moreno et al. 1994), and between 28 % to 34.3% in Gentoos (Williams and Croxall 1991).

The establishment of a sibling size hierarchy provides a potential mechanism for classical brood reduction to take place when food is scarce. Whereas in most years Chinstrap parents are able to raise two chick broods to similar final masses as single chick broods (Moreno et al. 1998), food has been shown to be limiting in some years, and starvation of chicks has been recorded (Moreno et al. 1994). However, although when food is limiting the smaller chick in two-chick Chinstrap broods is most likely to succumb to starvation first, subsequent starvation of the surviving chick is not uncommon (Moreno et al. 1994). In addition, likelihood of brood reduction is not related to the degree of sibling asymmetry, and growth of surviving chicks is not markedly better, and may be worse than that of unreduced broods (Moreno et al. 1994). Why should this be so? Pygoscelid chicks compete for food, either through direct aggression or indirectly (Davis and McCaffrey 1989), suggesting that this initial size asymmetry could be maintained throughout the dependent period. However, there is evidence that this is not the case, that in Chinstraps and Gentoos at least, there is a tendency for sibling asymmetries to decrease over time, and that reversals in the size hierarchy are not uncommon (Moreno et al. 1994; Williams and Croxall 1991).

Apparently then, asynchronous hatching does not lead to stable hierarchies between siblings. One reason for this may be the way in which Pygoscelid chicks are fed during the creche phase. Feeding chases, whereby the feeding parent runs and is chased by begging chicks (Sladen 1958; Thompson and Emlen 1968; Thompson 1981), are a characteristic of all species of Pygoscelid (Gentoo: Pettingill 1964; Adelie: Thompson 1981; Chinstrap: Bustamante et al. 1992). Before the creche stage the chicks are fed on the nest site and apportionment of food is determined primarily by sibling competition (Davis and McCaffrey 1989). Feeding chases in the creche stage have been variously proposed as a mechanism for parents to separate their own chicks from a creche; for parents to regulate the distribution of food between the siblings in order to either promote brood reduction (summary in Bustamante et al. 1992), or to increase feeding efficiency by reducing scramble competition (Boersma and Davis 1997). Feeding chases are reduced or non-existent when only one chick is present (Thompson 1981; Bustamante et al. 1992; Moreno et al. 1996; Boersma and Davis 1997), suggesting that chases serve to increase feeding efficiency and/or reduce harassment to the parents (Bustamante et al. 1992). Dee Boersma and Lloyd Davis (1997) hypothesised that if feeding chases were a mechanism for facilitating brood reduction, then the larger chick of the brood should be able to monopolize access to the food delivered throughout a feeding session. This was not the case, rather the chick being fed appeared more likely to switch as a result of the chase. Chases are thereby apparently acting as a parental mechanism to ensure equitable food distribution, reduce direct competition and inefficient transfer of food, and actually act to prevent the development of overlarge asymmetries (Boersma and Davis 1997) that could result in wasteful brood reduction, even when food is plentiful.

Some other evidence further weakens any case for adaptive brood reduction in Pygoscelids. Chick starvation has been recorded in Adelies, but is due primarily to failed nest reliefs by parents, and in general food may not be limiting for Adelies (Boersma and Davis 1997). Nor too are other adaptive explanations for hatching asynchrony supported by the evidence. Synchronously hatching broods may comprise 22-31% of Chinstrap broods, and chicks in synchronous broods reach the same final mass as chicks in asynchronous broods, implying that sibling size asymmetries do not improve feeding efficiency in this species (Moreno et al. 1994).

Asynchronous hatching and subsequent sibling size differences may arise incidentally in Pygoscelids as a result of selection pressure for early commencement of incubation to avoid exposure of eggs to low ambient temperatures (Williams and Croxall 1991), and feeding chases may therefore be a parental mechanism for ensuring that food allocation is equitable and that an overbalanced weight hierarchy does not develop from the incidental asymmetry caused by hatch order. There are three areas of interest for further investigation: (1) examination of post-fledging survival in relation to brood type (where symmetric and asymmetric broods either naturally occur or are created artificially), status of chicks within such broods, and fledging mass or timing; (2) the mechanics of sibling competition and food allocation in the absence of feeding chases (perhaps through erection of a barrier around the feeding group); (3) the proximate causes of intra-specific variation in the degree of hatching asynchrony.

In the four species of the genus Spheniscus, hatching asynchrony creates sibling size asymmetries that can lead to brood reduction (Lamey 1990; Boersma 1991; Boersma and Stokes 1995). Although slight egg-size dimorphism is common in these species, evidence suggests that its contribution to the establishment of sibling asymmetries may be negligible (Seddon and van Heezik 1991a).

The most detailed work has been done on Magellanic S. magellanicus and African S. demersus penguins. I assume that, because patterns of hatching asynchrony and sibling competition are similar amongst the four species, and seasonal variation in food availability is common (Boersma 1976, 1978, Boersma et al. 1990; Adams et al. 1992; Williams 1995), that the results from Magellanic and African penguins could be applied also to the Galapagos S. mendiculus and Humbolt S. humboldti penguins.

Magellanic penguins experience marked inter-annual variation in reproductive success, due primarily to variation in food availability (Boersma et al. 1990). Although symmetric broods do occur naturally, Magellanic chicks within a brood hatch on average two days apart, an interval that promotes the development of sibling size asymmetries (Boersma and Stokes 1995). Differential competitive ability means that starvation of the smaller, second-hatched (B) chick is not uncommon (Boersma and Stokes 1995). Observations indicate that the heavier chick (A) in a brood receives the most food, and that the difference in the amount of food obtained by the two chicks increases with increasing size asymmetry (Blanco et al. 1996). This differential feeding of the larger chick is believed to be due to the A-chick’s greater mobility and strength (Blanco et al. 1996), but it is possible that parents may also treat one chick preferentially (Boersma 1991). Boersma and Stokes (1995) note that fledging success in Magellanics is not well correlated with the degree of hatching asynchrony, i.e., greater degrees of hatching asynchrony are not, on average, associated with greater numbers of chicks fledged per nest, suggesting that parental quality, the timing of the first feeds after hatching and other factors may strongly influence the degree of sibling asymmetry. The significance of this resulting asymmetry is uncertain, though there could be some relationship between broods with mean levels of asynchrony (1.9 days) and both their fledging success and fledging mass (Boersma and Stokes 1995).

African penguin siblings hatch a mean of 2.1 days apart (Williams and Cooper 1984), so that at hatching, the B-chick is on average only 59% of the mass of the A-chick (van Heezik and Seddon 1991). While B-chicks in normal asynchronous broods were more likely to die of starvation than their larger sibling, comparisons of fledging success and survival in normally asynchronous broods and experimentally created synchronous broods which contained two same-sized siblings did not support the predictions of the brood reduction hypothesis (Seddon and van Heezik 1991a,b). However, chicks in experimental synchronous broods had depressed growth rates (van Heezik and Seddon 1991), so that chicks in asynchronous broods fledged earlier and at greater body mass (Seddon and van Heezik 1991a). Observations indicated that A-chicks in normal asynchronous broods out-compete B-chicks, not through direct aggression, but through scramble competition (van Heezik and Seddon 1996), whereby the A-chick uses its superior weight, mobility and co-ordination to gain first access to food regurgitated by the parents (van Heezik and Seddon 1997). A-chicks gain most food during the first part of each feeding session, but following A-chick satiation, the B-chick is able to feed with few interruptions; in this way, in times of abundant food, the development of over-large asymmetries would be avoided, preventing wasteful loss of the B-chick (van Heezik and Seddon 1996).

The evidence to date for Spheniscid penguins, the prime candidates for adaptive brood reducers, suggests that the brood reduction hypothesis is not an adequate explanation for the presence of sibling asymmetries established by hatching asynchrony. A number of other hypotheses have been put forward to explain avian hatching asynchrony, the most applicable of these to Spheniscus penguins (Seddon and van Heezik 1991a) is the sibling rivalry reduction hypothesis SRRH (Hahn 1981), which postulates that sibling asymmetries result in a feeding hierarchy within which the older chick is able to dominate the younger, and wasteful competition between two evenly matched siblings is avoided; these advantages should be evident even in years of good food availability. The SRRH however, also makes predictions about differential fledging success between synchronous and asynchronous broods, which are unsupported by the evidence from Magellanic and African penguins. Seddon and van Heezik (1991a) proposed extensions to the BRH and the SRRH, whereby sibling asymmetries may result in better quality fledglings, with higher post-fledgling survival (see also Offspring Quality Assurance Hypothesis, Slagsvold et al. 1995). Results from the African penguin most strongly support such an extension to the SRRH, so that sibling asymmetries create a feeding hierarchy that allows efficient use of resources and result in heavier (better?, but see Williams and Croxall 1991) fledglings (Seddon and van Heezik 1991a; van Heezik and Seddon 1996).

Recent work (Mock and Lamey 1991; Forbes 1993; Ploger 1997) suggests that avian parents may deliver less food to reduced broods, and that under such conditions brood reduction should not increase the survival rate of the remaining sibling(s). Seddon and van Heezik (1991b) suggested that this parental matching of food delivery to brood size may explain the slightly lower likelihood of survival of surviving chicks in reduced broods. This could be an explanation for the same phenomenon recorded in Chinstrap penguins (Moreno et al. 1994). However, this may also be a reflection of poor parents or parental problems in delivering food. Closer examination is needed to determine if decreased food delivery also means a decrease in food biomass delivered to the surviving chick. In a review of siblicidal raptors Robert Simmons showed that while food delivery to broods may decrease when brood size decreases, food biomass per individual chick actually increased in all cases (Simmons, unpublished manuscript).

I suggest therefore, that rather than search for the adaptive significance of penguin hatching asynchrony as a means to promote brood reduction, we should perhaps consider that hatching asynchrony, in Spheniscids if not in other genera facing intermittent food limitation, serves to establish a sibling size asymmetry within a brood. Sibling size differences may result in a feeding hierarchy that prevents wasteful competition. More efficient use of resources, in asymmetric broods compared with symmetric ones, may be expressed as improved chick fledgling quality, therefore higher post-fledging survival and greater likelihood of recruitment into the breeding population. Considerations of offspring quality, as opposed to quantity, have been the basis for hypotheses concerning the adaptive significance of canism in some raptor species (Newton, 1979; Simmons 1988, 1989, 1997) - the terrestrial, ecological equivalent of long-lived seabirds. Shorter fledging periods may have benefits for parents also, and could improve lifetime reproductive fitness by improving survival and therefore reproductive output, in terms of both numbers of breeding attempts, and numbers of surviving fledglings. If the adaptive significance of hatching asynchrony is the establishment of a feeding hierarchy, then it would be expected that there would be mechanisms that (a) maintained the sibling asymmetry, at least through the period of peak chick growth, and (b) prevented the asymmetry from becoming too great and risking wasteful brood reduction even when food was plentiful.

Little Penguins Eudyptula minor do not fit readily into the breeding pattern of either Pygoscelids or Spheniscids. While they share with Spheniscids a marked inter-annual variation in the timing of breeding (Fortescue 1995) and breeding success (Dann and Cullen 1990) due to variation in oceanographic conditions (Mickelson et al. 1992), food abundance (Hobday 1992), and therefore in food availability, they do not as a group clearly exhibit features which would facilitate the formation of sibling asymmetries. Eggs within a clutch may be similar in size (Stahel and Gales 1987), or the second-laid egg may be lighter (Fortescue 1995). Hatching is mainly synchronous (Williams 1995), but a hatching interval of up to seven days has been recorded (Reilly and Balmford 1975). Sibling asymmetries, presumably established by greater degrees of hatching asynchrony, result in competitive disadvantage to the B-chick, longer B chick fledging periods (Williams 1995) and higher mortality (Gales 1987). The possible adaptive significance of Little Penguin sibling asymmetry has not been examined explicitly, and could lie in improved fledging success or greater fledging mass for chicks in more asymmetric broods. Variation in food availability, breeding success and the degree of hatching asynchrony make the Little Penguin suitable for addressing such questions.

Much work remains to be done concerning the adaptive significance of penguin sibling asymmetries. While the predictions of classical brood reduction theory should continue to be borne in mind, as should the possibility that hatching asynchrony is a non-adaptive consequence of other factors which promote early onset of incubation, greater attention could be given to sub-lethal advantages for the establishment of intra-brood feeding hierarchies, including fledging weight, fledging period, and the relationship between fledgling condition and survival until breeding age. Details of the mechanisms by which food is apportioned between siblings of different competitive abilities could be examined to determine if similar patterns are maintained by different species, or under different feeding conditions or degrees of asymmetry. Closer attention could be paid to those species that show natural variation in the degree of brood reduction between pairs, between years or at different localities.

The adaptive significance of obligate brood reduction in the crested penguins is one of the more alluring of the penguin paradoxes (Croxall and Davis in press). Crested penguins have four traits that have largely defied a single adaptive explanation: (1) the first egg laid is markedly smaller than the second; (2) although two eggs are laid, two chicks are almost never fledged due to egg loss during laying, or to the starvation of the smaller sibling; (3) in instances when the first egg is retained throughout incubation, it hatches after the second-laid egg; (4) significant egg loss in at least two crested species may be due to the direct action of the attending adult (summaries in Williams 1995; St. Clair 1996).

In crested penguins the first-laid (A) egg is 15-45% smaller than, and is laid about four days before, the second-laid (B) egg (Warham 1975). If the A-egg is retained, as it may be in the Fiordland crested Eudyptes pachyrhynchus (Warham 1974a), Rockhopper E. chrysocome (Gwynn 1953, cited in Williams 1995) and Snares crested E. robustus (Warham 1974b), it typically hatches after the B-egg. When two eggs hatch there is no overt sibling aggression, but the smaller chick from the A-egg virtually always starves to death (Lamey 1990). In the other three of the six Eudyptes species, the Macaroni E. chrysolophus (Gwynn 1953, cited in Williams 1995), the Erect-crested E. sclateri (Richdale 1941) and the Royal E. schlegeli (Carrick 1972, cited in Williams 1995), the smaller A-egg disappears from the nest by a few days after B egg laying.

The apparent viability of A-eggs has led to the suggestion that the smaller first egg may be an insurance against egg loss (Lack 1968; Warham 1975; review in Williams 1995), or a response to high losses of first eggs due to conspecific male aggression (Johnson et al. 1987). However, neither of these ideas have been fully supported by direct field tests (Williams 1989; St. Clair 1992; Lamey 1993; St. Clair and St. Clair 1996). The insurance value of the A-egg is limited where both eggs do not remain in the nest together long enough, and timing of A-egg loss is unrelated to levels of male aggression, which result in only minor egg loss (Williams 1995). Often A-eggs are lost in the last day of the laying interval (Williams 1990), and observations have shown that, in Royal and Erect crested penguins at least, the A-egg may be deliberately ejected from the nest, most commonly immediately before the second egg has been laid (St. Clair et al. 1995). Egg ejection, by parent or sibling, may also play a role in egg loss in Macaronis (Williams 1989), and possibly also in Rockhoppers, where peak loss of A-eggs also takes place during peak periods of B-egg laying and hatching (St. Clair and St. Clair 1996).

Explanations for the unusual breeding pattern in crested penguins must take into account reversed egg-size dimorphism, reversed hatching asynchrony, and deliberate and early egg loss. Below I summarise a scenario by which these patterns may have evolved.

Recent phylogenetic studies have confirmed the close relationship between the Yellow-eyed penguin and the crested penguins, and have been used to suggest that traits present in both genera represent common ancestral features (Edge 1996). The Yellow-eyed penguin is a brood maximiser (Edge 1996); same-sized eggs, delayed brood patch formation, delayed onset of incubation and subsequent synchronous hatching, results in two same-sized/aged siblings. Lack of competition during feeding, and adequate food availability means Yellow-eyed penguin parents can raise two chicks to fledging in most years. Edge (1996) postulated that delays in incubation and in brood patch formation have probably arisen independently of reduction in the size of the first egg in crested penguins. If Eudyptid penguins were once, or derived from, similar brood maximisers, and were faced with a decrease in food availability (possibly associated with a postulated change from inshore- to off-shore-foraging, Williams 1980) that meant that two chicks could not be raised to fledging in most seasons, then in many years low food availability would result in chick starvation. The absence of any differential in competitive ability between siblings, in accordance with the brood reduction or sibling rivalry reduction ideas, would be expected to result in low productivity in broods with two same-sized siblings, and total breeding failure might occur. There would therefore be selection pressure for the creation of some degree of competitive difference between siblings, so that either feeding efficiency could be improved, or brood reduction facilitated, in times of food shortage. Given the previous establishment of physiological and behavioural features that delay incubation, if selection favoured unequal investment in offspring, the evolution of egg-size dimorphism may be easier than a change to early brood patch formation (Edge 1996). Thus one way to create this differential would be through differential provisioning of eggs. Crested penguins could either make A-eggs or B-eggs smaller. But if B-eggs were smaller, yet delayed incubation was retained, what would be the outcome? Development in both eggs would start after laying of the smaller B-egg. Hatching would be synchronous and the only differential between siblings would be that established through egg-size dimorphism. However, if the first laid egg were smaller, cooling during the laying period would be greater for the greater surface area to volume ratio (but see St. Clair 1996); incubation would begin with the laying of the B-egg, which would not have cooled, and may therefore get a head start in development through faster heating (e.g. Burger and Williams 1979) or other factors (e.g. St. Clair 1996). With egg-size dimorphism and delayed incubation both serving to disadvantage the A-egg, the B-egg would therefore hatch first, and produce a larger chick. This size differential could be further increased by early feeding of the B-chick before the smaller A-chick hatched, thus establishing a significant size hierarchy within the brood, and the potential for the benefits that may accrue from this, in times of food shortage.

So we have a facultative brood reducer with reversal of the usual egg-size dimorphism and hatching order. The move to offshore foraging may entail a decrease in the time available to feed two chicks, to the point where it becomes no longer possible to raise two chicks in any year (St. Clair pers. comm.). This would imply a selection pressure against the production of even a small egg/chick, unless there was some advantage in retaining the A-egg. It has been suggested that, since an A-egg may be relatively inexpensive to produce, there may be selective inertia against its elimination (Williams 1990), and even a weak secondary function during the laying period may be advantageous (St. Clair et al. 1995; St. Clair in press). There are four postulated secondary functions: (1) the A-egg is insurance against failure to produce a second egg (Williams 1989); (2) the A-egg is necessary to stimulate development of the brood patch in time for incubation to begin fully with the arrival of the B-egg (St. Clair 1992; but see St. Clair in press); (3) the presence of the A-egg ensures that the male will remain ashore, fasting with the female for long enough for the B-egg to develop and for full incubation to begin, i.e. to enhance mate fidelity (Williams 1995); (4) the A-egg may provide a signal to conspecifics that enhances synchrony between neighbouring pairs, and/or reduces contests for nesting space by signalling occupancy (Johnson et al. 1987).

If the value of the A-egg lies in its presence, for whatever reason, only until the B-egg arrives (or is about to arrive, an event that should be able to be predicted by the female as the egg moves down in the oviduct (St. Clair et al. 1995)), then its usefulness ends with the end of laying and the egg can be lost. The most efficient way to lose an unwanted egg is to eject it from the nest, and if this is done in response to the imminent arrival of the B-egg, the ejection would be expected to be done by the female before laying, as observed by Colleen Cassady St. Clair and her colleagues (1995), but possibly by either sex after laying. The observation that retained A-eggs are moved to an anterior position, where they are incubated less efficiently (Burger and Williams 1979; St. Clair 1992, 1996), hints that the action of egg ejection may derive as an extension, or exaggeration, of moving the smaller egg forward in the nest.

Given that an A-egg has been produced, and may be viable, why is this not retained throughout incubation and at least the first stages of chick rearing, for its insurance value? There would be selection to lose the A-egg before incubation if a two-egg clutch reduced B-egg survival, possibly via reduced efficiency of incubation resulting in longer incubation periods, or lower hatching success (St. Clair and St. Clair 1996). In such instances there would be pressure to lose the A-egg as soon as its usefulness during the laying period had been realised. Such factors may be important in Erect crested, Macaroni and Royal penguins, which lose the A-egg within 24 hours of laying of the B-egg. If, however, there was no disadvantage in retaining the A-egg during incubation, yet a possible advantage through its (even small) insurance value, the A-egg could hatch. But if the presence of an A-chick reduces the survival or quality of the B-chick, then there would be pressure for the A-chick to be lost as soon as its value during incubation had been realised. This occurs in Fiordland crested, Rockhopper and Snares crested penguins, where the A-chick is lost within a few days of hatching.

Generic patterns of brood size determination in penguins have been shaped by conditions experienced on land, and by food availability to adults during breeding. The single-egg clutch of King and Emperor penguins appears to be a response to the severe constraints on breeding posed by a long period of chick-rearing in the face of marked seasonal variation in food availability. But whereas the Emperor Penguin has reversed the usual physiological response to photoperiod, and commences a breeding attempt when daylight is decreasing in order to complete chick rearing by the austral summer, the King Penguin extends the chick-rearing period by virtual cessation of chick feeding over winter. The variation between the two Aptenodytes species appears to be due to the inability of Emperor chicks to survive an Antarctic winter, the fasting period ashore being borne by Emperor Penguin adult males, but by King Penguin chicks in the relatively milder sub-Antarctic conditions. The other 15 species of penguin lay two-egg clutches, but range from production of twins, through brood reduction under some conditions, to obligatory loss of one egg or chick. Brood reduction, whether facultative or obligatory, is facilitated by egg-size dimorphism and hatching asynchrony, and the resulting asymmetry between siblings. In potentially facultative brood-reducing genera such as Pygoscelis and Spheniscus, the adaptive significance of early onset of incubation and the establishment of sibling size asymmetries has been the subject of considerable study, although the results to date do not support the classical resource-tracking theories of brood reduction. Hatching asynchrony may result from pressure to protect eggs from predation, low ambient temperatures, or other factors, leading to the creation of non-adaptive sibling size differences. In Pygoscelids, feeding chases may serve to reduce the competitive disparity between two different-sized siblings to avoid loss of the less competitive smaller chick even when food is abundant. In Spheniscids it is possible that hatching asynchrony is adaptive, not in facilitating brood reduction, but in creating a feeding hierarchy within a brood which serves to reduce sibling competition, and may result in better quality fledglings, and perhaps also in greater lifetime reproductive success for adults. The significance of factors leading to obligate brood reduction in the Eudyptids eludes a single, parsimonious explanation. An evolutionary switch from inshore to offshore foraging may have reduced the ability of Eudyptid parents to raise two chicks, leading to selective pressure for the early elimination of one. The retention of a small, viable A-egg, in the face of almost certain loss, may be due to selective inertia for the elimination of an inexpensive egg with some secondary adaptive function. Hypotheses linking the retention of a small A-egg with secondary functions relating to physiological development, e.g. brood patch formation; social stimuli, or insurance value during the laying or early chick period, are currently the best candidates for a unifying theory.

I am grateful to the National Wildlife Research Center, and its parent body the National Commission for Wildlife Conservation and Development, for allowing me to spend at least part of my time thinking about penguins and oceans, instead of Houbara Bustards and Arabian deserts. The following people reviewed early drafts of the manuscript and made a number of valuable comments: Pat Monaghan, Juan Moreno, Graham Robertson, Robert Simmons, Colleen Cassady St. Clair, Yolanda van Heezik, and an anonymous reviewer. Keri-Anne Edge kindly gave me access to unpublished data, while John Croxall, Lloyd Davis, Pierre Jouventin and Colleen Cassady St. Clair facilitated access to material that was published or in press.

Adams, N.J. & Klages, N.T. 1987. Seasonal variation in the diet of the king penguin at sub-Antarctic Marion Island. Journal of Zoology, London 212: 303-324.

Adams, N.J., Seddon, P.J. & van Heezik, Y. 1992. Monitoring of seabirds in the Benguela Upwelling System: can seabirds be used as indicators and predictors of change in the marine environment? In: Payne, A.I.L, Brink, K.H., Mann, K.H & Hilborn, R. (Eds) Benguela Trophic Functioning. South African Journal of Marine Science 12:959-974..

Barrat, A. 1976. Quelques aspects de la biologie et de l’ecologie du manchot Royal (Aptenodytes patagonicus) des Iles Crozet. Comite National Francais de Rechrches Antarctique 40: 9-51.

Blanco, D.E., Yorio, P. & Boersma, P.D. 1996. Feeding behavior, size asymmetry and food distribution in Magellanic Penguin chicks. Auk 113(2): 496-498.

Boersma, P.D. 1976. An ecological and behavioural study of the Galapagos penguin. Living Bird 15: 43-93.

Boersma, P.D. 1978. Breeding patterns of Galapagos penguins as an indicator of oceanic conditions. Science 200: 1481-1483.

Boersma, P.D. 1991. Asynchronous hatching and food allocation in the Magellanic penguin Spheniscus magellanicus. Acta Congressus Internationalis Ornithologici, XX: 961-973.

Boersma, P.D., Stokes, D.L. & Yorio, P.M. 1990. Reproductive variability and historical change of magellanic penguins (Spheniscus magellanicus) at Punta Tombo, Argentina. Pp. 15-43. In: Davis, L. S. and Darby, J. T. (eds). Penguin Biology. Academic Press, San Diego.

Boersma, P. D. & Stokes, D. L. 1995 Mortality, hatching asynchrony, and size asymmetry in Magellanic Penguin chicks. In: Dann, P., Norman, I. and Reilly, P. (eds). The Penguins: ecology and management. Surrey Beatty and Sons, Chipping Norton, Australia.

Boersma, P.D. & Davis, L.S. 1997. Feeding chases and food allocation in Adelie penguins. Anim. Behav. 54: 1047-1052.

Burger, A.E. & Williams, A.J. 1979. Egg temperatures of the rockhopper penguins and some other penguins. Auk 96: 100-105.

Bustamante, J., Cuervo, J.J. & Moreno, J. 1992. The function of feeding chases in the chinstrap penguin. Animal Behaviour 44: 753-759.

Carrick, R. 1972. Population ecology of the Australian black-backed magpie, royal penguin, and silver gull. Population ecology of migratory birds, a symposium. U. S. Department of the Interior Wildlife Research Report 2: 41-98.

Challet, E., Bost, C.A., Handrich, Y., Gendner, J. P. & Le Maho, Y. 1994. Behavioural time budget of breeding king penguins. Journal of Zoology, London 233: 669-681.

Cherel, Y., Stahl, J-C. & Le Maho, Y. 1987. Ecology and physiology of fasting in king penguin chicks. Auk 104: 254-262.

Cherel, Y., Mauget, R., Lacroix, A. & Gilles, J. 1994. Seasonal and fasting related changes in circulating gonadal steroids and prolactin in King Penguins. Physiological Zoology 67(5): 1154-1173.

Clark, A. B. & Wilson, D. S. 1981. Avian breeding adaptations: hatching asynchrony, brood reduction and nest failure. Quarterly Review of Biology 56: 253-277.

Croxall, J.P. & Davis, L.S. in press. Penguins: paradoxes and patterns. Marine Ornithology.

Dann, P. & Cullen, J.M. 1990. Survival, patterns of reproduction and lifetime reproductive output in little blue penguins (Eudyptula minor) on Phillip Island, Victoria, Australia. Pp. 63-84. In: Davis, L. S. and Darby, J. T. (eds). Penguin Biology. Academic Press, San Diego.

Darby, J.T. & Seddon, P.J. 1990. Breeding biology of Yellow-eyed penguins (Megadyptes antipodes). Pp. 45-62. In: Davis, L. S. and Darby, J. T. (eds). Penguin Biology. Academic Press, San Diego, USA.

Davis, L. S. & McCaffrey, F. T. 1989. Recognition and parental investment in Adelie penguins. Emu: 89: 155-158.

Edge, K. A. 1996. Parental investment in penguins: a phylogenetic and experimental approach. PhD. Thesis. Department of Zoology, University of Otago, Dunedin, New Zealand.

Forbes, L.S. 1993. Avian brood reduction and parent-offspring ‘conflict’. American Naturalist 142: 82-117.

Fordyce, R.E., & Jones, C.M. 1990. Penguin history and new fossil material from New Zealand.Pp. 419-446, In: Davis, L. S. and Darby, J. T. (eds). Penguin Biology. Academic Press, San Diego.

Fortescue, M.E. 1995. Biology of the Little penguin Eudyptula minor on Bowen Island and at other Australian colonies. Pp. 364-392. In: Dann, P., Norman, I. and Reilly, P. (eds). The Penguins: ecology and management. Surrey Beatty and Sons, Chipping Norton, Australia.

Gales, R. P. 1987. Growth strategies in blue penguins, Eudyptula minor. Emu 87: 212-219.

Garcia, V., Jouventin, P. & Mauget, R. 1996. Parental care and the prolactin secretion pattern in the King Penguin: an endogenously timed mechanism? Hormones and Behaviour 30.

Gillespie, T.H. 1932. A Book of Penguins. Herbert Jenkins Ltd., London.

Groscolas, R., Jallageas, M., Goldsmith, A. & Assenmacher, I. 1986. The endocrine control of reproduction and moult in male and female Emperor (Aptenodytes forsteri) and Adelie (Pygoscelis adeliae) penguins. I. Annual changes in plasma levels of gonadal steroids and luteinizing hormone. General and Comparative Endocrinology 62: 43-53.

Gwynn, A.M. 1953. The egg-laying and incubation periods of Rockhopper, Macaroni and Gentoo Penguins. A.N.A.R.E. Report (Zoology). (B) 1: 1-29.

Hazard, R.G. 1894. Breeding habits of the King Penguin (Aptenodytes longirostris). Auk 11: 28-282.

Hahn, D.C. 1981. Asynchronous hatching in the laughing gull: cutting losses and reducing rivalry. Animal Behaviour 29: 421-427.

Hindell, M.A. 1988. The diet of the King Penguin at Macquarie Island. Ibis 130: 193-203.

Hobday, D.K. 1992. Abundance and distribution of pilchard and Australian anchovy as prey species for the Little Penguin Eudyptula minor at Phillip Island, Victoria. Emu 91: 342-354.

Hunter, S. 1991. The impact of avian predator-scavengers on King Penguins Aptenodytes patagonicus chicks at Marion Island. Ibis 133: 343-350.

Johnson, K., Bednez, J.C. & Zack, S. 1987. Crested Penguins: why are first eggs smaller? Oikos 49: 347-349.

Jouventin, P. 1971. Comportement et structure sociale chez le Manchot empereur. La Terre et la Vie 25: 510-586.

Jouventin, P., Barbraud, C. & Rubin, M. 1995. Adoption in the Emperor Penguin. Animal Behaviour 50: 1023-1029.

Jouventin, P. & Mauget, R. 1996. The endocrine basis of the reproductive cycle in the King Penguin. Journal of Zoology, London 238: 665-678.

Kirkwood, R. & Robertson, G. 1997. The foraging ecology of female Emperor Penguins in winter. Ecological Monographs 67(2): 155-176.

Kooyman, G.L. & Kooyman, T.G. 1995. Diving behavior of Emperor Penguins nurturing chicks at Coulman Island, Antarctica. Condor 97: 536-549.

Lack, D. 1947. The significance of clutch size. Ibis 89: 302-352.

Lack, D. 1954. The natural regulation of animal numbers. Oxford University Press, Oxford.

Lack, D. 1968. Ecological adaptations for breeding in birds. Methuen, London.

Lamey, T.C. 1990. Hatch asynchrony and brood reduction in penguins. Pp. 399-416 In: Davis, L. S. and Darby, J. T. (eds). Penguin Biology. Academic Press, San Diego.

Lamey, T.C. 1993. Territorial aggression, timing of egg loss, and egg-size differences in Rockhopper Penguins, Eudyptes chrysocome chrysocome, on New Island, Falkland Islands. Oikos 66: 293-297.

Le Maho, Y. 1977. The Emperor Penguin: a strategy to live and breed in the cold. American Scientist 65: 680-693.

Le Maho, Y., Delclitte, P. & Chatonnet, J. 1976. Thermoregulation in fasting Emperor Penguins under natural conditions. American Journal of Physiology 231: 913-922.

Magrath, R. D. 1990. Hatching asynchrony in altricial birds. Biological Review 65: 587-622.

Mauget, R, Jouventin, P, Lacroix A & Ishii, S. 1994. Plasma LH and steroid hormones in King Penguin during the onset of the breeding cycle. General and Comparative Endocrinology 93: 36-43.

Mickelson, M.J., Dann, P. & Cullen, J.M. 1992. Sea temperature in Bass Strait and breeding success of the Little penguin Eudyptula minor at Phillip Island, south-eastern Australia. Emu 91: 355-368.

Mock, D.W. & Lamey, T.C. 1991. The role of brood size in regulating egret sibling aggression. American Naturalist 138: 1015-1026.

Moreno, J., Carrascal, L.M., Sanz, J.J., Amat, J. A. & Cuervo, J.J. 1994. Hatching asynchrony, sibling hierarchies and brood reduction in the Chinstrap Penguin. Polar Biology 14: 21-30.

Moreno, J., Carrascal, L.M. & Sanz, J.J. 1996. Parent-offspring interactions and feeding chases in the Chinstrap Penguin. Bird Behavior 11: 31-34.

Moreno, J., Vinuela, J., Belliure, J. & Ferrer, M. 1998. Brood-size dependent growth in the Chinstrap Penguin: a field experiment. Journal of Field Ornithology 69: 269-275.

Newton, I. 1979. Population ecology of raptors. T & AD Poyser, Berkhamsted, U.K. 399 pp.

O’Connor, R.J. 1978. Brood reduction in birds: selection for fratricide, infanticide and suicide? Animal Behaviour 26: 79-96.

Pettingill, O.S. Jr. 1964. Penguins ashore at the Falkland Islands. Living Bird 3: 45-64.

Ploger, B.J. 1997. Does brood reduction provide nestling survivors with a food bonus? Animal Behaviour. 54: 1063-1076.

Prevost, J. 1961. Ecologie du manchot empereur. Hermann, Paris.

Putz, K. & Bost, C.A. 1994. Feeding behaviour of free-ranging King Penguins. Ecology 75(2): 489-497.

Reilly, P.N. & Balmford, P. 1975. A breeding study of the Little Penguin, Eudyptula minor, in Australia. Pp. 161-187. In: Stonehouse, B. (ed.). The Biology of Penguins. Macmillan, London.

Richdale, L.E. 1941. The Erect-crested Penguin (Eudyptes sclateri). Emu 41: 25-53.

Richdale, L.E. 1957. A population study of penguins. Oxford University Press, Oxford.

Ricklefs, R.E. 1965. Brood reduction in the Curve-billed Thrasher. Condor 67: 505-510.

Ridoux, V., Jouventin, P., Sahl, J-C. & Weimerskirch, H. 1988. Ecologie alimentaire comparee des manchots nicheurs aux Iles Crozet. Revues Ecologie 43: 345-355.

Robertson, G. 1990. Huddles. Australian Geographic 20: 74-97.

Robertson, G.G. 1995. The foraging ecology of Emperor Penguins (Aptenodytes forsteri) at two Mawson Coast colonies, Antarctica. A.N.A.R.E. Reports No. 138, 139 pp.

Seddon, P.J. 1989. Patterns of nest relief during incubation and incubation period variability in the Yellow-eyed Penguin (Megadyptes antipodes). New Zealand Journal of Zoology 16: 393-400.

Seddon, P.J. 1990. Behaviour of the Yellow-eyed Penguin chick. Journal of Zoology, London 220: 332-343.

Seddon, P.J. & van Heezik, Y. 1991.a. Hatching asynchrony and brood reduction in the Jackass Penguin: an experimental study. Animal Behaviour 42: 347-356.

Seddon, P.J. & van Heezik, Y. 1991.b. Effects of hatching order, sibling asymmetries, and nest site on survival analysis of Jackass Penguin chicks. Auk 108: 548-555.

Simmons, R.E. 1988. Offspring quality and the evolution of cainism. Ibis 130: 339-357.

Simmons, R.E. 1989. The Cain and Abel riddle in eagles and other birds. African Wildlife 43: 35-43.

Simmons, R.E. 1997. Why don’t all siblicidal eagles lay insurance eggs? The egg quality hypothesis. Behavioural Ecology 8: 544-550.

Sladen, W.J.L. 1958. The Pygoscelis penguins. 1. Methods of study. 2. The Adelie penguin. Falkland Islands Dependency Survey Science Report 17: 1-97.

Slagsvold, J., Amundsen, T. & Dale S. 1995. Costs and benefits of hatching asynchrony in Blue Tits Parus caeruleus. Journal of Animal Ecology 64: 563-578.

St. Clair, C.C. 1992. Incubation behaviour, brood patch formation and obligate brood reduction in Fiordland crested penguins. Behavioural Ecology and Sociobiology 31: 409-416.

St Clair, C.C. 1996. Multiple mechanisms of reversed hatching asynchrony in Rockhopper Penguins. Journal of Animal Ecology 65: 485-494.

St. Clair, C.C, Waas, J.R., St. Clair, R.C. & Boag, P.T. 1995. Unfit mothers? Maternal infanticide in royal penguins. Animal Behaviour 50: 1177-1185.

St Clair, C.C. & St. Clair, R. C. 1996. Causes and consequences of egg loss in Rockhopper Penguins. Oikos 77.

St. Clair, C.C. in press. What is the function of the first eggs in Crested Penguins? Auk.

Stahel, C. & Gales, R. 1987. Little Penguin. Fairy penguins in Australia, University of New South Wales Press.

Stonehouse, B. 1960. The King Penguin Aptenodytes patagonicus of South Georgia. I. Breeding behaviour and development. Falkland Islands Dependencies Survey Scientific Report 6: 1-33.

Thompson, D.H. 1981. Feeding chases in the Adelie Penguin. Antarctic Research Series 30: 105-122.

Thompson, D.H. & Emlen, J.T. 1968. Parent-chick individual recognition in the Adelie Penguin. Antarctic Journal of the United States 3: 132.

van Heezik, Y. 1990. Patterns and variability of growth in the Yellow-eyed Penguin. Condor 92: 904-912.

van Heezik, Y. & Davis, L. 1990. Effects of food availability on growth rates, fledging sizes and reproductive success in the Yellow-eyed Penguin Megadyptes antipodes. Ibis 132: 354-365.

van Heezik, Y. & Seddon, P.J. 1991. Influence of hatching order and brood size on growth in Jackass Penguins. South African Journal of Zoology 26(4): 199-203.

van Heezik, Y., Seddon, P.J. du Plessis, C.J. & Adams, N.J. 1993. Differential growth of King Penguin chicks in relation to date of hatching. Colonial Waterbirds 16:71-76.

van Heezik, Y., Seddon, P.J., Cooper, J. & Plos, A. 1994. Interrelationships between breeding frequency, timing and outcome in King Penguins Aptenodytes patagonicus: are King Penguins biennial breeders? Ibis 136: 279-284.

van Heezik, Y. & Seddon, P.J. 1996. Scramble feeding in Jackass Penguins: within-brood food distribution and the maintenance of sibling asymmetries. Animal Behaviour 51: 1383-1390.

van Heezik, Y. & Seddon, P.J. 1997. Penguins under the sun. Natural History 106: 30-35.

Warham, J. 1974a. The Fiordland Crested Penguin Eudyptes pachyrhynchus. Ibis 116: 1-27.

Warham, J. 1974b. The breeding biology and behaviour of the Snares Crested Penguin. Journal of the Royal Society of New Zealand. 4: 63-108.

Warham, J. 1975. The Crested Penguins. Pp. 189-269. In: Stonehouse, B. (ed.). The Biology of Penguins. Macmillan, London.

Weimerskirch, H., Stahl, J.C. & Jouventin, P. 1992. The breeding biology and population dynamics of King Penguins on the Crozet Islands. Ibis 134: 107-117.

Williams, A.J. 1980. The breeding biology of Eudyptes penguins with particular reference to egg-size dimorphism. PhD. Thesis. University of Cape Town, South Africa.

Williams, A.J. & Cooper, J. 1984. Aspects of the breeding biology of the jackass penguin, Spheniscus demersus. Pp. 841-853. Proceedings of the Fifth Pan-African Ornithological Conference.

Williams, T.D. 1989. Aggression, incubation behaviour and egg loss in Macaroni Penguins, Eudyptes chrysolophus, at South Georgia. Oikos 55: 19-22.

Williams, T.D. 1990. Growth and survival in Macaroni Penguin, Eudyptes chrysolophus, A- and B-chicks: do females maximise investment in the large B-egg? Oikos 59: 349-354.

Williams, T.D. 1995. The Penguins. Oxford University Press, Oxford.

Williams, T.D. & Croxall, J.P. 1991. Chick growth and survival in Gentoo Penguins (Pygoscelis papua): effect of hatching asynchrony and variation in food supply. Polar Biology 11: 197-202.