This study examines the relationships between metabolic status and behavior in spontaneously fasting birds in the context of long-term regulation of body mass and feeding. Locomotor activity, escape behavior, display songs, body mass, and metabolic and endocrine status of captive male emperor penguins were recorded during a breeding fast. We also examined whether body mass at the end of the fast affected further survival. The major part of the fast (phase II) was characterized by the maintenance of a very low level of locomotor activity, with almost no attempt to escape, by an almost constant rate of body mass loss, and by steady plasma levels of uric acid, β-hydroxybutyrate, and corticosterone. This indicates behavioral and metabolic adjustments directed toward sparing energy and body protein. Below a body mass of ∼24 kg (phase III), spontaneous locomotor activity and attempts to escape increased by up to 8- and 15-fold, respectively, and display songs were resumed. This probably reflected an increase in the drive to refeed. Simultaneously, daily body mass loss and plasma levels of uric acid and corticosterone increased, whereas plasma levels of β-hydroxybutyrate decreased. Some experimental birds were seen again in following years. These findings suggest that at a threshold of body mass, a metabolic and endocrine shift, possibly related to a limited availability of fat stores, acts as a “refeeding signal” that improves the survival of penguins to fasting.
- energy balance
- spontaneous fasting
- body mass threshold
- locomotor activity
- feeding behavior
the long-term regulation of body mass and energy stores is achieved principally through the control of energy intake (20, 25,36) and is most likely based on metabolic signals for control of eating behavior (16). Thus for changes in energy stores and expenditure to affect energy intake, there must be a mechanism that translates alterations in energy metabolism into this behavior. To date, the control of food intake by metabolic cues has been examined mainly using laboratory mammals (3, 22, 28), and the long-term signals that serve to maintain energy balance are still obscure (26).
Undoubtedly, a better understanding of the long-term metabolic control of feeding behavior can be obtained by using wild mammals and birds that spontaneously fast at some stage of their annual cycle (29). These spontaneous fasts occur when animals are engaged in specific activities that compete with feeding, such as hibernation, migration, molt, incubation, or defense of a territory, and are observed even though food is readily available (see Ref. 30 for review). These periods of negative energy balance are anticipated by accumulation of sufficient nutrient stores to cope with future needs. However, disruption of the natural development of their activity and refeeding, or attempts to refeed, have been described in fasting hibernators (12) and in incubating (1, 32, 38) and trans-desert migratory birds (4). Why these behavioral changes occur, how they are triggered, and notably whether they are related to the attainment of some critical energy and/or metabolic status and whether such a disruption of the fast improves further survival are largely unknown.
To examine these questions, the present study was performed on emperor penguins (Aptenodytes forsteri), an antarctic seabird whose males fast for up to 4 mo while breeding on the sea ice (32). Previous studies have shown that in this species, as in other long-term fasting animals, fasting is characterized by a long period of protein sparing and preferential mobilization of fat stores (phase II), followed by a period of increased net protein catabolism (phase III; 19, 33). Since body mass of the leanest emperor penguins leaving their breeding colony to refeed at sea is close to, or only slightly less, than the body mass at the transition betweenphases II andIII, and since no emperor penguin has ever been found fasted to death in the colony, it has been asked whether this metabolic shift could trigger behavioral changes, i.e., egg abandonment and departure to refeed at sea (19, 27, 33). That such a connection between behavioral changes and fasting phases might exist is suggested by the observation that locomotor activity (an index of the motivation to search for food) in the fasting laboratory rat began to increase markedly at the end of phase II (23).
In the present work, possible relationships between behavior and metabolic status of spontaneously fasting emperor penguins were investigated. Emperor penguins feed at sea and are colonial birds, so the feeding activity, or motivation of free-living individuals, cannot be directly assessed on land. Thus the locomotor activity of spontaneously fasting captive emperors was monitored and used as an index of feeding motivation.
MATERIALS AND METHODS
The study was conducted at Dumont d’Urville Station, Pointe Géologie Archipelago, Adélie Land, Antarctica (66°40’S-140°01’E). According to the “Agreed Measures for the Preservation of Antarctic Fauna,” the project was approved by the French Committee for Antarctic Research.
Over three breeding seasons (1981, 1991, 1992), four studies were run on male emperor penguins obtained in the nearby breeding colony. Three studies were made to determine metabolic and behavioral changes during fasting. The fourth study attempted to determine whether the body mass reached at the end of the breeding fast affects survival. Studies were performed on males caught in April within 2 wk of their arrival in the breeding colony. Birds were sexed from their display song and/or body mass, males being fatter than females. They were penned outdoors in a wire enclosure ∼1 km away from the colony and continued their spontaneous fast. Snow was given ad libitum as a water source.
From April to July 1981, the time course of body mass loss, overall behavior, and plasma concentrations of uric acid, β-hydroxybutyrate, and corticosterone were monitored in six penguins. In fasting birds, plasma uric acid and β-hydroxybutyrate concentrations reflect the intensity of protein catabolism and fatty acid utilization, respectively (10). The birds were regularly weighed on a platform balance (±20 g). Blood was regularly sampled from a flipper vein and centrifuged after being heparinized. The collected plasma was frozen until analysis. Uric acid was assayed on whole plasma and β-hydroxybutyrate after deproteinization of the plasma with 7% HClO4 and neutralization with 20% KOH. Enzymatic methods were as in Cherel et al. (11). Corticosterone was extracted from 0.35 ml of plasma with dichloromethane and measured by radioimmunoassay (11). The behavior of birds was scored daily in the middle of the daylight period. For 2 h, the birds were continuously watched from a nearby shelter by a single observer. Display songs were noticed, and the birds were categorized as quiet or active. Quiet birds remained standing or lying, usually in the middle of the enclosure during the whole monitoring period. Birds were considered active if they were observed to walk along the fence wire for several minutes. From this observation, a behavioral index was calculated (see Fig. 2legend).
From April to July 1991, the locomotor activity of six penguins was measured daily to better define the comportmental changes seen in the first study. As in the earlier study, the pen was next to a shelter from which an observer could watch the birds through a small window without interfering with their behavior. After 1 wk of adaptation to captivity, the number of steps and of pecks in the fence wire (an index of escape behavior) made by each bird were counted for 1 h in the middle of the daylight period. All counts were made by the same observer simultaneously on the six birds. Blind countings by other observers on only a single bird yielded the same results. Birds were weighed regularly on a platform balance (±50 g) in the evening and after measurement of their activity so as to avoid any bias in the activity data that might result from handling. Then the birds remained undisturbed until the next weighing.
This study was run on seven penguins to observe diurnal and nocturnal locomotor activity (number of steps) for 24-h periods during the last part of the fast when marked behavioral and metabolic changes occurred (see below). The birds, caught on arrival in the breeding colony (April 1992), were kept in a pen as in the previous studies. At a 26.10 ± 0.20 kg body mass, i.e., after 50 ± 3 days of fasting, the birds were equipped with a 30-g pedometer having a digital display (model 201, Azimut). It senses the shock of the foot hitting the ground at each step. Pedometers were stuck to the feathers with epoxy at the level of the hip. They were wrapped in a transparent plastic sheet to avoid icing by blizzards. Each pedometer was calibrated by comparing the indication given by the apparatus to visual countings during 50- and 75-step walks. A correction factor (walked steps/pedometer indication) was determined for each individual and was applied in the calculations. Pedometers were read twice a day, at the onset of daylight (1000) and at the onset of darkness (1700). Birds were weighed daily (±50 g) at 1700.
The same year as study 1, 13 males (initial body mass 36.5–41 kg) were caught at the onset of breeding and were fasted in a pen until reaching a body mass ranging from 24 to 18 kg. Then they were banded and released. In addition, 18 free-living males were weighed and banded at the end of the incubation shift, when they left the breeding colony to refeed at sea after being relieved by their mate. The presence or absence of these banded birds in the breeding colony was checked over the next five years. Since some birds might have been missed, including because of band loss, the data from this study cannot be used to estimate survival rate. The presence of a bird indicates that it survived, but that a bird was not seen again does not mean that it died.
Data Analysis and Statistics
Values are means ± SE. Values were compared according to the Peritz’ F test for multiple comparisons. Data were plotted versus body mass instead of time of fasting because the procedure allows a much better interindividual adjustment as is illustrated in Fig. 1 for specific daily body mass loss and plasma uric acid. Each different parameter was fitted to body mass using nonlinear regression analysis. The fits were composed of two successive linear segments, the intersection of which yielded the body mass at which changes in the parameter occured.
Body mass loss. At the beginning of the fast in the pen, the average body mass of the six birds was 38.5 ± 1.1 kg. During the major part of the fast, i.e., for 91.5 ± 6.6 days, the specific daily body mass loss remained at the low value of 6.7 ± 0.4 g ⋅ kg−1 ⋅ 24 h−1 (Figs.1 and 2). For 2 wk afterwards, it increased by a factor of 3.5 (P < 0.001), reaching 23.5 ± 1.1 g ⋅ kg−1 ⋅ 24 h−1 when the penguins were released at 18.1 ± 0.3 kg body mass (Fig. 2). In accordance with previous studies (10, 33), periods corresponding to low or increasing specific daily body mass loss were named phase II and III of fasting, respectively.
Uric acid. Duringphase II of fasting, plasma uric acid level was maintained at the low value of 0.18 ± 0.02 mmol/l (Fig.2). Thereafter, it rose by a factor of 5–10 (P < 0.001), reaching 1.53 ± 0.15 mmol/l at the end of phase III.
Corticosterone. Duringphase II of fasting, plasma corticosterone remained at the low level of 31.9 ± 4.9 nmol/l (Fig.2). Phase III was characterized by a 2.2-fold rise of corticosterone (P < 0.001) to 71.4 ± 10.4 nmol/l when the birds were released.
β-Hydroxybutyrate. Duringphase II of fasting, the plasma concentration of β-hydroxybutyrate remained at 0.81 ± 0.11 mmol/l (Fig. 2). During phase III, it decreased twofold (P < 0.05) and reached 0.49 ± 0.05 mmol/l at release of the birds.
Behavioral index. During the first 3 mo of fasting (phase II), the behavioral index remained at low values (Fig. 2). It increased by a factor of two during phase III(P < 0.001).
Display song. Display songs were heard during the first 3 wk after capture and confinement (Fig. 2). Thereafter, the penguins remained silent for 2 mo, up to the end ofphase II. Display songs were heard again during phase III.
Body mass loss. When penned, the six birds weighed an average of 40.9 ± 0.4 kg. Duringphase II, the specific daily body mass loss remained at a steady low value of 5.3 ± 0.3 g ⋅ kg−1 ⋅ 24 h−1 (Fig.3). During phase III, it increased progressively (P < 0.001) and reached twice thephase II level when the birds were released at a 21.5 ± 0.5 kg body mass (Fig. 3).
Locomotor activity. Duringphase II, locomotor activity remained at a low value (84 ± 14 steps/h; Fig. 3). Duringphase III, the number of steps rose progressively and significantly (P < 0.001) by up to eightfold. During the first 3 mo of fasting, attempts to escape remained few and erratic (Fig. 3). However, this activity began to increase once the penguins were in phase III of fasting and reached values 15 times higher than during the period of quiescence.
This study focused on the details of the behavioral changes at the transition between phase II andphase III.
Body mass loss. At capture, the seven penguins weighed an average of 39.2 ± 0.9 kg. Duringphase II, the body mass loss remained low (Fig. 4). The fact that the pedometers were glued on the birds during this phase of fasting (seematerials and methods) did not significantly change the course of the body mass loss. The specific daily body mass loss was 7.2 ± 0.4 g ⋅ kg−1 ⋅ 24 h−1 and 6.8 ± 0.2 g ⋅ kg−1 ⋅ 24 h−1, respectively, before and after the pedometers were used. During phase III, specific daily body mass loss increased significantly (P < 0.001) and reached a three times higher level when the birds were released at a 21.8 ± 0.2 kg body mass.
Total daily activity. In agreement with the first two studies, total locomotor activity remained at a low level as long as the birds were in phase II of fasting, with the number of steps over 24 h equal to 1,291 ± 155 (Fig. 4). During phase III, the number of steps significantly increased (P < 0.001), the birds being about three times more active when they were released.
Nycthemeral pattern of activity. During phase II, the birds were about twice as active during the 7 h of daytime (899 ± 129 steps) than during the 17 hours of nighttime (381 ± 33 steps) (Fig. 4). Duringphase III, both diurnal and nocturnal locomotor activity significantly increased. However, there was no change in the partition of total activity between day and night, ∼70% remaining diurnal (Fig. 4).
Horary activity. As illustrated in Fig. 5 for a single bird, the difference between daytime (1000–1700) and nighttime (1700–1000) activity is amplified when expressed on an horary basis. Over the entire fasting period, the horary locomotor activity was 5 ± 1 times higher during daytime than during nighttime. During thephase II of fasting, the horary number of steps during daytime was 128 ± 19 (n = 7), a value not significantly different from the one found in study 2 during the hour of counting in the middle part of daytime.
Body mass for changes of measured parameters in studies 1, 2, and 3. All the measured parameters (specific daily body mass loss, locomotor activity, display songs, plasma metabolites, and corticosterone) showed a prolonged period of steady state (phase II) followed by a period of rapid change (phase III). The corresponding body masses at the transition between phase II andphase III are shown in Table1 for the three studies. They ranged from ∼22 to 26 kg. No significant body mass differences appeared for any of the measured parameters at the phase II-phase IIItransition for a given study. However, the body mass at transition determined from daily body mass loss was significantly higher duringstudy 2 than instudies 1 and3. This can be related to a higher initial body mass of birds in study 2. The body mass at transition was directly related to the body mass at the beginning of the fast (Fig.6 A). In contrast, it was not related to the duration of phase II (Fig. 6 B). The total body mass loss at the phase II-phase IIItransition differed among individuals (from 13 to 20 kg) and was also directly related to initial body mass (r = 0.62,n = 19,P < 0.01).
On average, the body mass at the beginning of fasting was 39.5 ± 0.6 (n = 19, birds fromstudies 1,2, and3 pooled). The corresponding averaged body mass at transition determined from specific daily body mass loss was 23.8 ± 0.4 kg.
This study sought to determine whether the body mass of a bird at the end of the fast affects its reappearance in the colony during following years.
Free-living birds. Among the 18 birds that were banded at their departure from the breeding colony, six (33%) were seen again during following years (Fig.7). At departure, some of these six birds were among the leanest. Their mean body mass at departure was not significantly different from that of the birds that were not seen again (Table 2).
Captive birds. The body mass at release of the 13 birds was significantly lower than that of free-living birds at departure (Table 2). Six of the 13 birds (46%), including some of the leanest, were seen again in the breeding colony in following years (Fig. 7). Their body mass at release was not significantly different from that of the birds not seen again (Table2).
Metabolism and Behavior During Phase II of Fasting
In accordance with previous findings in birds (10) and mammals (8), this study shows that the major part (phase II) of long-term fasting is characterized by1) the maintenance of low rates of body mass loss, 2) a low rate of protein catabolism, as reflected here by the low plasma uric acid level, and 3) the preferential reliance on fat stores as the major source of energy fuel, as indicated by the high plasma level of β-hydroxybutyrate. In this study,phase II was also characterized by the maintenance of a low plasma level of corticosterone. This has been previously reported in long-term fasting king penguins and has been suggested as contributing to protein sparing through regulation of gluconeogenesis at a low rate (11).
This study is the first to report on activity in relationship to metabolic status in spontaneously fasting penguins. It shows that in the emperor penguin, phase II is associated with a very low and steady level of locomotor activity. This is in agreement with previous findings in laboratory rats (23) and also with the observation that in the emperor penguin fasting induces an increase of slow wave sleep at the expense of wakefulness (15). Low locomotor activity probably contributes to energy saving, since the walking metabolic rate in fasting emperor penguins is three- to fourfold higher than the resting metabolic rate (14). From data on walking energetics in the latter study and from the energy equivalent of body mass loss (19), it can be estimated that the ∼1,300 steps walked daily by a 25-kg emperor penguin (from study 3) represents only 3.5% of its daily energy expenditure. Reduced walking has been previously observed in free-living fasting emperor penguins during incubation (21). However, this might be attributable to the fact that they incubate their single egg on their feet, which limits motion, and to the need to huddle together to save energy during cold weather (32). We show here that during phase II of fasting, the locomotor activity is reduced even in nonincubating, nonhuddling captive emperors.
Behavioral Changes Reflect an Increased Drive for Refeeding and are Triggered by Attainment of a Threshold Body Mass
Following phase II and below a body mass close to 24 kg, marked changes in behavior occurred, including increased locomotor activity, attempts to escape, and resumption of display songs. Thus below this body mass, one of the strategies allowing energy saving, i.e., very low activity, is abandoned. Such an increase in locomotor activity has been previously observed in fasting rats (35), notably in coincidence with phase III (23). It has been interpreted as reflecting an increase in the refeeding drive, the increase in locomotor activity being immediately suppressed when food is given (23). The same explanation probably holds here, that is, the increase in locomotor activity and in escape attempts reflect an internal pressure for refeeding that outweighs the motivation to stay inactive and to save energy. In fasting rats, the increase in locomotor activity duringphase III is associated with an increase in the proportion of diurnal activity (23). In contrast, the day-night pattern of locomotor activity was maintained in the emperor. This observation is somewhat surprising given the connection between increased activity and increased drive to refeed. Indeed, it has been observed that free-living emperor penguins move toward the sea where they feed only during daytime (32), so that a proportionally higher diurnal activity could have been expected. In addition to increased locomotor activity, phase III was associated with a resumption of display songs. This could also be interpreted as reflecting a motivation to depart to sea to refeed since in the wild this departure, which is usually triggered by the return of the female, is immediately preceded by a brief period of intense display songs (32). Within the hours before spontaneous abandon of the egg and departure to refeed at sea, display songs were also observed in fasting incubating king penguins not relieved by their partner (Groscolas, unpublished data). It is remarkable that despite having been nonincubating for ∼3 mo, captive male emperor penguins retain this behavior that signals their readiness to depart for feeding.
The body mass below which behavioral changes occurred was very similar for three different years (22.4–25.4 kg) and also was similar to that reported previously for metabolic changes, i.e., 23 (14) and 23.6 kg (33). Evidently, the changes in behavior correspond to a threshold body mass. In contrast, the duration of fasting before the occurrence of these changes was variable (from 60 to 110 days, see Fig. 6), as was the total body mass loss, and there was no relationship betweenphase II duration and threshold body mass. This strongly supports the idea that the stimulation of the refeeding drive is related to the attainment of a given energy status rather than to a given duration of fasting or body mass loss. Such a conclusion would be in agreement with the fact that, based on various behavioral indexes such as activity, in deprived mammals and birds body mass is a better indicator of feeding motivation than the duration of deprivation (31). However, in emperor penguins body mass seems to be only a rough indicator of the energy status associated with the stimulation for refeeding. There were marked interindividual differences in the threshold body mass, which ranged from 21 to 27 kg. The higher values were associated directly with higher body mass at the onset of the fast (see Fig. 6). This might be explained by interindividual differences in structural size (32), the heavier birds at the onset of fasting being also the larger, i.e., those with the higher fat-free body masses. Thus the stimulation for refeeding could correspond to a threshold level of energy stores relative to the size of the bird rather than to an absolute mass of energy stores or body mass. Given that fasting penguins rely almost entirely on stored fatty acids as energy fuel (19, 33), we suggest that the increase in the drive to refeed is associated with the attainment of a threshold adiposity (fat mass/body mass). From previous studies (19, 33), this adiposity can be estimated as close to 10%. It could correspond to a “defended energy level,” as postulated in animal anorexia (30).
Evidence for a Metabolic and Endocrine Refeeding Signal that Contributes to Survival of Long-Term Fasting
We have previously shown in breeding fasting emperor penguins that reaching a 24-kg body mass corresponds to a metabolic shift characterized by a decrease in the contribution of lipids to energy expenditure and to a compensatory increase in the contribution of body protein (19, 33). This shift is reflected by an increase in daily body mass loss and plasma uric acid and by a decrease in plasma β-hydroxybutyrate (Fig. 2). A decrease in plasma nonesterified fatty acids has also been observed (19). Substantial evidence for the hypothesis that feeding in birds is sensitive to circulating free fatty acid levels or related parameters of lipid metabolism has been found recently (5). In rats fed high-fat diets and thus massively using lipids as energy fuels as do fasting emperor penguins, the inhibition of fatty acid oxidation has been shown to stimulate food intake (17,24). We therefore hypothesize that in fasting emperors a decreased utilization of lipid fuel, possibly resulting from a decreased release of fatty acids from adipose tissue at a threshold adiposity, is at the origin of metabolic changes that ultimately lead to the stimulation of feeding behavior. This would allow integration of the control of body fat with the control of food intake, as proposed by the lipostatic theory (18, 22).
As in previous observations on king penguins (11), the present study shows that the phase III of fasting is associated with an increase in the plasma concentration of corticosterone. This increase could be responsible for the increase in protein catabolism, since corticosterone is known to mobilize peripheral calorie stores for glucose production and energy utilization (37). The increase in plasma corticosterone could also intervene in the rise of locomotor activity, as recently shown in fasting rats (9), and may play a major role in the stimulation of feeding behavior. In fasting passerine birds, corticosterone has been shown to stimulate escape behavior and perch hopping, which suggests that it is important for initiation of food searching (2). In mammals, glucocorticoids stimulate caloric intake at the brain level (6) and have been proposed as major long-term regulators of both energy intake and storage (34,37). More specifically, the ratio between circulatory corticosteroids and insulin over the long term was hypothesized as a major determinant of the level of food intake (13, 37). In fasting penguins, insulin is maintained at low levels during phase III (11). The enhancement of locomotor activity and of the refeeding drive observed in the present work is therefore concomitant with a pronounced increase in the corticosterone-to-insulin ratio, which suggests that the above hypothesis could apply to birds. Plasma corticosterone is also known to increase in response to a wide range of stressors in birds and mammals; this increase serves to mobilize energy stores via protein catabolism, which is an adaptative measure in the short term (2). There are therefore strong parallels between the response to stress and that observed presently in an emergency state. This suggests that an increase in corticosterone secretion is a basic adaptative mechanism allowing vertebrates to face a wide range of critical energy situations.
The rapid transition in behavior triggered at the threshold depletion in fat stores has adaptive value for free-living penguins by redirecting behavior toward the search for food and restoration of energy reserves and survival. This is illustrated here by the finding that many free-living birds departing to refeed at a body mass similar to the threshold body mass, as well as those released at a body mass slightly below this threshold, succeeded in refeeding and were seen again during following years. Also, it appears that at the point when feeding behavior is triggered, birds have an energy safety margin. In our study, captive birds released to refeed at an average 19.2 kg body mass were seen again in proportion similar to free-living birds departing at an average 23.5 kg body mass. This safety margin probably helps the birds to cope with the unpredictable extent of sea ice they have to cross at a high-energy cost before they reach the open sea where they feed (19, 33).
With the use of spontaneously fasting birds, we show that there is a close connection between metabolic status and behavior throughout the course of a long-term fast. Whereas locomotor activity of emperor penguins is very reduced throughout the major part of a fast, it markedly increases below a threshold body mass characterized by metabolic and endocrine changes. This suggests that a metabolic shift associated with a threshold depletion of fat stores triggers refeeding. This could also explain the spontaneous disruption of activities in fasting birds and mammals (see introduction and Refs. 1, 4, 12, 32,38). A limitation of the contribution of fatty acids to energy flux could be the key metabolic signal. This hypothesis could be tested by examining fatty acid release and oxidation during thephase II to phase III transition and by studying the effect of an experimental blockade of fatty acid utilization on the metabolic and endocrine status and on behavior. Another, but not exclusive, possibility would be that adipose tissue directly informs the central nervous system of its status and thereby intervenes in the control of feeding behavior. In this context, it will be interesting to examine if adipose tissue of spontaneously fasting birds secretes leptine or a similar hormone. In mammals, leptine is secreted in proportion to adipose mass and plays a major role in the control of food intake and energy balance (7). Further studies on the relationships between metabolic, hormonal, and behavioral changes in spontaneously fasting birds or mammals will allow a better elucidation of the long-term control of feeding behavior and body fat.
We thank the members of the 31st, 41st, and 42nd French Expedition in Adélie Land for assistance in parts of the experiments and Dr. H. Weimerskirsh for providing us the data on the recovery of banded penguins. We are grateful to Dr. Y. Le Maho for stimulating discussion at various stages of this study and Dr. Y. Cherel for help in hormonal determination.
Address for reprint requests: J.-P. Robin, Centre d’Ecologie et Physiologie Energétiques, Centre National de la Recherche Scientifique, 23 rue Becquerel, 67087 Strasbourg, France (E-mail:).
This work was supported by grants from Terres Australes et Antarctiques Françaises and the Institut Français pour la Recherche et la Technologie Polaires.
- Copyright © 1998 the American Physiological Society