INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE METABOLISM DURING FASTING and the regulation of fasting duration have been extensively studied in resting animals (3, 10, 12, 13). In particular, wild animals fasting spontaneously during part of their annual cycle have provided invaluable information on the ecological context and significance of the types of fuel substrates used and the regulation of termination of fasting (e.g., see Refs. 12 and 36). After an initial transient phase one of adaptation, long-term fasting birds and mammals enter a steady-state phase two, during which the animal's energy expenditure is covered by a maximum proportion of lipids and a minimum proportion of protein. The duration of phase two depends on the initial amount of lipid stores and does not exist in lean fasting animals (10). Phase two ends spontaneously when lipid stores drop to a certain level (threshold adiposity) (36). The subsequent phase three is characterized by a metabolic and endocrine shift, resulting in a dramatic increase in protein catabolism, a rise in glucocorticosteroids and a change in behavior (3, 11, 13, 14, 36); resting fasting animals usually restart foraging activity and may desert their brood (9, 18, 30, 32, 36). This closely resembles observations in birds that show that unpredictable "stressful" events activate the hypothalamo-adenohypophyseal-adrenal axis, which results in the release of glucocorticosteroids and disrupts the current "normal activity." Birds then enter a transitory emergency state to cope with the situation (44, 45).

Migrating animals also fast during their extensive journeys, but in contrast to resting animals, their metabolism and the regulation of fasting during endurance exercise are largely unknown. During endurance flights, migrant birds are in the following particular physiological situation: they fast for up to several days when crossing oceans or deserts, and they exercise at a very high energetic level (2.3 times the maximum achieved by small mammals) (7). Therefore, their metabolism has to deal simultaneously with a long fasting period without food or water intake and a very high energy expenditure. This prompts the question whether fasting during endurance exercise is governed by the same metabolic and endocrine shifts that apply to all fasting resting birds and mammals or whether endurance exercise entails different metabolic and endocrine regulations.

The proportion of lipids and proteins catabolized during migratory flights is of foremost importance for the success of these long nonstop flights (25). It has been suggested that, just as fasting resting birds, migratory birds rely on a similarly high proportion of energy derived from lipids (25). Hence, despite high energy expenditure during flight, a metabolic state is attained that accords with phase two of inactive fasting birds in terms of the proportion of fuel types used. However, it is unclear whether migrants also show an increase in protein use before lipid stores are exhausted, which parallels phase three of fasting birds at rest.

Results about the adrenocortical response to endurance exercise in birds are controversial so far. Birds caught after long flights have been found to have low [tropical passerine migrants (19, 40)] and medium to high corticosterone levels [homing pigeon (20); migrant wader species (34)]. This may be due to the fact that corticosterone, the main glucocorticoid in birds, has different effects depending on the level secreted and the physiological condition of the bird (44-46). Fluctuations in low levels of plasma corticosterone may be associated with, among other things, the regulation of the metabolism under predictable environmental fluctuations. Levels surpassing these normal levels considerably may be associated with transitory, facultative stressful events and may trigger an emergency state (reviewed in Refs. 44 and 45).

Elevated levels of corticosterone promote gluconeogenesis and if levels are high for prolonged periods or during a fasting state, they result in breakdown of muscle protein (21, 29, 45). Hence, corticosterone is likely to be involved in regulating a change in the proportions of fuel types used during endurance flight (transition from phase two to three), and we predicted elevated corticosterone levels in situations when birds need to make metabolic or behavioral adjustments (19).

However, it is unknown whether a decrease in body stores below a certain level during flight represents a predictable event to which the bird is adapted during the life-history stage of migration or whether it is perceived as an unpredictable event that is handled through a high secretion of corticosterone and an emergency state. Again, there are controversial findings in migrating birds as to the adrenocortical response to an acute stressor (23, 34, 40). It has been postulated that the adrenocortical stress response in migrants is suppressed, because the bird does not benefit from an emergency state when flying, particularly over inhospitable areas (34).

This study investigated 1) whether a metabolic shift similar to phase three of fasting birds at rest occurs during endurance flight, 2) whether there is an adrenocortical response associated with a change in fuel types used and with fuel stores remaining, and 3) whether there is an adrenocortical response to an acute stressor (handling) during the first minutes after capture. We investigated these questions in migrant small passerine birds that arrived at an Italian island after having crossed the Sahara desert and, in a nonstop flight of ~500 km, the Mediterranean Sea. We used circulating levels of uric acid as a measure of current protein breakdown and the visible amount of subcutaneous fat stores and an estimate of breast muscle thickness as measures of current lipid and protein stores. Adrenocortical activity was assessed by measurements of plasma corticosterone levels.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Ventotene Island is situated 50 km off the Tyrrhenian coast of Italy (40°48'N/13°25'E). In spring, tropical migrants arrive at the island with northeastern directions, which indicates that they came from North Africa and had crossed at least 500 km of open sea. We took advantage of the possibility to study freshly landed migrants varying widely in energy stores. The amount of energy stores remaining on Ventotene varied between species and was, among other factors, correlated with the latitude of the northernmost preferred habitat south of the Sahara, i.e., migration distance from sub-Saharan Africa (33). In species fuelling at a considerable distance south of the Sahara, mostly lean individuals were caught on Ventotene. In species fuelling closer to the Sahara, mostly fat individuals were caught, whereas some species with intermediate energy stores showed a larger intraspecific variation. Birds were caught at the southwestern tip of the island in mist nets. The following observations indicate that most migrants were caught within a few hours (usually within 30 min) after landing. Retraps of birds ringed at Ventotene are scarce, indicating that the majority leave the island the same day. Migrants can be observed arriving at the island from the sea. They move through the trapping site in waves, and their behavior is different from that of the few foraging individuals resting on the island (unpublished observation). Because we included in this study only birds that did not show signs of recent food intake (defecation of insect or nectar remains during capture and handling, fresh pollen around the bill), we probably eliminated almost all individuals resting on the island for more than about 1 h.

Ten species of tropical migrants were investigated during peak spring migration (29 April-9 May, 1995; 29 April-7 May, 1996): One species of the columbiformes, the turtle dove (Streptopelia turtur) and nine species of the passeriformes: redstart (Phoenicurus phoenicurus), whinchat (Saxicola rubetra), icterine warbler (Hippolais icterina), whitethroat (Sylvia communis), garden warbler (S. borin), wood warbler (Phylloscopus sibilatrix), willow warbler (P. trochilus), spotted flycatcher (Muscicapa striata), and the pied flycatcher (Ficedula hypoleuca). Mist nets were observed continuously from a hide, and birds trapped were blood-sampled within 6 min (usually 2-4 min) after flying into the net. Blood was obtained by puncturing the alar vein and collected with a capillary system (Microvette CB300 Fluore, Sarstedt). Then, birds were banded, and, among others, body mass, fat score (27), and muscle score (2) were taken before release. In the turtle dove, the dense feathering of the underparts prevented the scoring of fat stores.

Fat scores were obtained by estimating the visible amount of subcutaneous fat deposits between the furcula and on the abdomen in live birds (9 level score) (27). These scores correlate well with the amount of fat extracted from whole birds (27 and unpublished data). Muscle scores estimate the thickness of the breast muscle (4 levels: 0 = breast muscle emaciated and its cross section shaped concavely, 3 = breast muscle bulging and shaped convexely) (2). Because the fat extracted from breast muscles of small passerines across all muscle and fat score classes on Ventotene Island is only 1.9% ± 1.2 (SD; n = 84, range 0.2-5.5%) of fresh breast muscle mass and decreases linearly with decreasing lean dry breast muscle mass, muscle scores represent an estimate of breast muscle protein mass. Fat score was repeatable to 98% (n = 1,319), and muscle score to 75% (n = 605), in a calibration study involving 43 persons (F. Bairlein, unpublished data). Because fat and muscle scores for this study have been taken by three experienced persons only, repeatability is likely to be even higher.

The blood, kept in a cooler with ice, was centrifuged within 30 min in the field, and the plasma was stored in liquid nitrogen. After shipment to Switzerland, samples were stored at 20°C. Uric acid was determined by an enzymatic colorimetric kit (Sigma Diagnostics, procedure number 685) adapted for 5 µl of plasma. Corticosterone levels were measured by radioimmunoassay: 2,000 counts/min of tritiated corticosterone were added to each plasma sample of 10 µl and allowed to equilibrate for 24 h at 4°C. Samples were then extracted with 4 ml of redistilled methylene chloride. Extracts were dried under a stream of nitrogen and redissolved in 550 µl of phosphate-buffered saline (pH 7.1). Two hundred microliters were transferred to each of two assay tubes, and 100 µl were transferred to a scintillation vial to estimate recovery. The remainder followed a previously published protocol (e.g., see Ref. 39). Recoveries were 87%, and the lowest detectable dose was 8 pg/tube. The samples of the two study seasons were measured in two different assays. A pool containing 250 pg of corticosterone measured 241 and 255 pg, respectively, in these assays. Intra-assay variation was 6.5%.

Because the amount of blood collected varied and because also other metabolites were determined, not all measurements could be made in all individuals. For statistical analyses, we used SPSS (PC version 2.0).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The following observations suggested that our results from birds of unknown landing time represent the situation of newly grounded birds. We observed two garden warblers, five willow warblers, and nine turtle doves, which arrived from a considerable height above the sea with northeastern directions, landed on the island, and were caught and blood-sampled within <10 min after landing. The plasma uric acid levels of these willow warblers were indistinguishable from conspecifics we considered freshly landed, but whose landing time was not known exactly [uric acid: means 1.07 ± 0.13 (SD) (5) against 1.02 ± 0.45 (57), P = 0.8], and the same applied for uric acid [0.38 ± 0.22 (9) against 0.53 ± 0.28 (46), P = 0.13] and corticosterone levels of turtle doves [3.68 ± 1.98 (9) against 5.40 ± 3.49 (42), P = 0.16]. There were only one or two corticosterone values for garden and willow warblers, respectively, and only two uric acid values for garden warblers, which also were in the range of values of conspecifics considered freshly landed.

In four species in which the sex could be determined in the majority of individuals, there was no difference between sexes in plasma corticosterone levels. In none of the species was there a significant trend with date or time of day for uric acid and corticosterone levels (except for uric acid in pied flycatcher; see Plasma uric acid levels). For subsequent analysis, we, therefore, did not distinguish between sexes and did not include date and time of day as a factor (except in the pied flycatcher).

Plasma uric acid levels. Plasma uric acid levels were not dependent on time elapsed between capture and sampling in any of the species.

Plasma uric acid levels differed significantly between species [ANOVA, F(9,315) = 28.7, P < 0.001]. Three species had low mean uric acid levels: the frugivorous and nectarivorous whitethroat and garden warbler and the granivorous turtle dove (Fig. 1). In the remaining seven insectivorous species, there was a significant negative correlation of mean uric acid level and mean fat score (Fig. 1A, r = 0.92, P = 0.003) and mean muscle score (Fig. 1B, r = 0.78, P = 0.04). Because mean fat and mean muscle scores were correlated (r = 0.85, P = 0.015, n = 7), it remained unclear whether fat or protein stores or both influenced plasma uric acid levels.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Mean plasma concentrations of uric acid in 10 species in relationship with mean fat score (A) and mean muscle score (B). Horizontal and vertical lines give SE of means and, if not shown, are smaller than symbol for mean. Closed symbols, insectivorous species; open symbols, species feeding also on plant materials. Fh, pied flycatcher, Ficedula hypoleuca (n = 27); Hi, icterine warbler, Hippolais icterina (26); Ms, spotted flycatcher, Muscicapa striata (19); Pp, redstart, Phoenicurus phoenicurus (8); Ps, wood warbler, Phylloscopus sibilatrix (23); Pt, willow warbler, Phylloscopus trochilus (62); Sr, whinchat, Saxicola rubetra (28); St, turtle dove, Streptopelia turtur (in B only) (55); Sb, garden warbler, Sylvia borin (52); Sc, whitethroat, S. communis (23).

Because the seven insectivorous species had similar uric acid levels, we combined them for the analysis of the dependence of uric acid levels on both fat and protein stores (Fig. 2). Excluding birds with muscle score of zero, plasma uric acid levels were significantly different between fat scores, but not between muscle scores and species [3-way ANOVA; fat score F(5,164) = 2.6, P = 0.026; muscle score F(2,164) = 0.01, P = 0.9; species F(6,164) = 0.8, P = 0.5; interactions not significant]. Uric acid levels were at ~1.0 mM in birds with fat scores three to six and were increased to ~1.5 mM in birds with fat score zero (Fig. 2). In birds with fat scores one and two, uric acid levels seemed to increase with decreasing muscle score (Fig. 2), but this was not significant. Birds with a muscle score of zero also had fat scores of zero, and their uric acid levels were not significantly different from birds with muscle scores one or two and fat score zero [F(2,50) = 1.5, P = 0.23].

View larger version (11K): [in this window] [in a new window]   Fig. 2.   Mean plasma uric acid concentrations (±SE) of 7 insectivorous species combined according to fat and muscle score. Numbers above graph denote sample sizes.

Because species generally arrived at the island with either small or large energy stores (33), the relationship of uric acid levels and energy stores could only be investigated in those species that showed large variation in energy stores. As measures of energy stores, we used fat score, muscle score, and body mass. Uric acid levels were negatively correlated with fat score [pied flycatcher ANOVA F(1,24) = 6.5, P = 0.02; wood warbler F(3,18) = 3.4, P = 0.04; whinchat F(1,26) = 5.8, P = 0.02], muscle score [wood warbler ANOVA F(3,18) = 5.4, P = 0.01], and with body mass (wood warbler, linear correlation r = 0.53, P = 0.01, n = 22; whinchat r = 0.41, P = 0.03, n = 28). Similar tendencies were observed in other species. Furthermore, pied flycatchers arriving at Ventotene later during the day showed higher uric acid levels than individuals arriving earlier during the day (r = 0.59, P = 0.002, n = 26), but not lower body mass (r = 0.02, P = 0.9).

Plasma corticosterone levels. Plasma corticosterone levels differed significantly between species [ANOVA, F(9,168) = 12.1, P < 0.001]. Mean corticosterone levels per species decreased significantly with mean fat score (r = 0.76, P = 0.02, n = 9) and showed a tendency to decrease with mean muscle score (r = 0.58, P = 0.08, n = 10; Fig. 3, A and B). Mean corticosterone levels per species also increased with mean uric acid levels (Fig. 3C; r = 0.67, P = 0.03, n = 10; for insectivorous species only: r = 0.90, P = 0.005, n = 7).

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Mean plasma corticosterone concentrations of 10 species in relationship with mean species-specific fat score (A), mean muscle score (B), and mean plasma uric acid levels (C). Horizontal and vertical lines give SE of means and, if not shown, are smaller than symbol for mean. See Fig. 1 for abbreviations of species. Sample sizes are Fh (n = 13), Hi (11), Ms (10), Pp (5), Ps (7), Pt (19), Sr (19), St (51), Sb (29), Sc (9).

Because there were no obvious differences between passerine species when accounting for fat and muscle scores, we combined them to analyze the relationship of corticosterone levels with both fat and muscle stores. When excluding birds with muscle score zero, plasma corticosterone levels were significantly different between fat scores, but not between muscle scores and species [3-way ANOVA; fat score F(5,104) = 3.6, P = 0.005; muscle score F(2,104) = 0.7, P = 0.5; species F(8,104) = 1.34, P = 0.23; interactions not significant]. Corticosterone levels were low in birds with fat scores three or higher and increased when fat scores were below three (Fig. 4). Within fat scores, muscle score had no effect on corticosterone levels, except in birds with fat score zero. All birds with muscle scores of zero had no visible subcutaneous fat (fat score 0) and showed very high corticosterone levels compared with birds of fat score zero, but muscle scores of one or two [ANOVA, F(2,260) = 6.8, P = 0.005]. The three highest corticosterone levels (> 50 ng/ml) occurred in birds with both fat and muscle scores of zero (2 spotted flycatchers, 1 icterine warbler). There was no significant relationship between plasma uric acid levels and corticosterone levels in all passerines (n = 109, P = 0.8) or in all seven insectivorous species (n = 74, P = 0.7).

View larger version (11K): [in this window] [in a new window]   Fig. 4.   Mean corticosterone concentrations (±SE; if not shown, SE is smaller than symbol for mean) of 9 passerine species combined according to fat and muscle score. Numbers above graph denote sample sizes.

Within species, the same general relationships were observed. In the garden warbler [ANOVA F(2,26) = 3.7, P = 0.04] and the whinchat [F(1,17) = 5.1, P = 0.04], birds with low fat stores had significantly higher corticosterone levels than birds with high fat stores, and a similar tendency was observed in whitethroat and spotted flycatcher. In the spotted flycatcher, there was a significant negative correlation of corticosterone levels and body mass (r = 0.71, P = 0.02, n = 10). In this species, 40% of the individuals examined had emaciated breast muscles (muscle scores 0 or 1) and very low fat scores (0 or 1); body mass was apparently a better indicator of muscle mass than muscle score, which is unable to differentiate among different degrees of emaciated breast muscles, and this may explain the extreme position of this species in Fig. 3.

Plasma corticosterone levels increased significantly within 6 min after capture in icterine warbler (r = 0.74, P = 0.02, n = 10), turtle dove (r = 0.42, P = 0.003, n = 50), and almost significantly in the willow warbler (r = 0.44, P = 0.06, n = 19). In the remaining seven species, there was no indication of an increase of corticosterone levels with time lapse after capture (all P > 0.26). The three species that showed an increase of corticosterone levels with handling time all had a high proportion of birds with fat scores greater than two. When combining the nine insectivorous species (Fig. 5), birds with fat scores three to six showed a positive correlation of corticosterone levels with duration of capture stress (r = 0.45, P = 0.003, n = 41, no significant effect of species P = 0.9; Fig. 5B). In contrast, in birds with fat scores two or lower (excluding birds with muscle score 0) such a positive correlation was absent (r = 0.05, n = 78, P = 0.7; Fig. 5A).

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Plasma corticosterone concentrations of 9 passerine species combined in relation to time elapsed between capture and blood sampling for birds with fat scores 0-2 (A; excluding birds with muscle score 0) and for birds with fat scores 3-6 (B).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Whereas there is ample evidence that plasma uric acid levels reflect body protein breakdown in long-term fasting birds at rest (e.g., see Refs. 11, 12, 14, 36, 37), direct evidence that this is also the case for birds during endurance exercise is lacking. However, it is likely, because both protein breakdown and uric acid levels are increased in birds during flight compared with fasting birds at rest (4, 17, 25, 26, 41).

Uric acid levels are correlated with dietary protein intake for at least 6 h after feeding in chicken (31) and in absorptive birds of many other species. This suggests that an effect of dietary protein content on plasma uric acid levels is still noticeable after a considerable flight time (Fig. 1).

Protein breakdown: indications of a phase three. In long-term fasting resting birds, phase two is characterized by low daily nitrogen loss and, as a result, low uric acid levels. The initial amount of body fat stores determines the length of this phase and the level of protein breakdown (e.g., Refs. 10 and 25). Phase three is characterized by dramatic increases in daily nitrogen loss and uric acid levels. In birds and mammals with high initial fat stores, allowing for an extended phase two, phase three begins typically when there are still substantial amounts of fat stores left (threshold adiposity) (10, 12, 13, 36).

The situation in migrating birds is similar to that of long-term fasting resting birds, at least in terms of uric acid levels. Birds embarking on a nonstop flight across the Mediterranean have large fat stores comparable to those of long-term fasting birds. Uric acid levels increased when fat score dropped to two (Fig. 2). In species arriving at Ventotene Island with a high variation in fat scores, a similar pattern was found within species. Body composition analyses of dead birds from Ventotene Island showed that fat score of two, the probable threshold adiposity, represents 4-5% fat of the total body mass (R. Schwilch, A. Grattarola, F. Spina, and L. Jenni, unpublished data). Thus the threshold adiposity may be lower than in Emperor Penguins (10%) (36).

The plasma uric acid levels of birds with fat scores three to six after landing at Ventotene Island (~1 mM, Fig. 2) were similar to the levels of birds caught during night migration at an Alpine pass (26), but higher than in overnight fasted resting small passerines (~0.5 mM) (26), whereas values >1 mM are typical for birds with low fat stores (19).

There was considerable variation in muscle scores for a given fat score (Fig. 2; F. Spina, unpublished data). It remains unclear whether this variation was caused by variation in the ratio of fat to protein stored before the onset of endurance flight or whether it is the result of a variation in the amount of protein breakdown during flight. The possible, although not significant, negative correlation of uric acid levels and muscle score within fat scores one and two (Fig. 2) suggests that there might be individual differences in the amount of protein breakdown and in the onset of phase three relative to body fat stores. As suggested by the increase in uric acid levels of pied flycatchers with time of day in the absence of a concurrent decrease in body mass, such individual differences might be related with the duration aloft (assuming that all birds started from North Africa the evening before) and possibly adverse conditions encountered during flight (causing a later arrival at Ventotene).

During fasting, protein is not catabolized uniformly from all organs. Whereas migratory garden warblers and blackcaps fasted for 1.5-3 days in the laboratory predominantly reduced the digestive tract (24, 28), we found that birds on Ventotene Island had reduced the flight muscles more than other organs when arriving in phase two of fasting and the digestive organs more than other organs when arriving in phase three of fasting (R. Schwilch, et al. unpublished data). Hence, whether birds enter phase three of increased protein breakdown during migratory flight or not may determine the functioning of the digestive organs at the onset of subsequent stopover and refuelling.

Regulation of protein breakdown: role of corticosterone. In long-term fasting penguins, plasma corticosterone levels increase with the onset of phase three (11, 14, 36). Plasma corticosterone levels in relation to fat stores found in the present study followed a similar pattern, as circulating corticosterone levels were elevated when fat stores dropped to a score of two (Fig. 4). However, the pattern of migrating birds can be differentiated further. Three levels of circulating corticosterone were indicated: a low level in birds with fat scores three to six, a medium level in birds with fat scores zero to two and muscle scores greater than zero, and a high level in birds with fat score zero and muscle score zero (Fig. 4). Although both uric acid and corticosterone levels increased when fat score dropped to two, the individual values were not correlated, but only the mean values of the species (Fig. 3C). Reasons for this absence of correlation were that uric acid and corticosterone had different relationships with fat and muscle scores. Whereas uric acid concentrations showed two levels according to fat score and, within fat scores one and two, seemed to be correlated with muscle score, corticosterone concentrations showed three levels and no correlation with muscle score for a given fat score.

The low corticosterone levels found in birds with abundant fat stores are in agreement with earlier findings in garden warblers caught in the Sahara after an endurance flight (40), in various passerine migrants caught out of nocturnal or diurnal migratory flight in the Swiss Alps (19), and in garden warblers fasting in the laboratory for several days during the autumn migration season (M. Ramenofsky, unpublished data). This suggests that fasting and simultaneous high-level exercise do not result in elevated corticosterone secretion and, hence, are not perceived as stressful by well-adapted migrants. This is in contrast to domesticated species and homing pigeons, in which nonvoluntary fasting or endurance exercise results in increased corticosterone levels (20, 21, 35) and may have been perceived as stressful.

The medium corticosterone levels found in birds with low or almost depleted fat stores can be interpreted as a means to regulate the types of fuel used during endurance flight, i.e., the onset of phase three with an increased protein breakdown. Hence, migrants with low or depleted fat stores, but intact flight muscles, still do not enter a stressful emergency state. Corticosterone is known to regulate metabolism and, in particular, to increase protein breakdown in fasting animals (29, 45). Therefore, the catabolism of body energy stores in a normal, although large, range in migrants during endurance flight seems to be regulated by variation in low and medium corticosterone levels. A dependence of circulating corticosterone levels on body mass or energy stores is a well-known phenomenon (44) and occurs also in migrants after endurance flight (34, 40, and this study).

In contrast, corticosterone levels increased dramatically when fat stores were depleted and flight muscles (representing overall protein stores) were critically emaciated (Fig. 4), and this situation seems to be a stressful event for migrants. It resembles the corticosterone response of birds to unpredictable events, such as storm or prolonged human disturbance, which disrupt the current normal activity and result in an emergency state (44, 45). In the latter situation, birds disrupt their normal behavior (e.g., territorial behavior, reproduction, flocking during winter) and move away from the source of stress, seek refuge, or increase foraging (44, 45). However, the behavioral effects of high corticosterone levels in migrating birds aloft are unclear. The most appropriate behavior for a bird during endurance flight might be to reduce flight altitude (thereby gaining potential energy from earlier climbing), land as soon as possible in an acceptable habitat, and start foraging.

Whether medium levels of corticosterone redirect behavior in migrants during endurance flight is not known. But it is possible that medium levels of corticosterone trigger landing. Nocturnal migrants flying over land generally end a flight bout before sunrise (5) and, in the laboratory, they show a peak in corticosterone level in the early morning during the migratory season (40).

Adrenocortical response to handling stress. A bird's responsiveness to an acute stressor, measured by corticosterone release into the circulation, is commonly evaluated by repeated blood sampling, assuming that this procedure is stressful (e.g., see Refs. 38 and 46). The stress response may also be evaluated by examining the relationship between corticosterone levels and time elapsed between capture and blood sampling (e.g., see Refs. 34 and 40). It has been suggested that birds in severe environments (arctic and desert breeders) (44) and on migration (23, 34) adaptively inhibit the adrenocortical response to environmental stressors, because their current normal activity (breeding; refueling or flying during migration) has to continue even under stressful conditions.

The present study showed that the response to handling stress during the first 6 min depends on fat stores and the baseline corticosterone levels (Fig. 5). Fat birds had low initial corticosterone levels and an increase with handling time, whereas lean birds had medium initial corticosterone levels and showed no increase with handling time. These results may, at least in part, explain the contradictory results obtained in previous studies of birds after landing. During the migratory period, fat free-living and captive garden warblers had low corticosterone levels and they showed increases in levels with handling time (40). In contrast, lean bar-tailed godwits, Limosa lapponica, captured after a long-distance flight, had medium corticosterone levels and did not increase corticosterone levels in response to handling (34). Hence, fat migrants may show increased corticosterone levels in response to a stressor, a pattern found normally in birds (46). In contrast, lean migrants that are upregulating their protein breakdown through moderate corticosterone levels may inhibit further corticosterone secretion in response to a stressor, because there is apparently no advantage in redirecting their behavior from landing and refuelling.

Perspectives

It emerges that fasting during both low- and high-energy expenditure may have a common regulatory mechanism. A metabolic and hormonal shift is associated with a threshold of energy stores (energy safety margin) and triggers a change in behavior. Whereas fasting birds at rest restart foraging activity, the behavior in flying birds remains to be studied. It seems likely that in both resting and exercising birds, the metabolic and/or hormonal shift triggers a refeeding signal (36). This may be via a decreased release of fatty acids from adipose tissue at a threshold adiposity (36) or via the secretion of leptin, which, in mammals, is secreted in proportion to adipose mass (8). In the present study, there was no significant decrease in plasma free fatty acid levels with decreasing fat score (unpublished results), which favors the second hypothesis. To our knowledge, a leptin gene (43, see also Ref. 16) and its protein have not yet been unequivocally identified in birds, and intracerebroventricular administration of mouse leptin failed to reduce food intake in the chicken (6). Should a leptin-like adipocyte hormone be demonstrated in birds, and should it have regulatory roles similar to those in mammals (for example, see Ref. 15), including inhibition of the hypothalamic-pituitary-adrenal axis in response to restraint stress (22) and fasting (1), then our results would suggest a putative, but to-date unidentified interaction between leptin, corticosterone, and protein breakdown.

The present study supports the idea that plasma corticosterone levels are dependent, in a nonlinear manner, on body energy stores (threshold adiposity). Moreover, this study demonstrates that the adrenocortical response to an acute stressor may not only vary in amplitude (suppression or enhancement) (38, 44, 45), but may actually change its relationship with energy stores (positive, negative, or no relationship with fat stores) in relation to the ecological context the bird is experiencing (for example, see Ref. 42). Hence, differences between species, populations, seasons, stage of migration, and individuals in the adrenocortical stress response may be explained by differences in body energy stores relative to the current (possibly variable) threshold adiposity and the ecological context (for example, see Ref. 38).