Daily energy expenditure increases in response to low nutritional stress in an Arctic-breeding seabird with no effect on mortality



1. The regulation of energy expenditure in relation to food availability and its consequences for individual fitness in free-ranging animals are poorly understood. Increased daily energy expenditure (DEE) may be viewed as the result of two different processes: expenditure may be forced upwards by low food availability (forcing hypothesis) or enabled to increase by high levels of food resources (enabling hypothesis). Several studies have suggested long-term fitness costs due to increased mortality as a trade-off to increased DEE.

2. We examined the relationship between energy expenditure and an indirect measure of food availability, and the short-term fitness consequences associated with changes in DEE in a small, Arctic seabird, the little auk (Alle alle). We measured DEE of 43 parent little auks by the doubly labelled water method during two consecutive breeding seasons and inferred food availability from plasma concentrations of the stress hormone corticosterone (CORT).

3. We found that DEE was elevated by 26·7% in the year with reduced levels of CORT, indicating that little auks responded to increased food availability by increasing their DEE. These results support the enabling hypothesis. Elevated DEE was presumably caused by increased parental effort as reflected by higher chick provisioning rates and larger chick meals, and was associated with fitness benefits in terms of enhanced current reproductive success.

4. Contrary to earlier studies, our data did not indicate adverse effects associated with elevated DEE; there was no negative relationship between DEE and the probability of adults returning to the colony the following year. Instead, adult return rate was positively related to body mass, with lower return rates when food was limited.

5. These results suggest that ecological consequences associated with limited resource availability may outweigh possible direct negative physiological effects of elevated DEE.


All living processes require energy, and the rate of energy expenditure is a fundamental quantity integrating all aspects of the ecology of a species (Brown et al. 2004). Although interspecific variation in energy expenditure can largely be explained by a number of environmental and biological features, such as body mass, phylogeny, ambient temperature or latitude (see Bryant 1997; Nagy, Girard & Brown 1999; Speakman 2000 for reviews), temporal and spatial variations within species are less well understood. One key factor that is likely to play an important role in the regulation of energy expenditure in free-ranging animals is food availability. Theoretically, food availability may affect daily energy expenditure (DEE) through two different processes: DEE may be forced upwards by poor food supply or enabled to increase by high levels of food availability (Speakman et al. 2003). The direction of adjustment in energy expenditure associated with changes in food availability has important implications for an individual’s total energy budget and the amount of energy available that can be expended on competing fitness-related activities, such as reproduction or self-maintenance. Hence, better knowledge about the relationship between food availability and energy expenditure may help in understanding the evolution of life-history traits and trade-offs.

A forced elevation in energy expenditure may reflect the need of an individual to compensate for low food availability by an increase in foraging effort (forcing hypothesis). There is a positive feedback in this scenario, with DEE increasing disproportionately with diminishing food resources because the additional energy spent on foraging would require more food and consequently even more foraging effort and energy expenditure (Gorman et al. 1998). Alternatively, elevated DEE may be the result of an enabling effect, with high metabolism being fuelled by the excess energy available in the environment (enabling hypothesis). Under this scenario, the positive relationship between food availability and DEE would be driven by fitness benefits associated with the extra energy spent (Speakman et al. 2003). For example, a high level of energy expenditure during the reproductive season may benefit offspring through increased parental effort (Bryant & Tatner 1988) or benefit the parent by an increased allocation of energy to self-maintenance.

However, a high rate of energy expenditure, whether a consequence of forcing or enabling, may also entail a fitness cost. For example, both experimental (Deerenberg et al. 1995; Daan, Deerenberg & Dijkstra 1996; Golet et al. 2004) and empirical studies (Bryant 1991) have shown that an increase in DEE may be associated with a reduction in future survival. The ‘free radical damage hypothesis’ (Harman 1956) has been proposed as a proximate mechanism to explain these findings. This hypothesis posits that an increased metabolic rate leads to augmented generation of superoxide free radicals, inevitable byproducts of oxidative metabolism, which may lead to increased somatic damage and consequently to ageing and death (Harman 1956; Beckman & Ames 1998).

In free-ranging animals, the regulation of DEE in response to variation in food resources has rarely been studied and results are equivocal. For example, in birds a positive relationship between food availability and DEE has been found in house martins (Delichon urbica; Bryant & Tatner 1988) and kittiwakes (Rissa tridactyla; Jodice et al. 2006), but a negative association has been reported in the blue tit (Parus caeruleus; Thomas et al. 2001) and great tit (P. major; Tinbergen & Dietz 1994), and in a second study on kittiwakes (Kitaysky et al. 2000). The direction of the relationship between DEE and food availability may therefore vary both within and between species and life history stages depending on costs and benefits incurred by elevated DEE under different circumstances.

Food availability can be regarded as the result of a complex interaction of factors, such as abundance, accessibility, patchiness and nutrient composition of prey (Cairns 1987), and the difficulty to obtain a direct measure of it has been a major limitation for studies examining ecological mechanisms that may explain variation in DEE. However, recent research has indicated that the plasma concentration of the stress hormone corticosterone (CORT), the main avian glucocorticoid, constitutes a good indirect proxy of food availability (Pravosudov et al. 2001; Lynn, Breuner & Wingfield 2003; Schoech, Bowman & Reynolds 2004). In seabirds, a number of both experimental (Kitaysky et al. 1999a, 2001) and observational studies (Kitaysky, Wingfield & Piatt 1999b; Wingfield et al. 1999; Kitaysky, Piatt & Wingfield 2007) have shown that the secretion of CORT is largely driven by changes in food availability. Under controlled experimental conditions, plasma CORT levels increase in proportion to the severity of the food deprivation (reviewed in Kitaysky et al. 2003) and decrease when food is supplemented (Clinchy et al. 2004; Schoech, Bowman & Reynolds 2004). Under natural conditions, a strong negative correlation between CORT and fish abundance, for example, has been demonstrated in the common guillemot (Uria aalge; Kitaysky, Piatt & Wingfield 2007). Based on 3 years of concurrent measurement of the abundance of zooplankton prey and CORT secretion, a negative relationship has also been found in the little auk (Alle alle; N. Karnovsky, unpublished data). Short-term variation in the nutritional status of an individual is reflected in baseline concentrations of CORT which increase rapidly (days) in response to food stress (Kitaysky, Piatt & Wingfield 2007). Chronic food shortage, however, causes an enhancement of the adrenal function and thus the capacity of birds to release CORT. Maximum CORT production, measured in response to a standardized stressor, therefore corresponds to food-related stress integrated over longer time periods (weeks; Kitaysky et al. 2001; Kitaysky, Piatt & Wingfield 2007). Thus, measuring both baseline and maximum levels of CORT provides information on long- and short-term variation in food stress.

The aim of this study was: (1) to determine the relationship between energy expenditure and food availability, and (2) to evaluate fitness consequences associated with changes in DEE in a small (body mass ∼160 g) diving seabird, the little auk (Fig. 1). Little auks have a high field metabolic rate which, during reproduction, exceeds predictions for marine birds based on body mass by more than 70% (Gabrielsen et al. 1991; Nagy, Girard & Brown 1999). This high rate of energy expenditure is likely to be caused by high energetic costs of foraging and chick provisioning due to wing-propelled locomotion during diving, and a combination of energy intensive flapping flight, high wing loading and long foraging distances (Gabrielsen et al. 1991; Stempniewicz 2001). Changes in food availability are, therefore, likely to have a strong impact on DEE in this species.

Figure 1.

 A parent little auk returning to its colony with food for the chick stored in a throat pouch. Photo by Carsten Egevang ©.

We measured DEE by the doubly labelled water method (Lifson & McClintock 1966; Butler et al. 2004) in chick-rearing little auks during two consecutive breeding seasons, and inferred food availability during these seasons from plasma levels of CORT obtained from individuals simultaneously sampled at the same colony. In addition, we estimated chick fledging success and the return rate of parents to the colony the year following DEE measurements to assess potential costs and benefits to current and future reproduction in connection with changes in DEE. Finally, we recorded chick provisioning behaviour with respect to feeding rate and meal size as a possible mechanistic link between energy expenditure and reproductive success.

We predicted that if elevated DEE was an adaptive response enabled by high food availability, a positive relationship between DEE and food availability would be associated with enhanced chick provisioning and reproductive success. However, if DEE was negatively related to food availability, e.g. forced upwards by poor conditions, we expected DEE to be inversely related to food delivery and offspring survival. In both cases, we expected a negative correlation between DEE and the return rate of adults the following year.

Materials and methods

Study area and species

The study was conducted at a little auk colony of c. 2000 breeding pairs at Kongsfjorden (79° 01′ N, 12° 25′ E), Svalbard, Norway. Main field work took place during the chick rearing period of two consecutive breeding seasons, from June to August in 2006 and 2007. Additional data (adult return rate) were collected in June and July 2008. The breeding distribution of little auks is confined to high Arctic latitudes. Little auks establish their nests in rock crevices in scree slopes where they lay a single egg clutch. Both sexes are similar in size (Jakubas & Wojczulanis 2007) and share parental duties until the female leaves the breeding site shortly before fledging and the male leads the chick to sea (Stempniewicz 2001; Harding et al. 2004). Little auks are specialized plankton feeders, with calanoid copepods constituting up to 90% of their diet (Mehlum & Gabrielsen 1993; Pedersen & Falk 2001). Food for the chick is brought back to the colony in an extensible gular pouch which can contain more than 3000 prey items (Karnovsky et al. 2003). During chick rearing, little auks use a bimodal foraging strategy in which they alternate single foraging trips of long duration (c. 18 h) with several short trips (c. 2 h; Welcker, Harding, Karnovsky et al., in press). During short trips, little auks are thought to mainly collect food for their chicks in near-shore areas. During long trips the birds may utilize distant foraging areas to replenish body reserves which are then depleted during short trips. Chick fledge at an age of 25–28 days, when they have reached about 70–80% of the adults’ body mass (Stempniewicz 2001; Harding et al. 2004).

Energy expenditure

We obtained estimates of daily energy expenditure of little auks with the doubly labelled water (DLW) method (Lifson & McClintock 1966; Speakman 1997; Butler et al. 2004). This method has been previously validated by comparison with indirect calorimetry in a range of small birds (Visser & Schekkerman 1999). In both seasons, measurements were taken during the chick rearing period when chicks were between 2 and 17 days old. Only adults with a live chick in the nest, as indicated by a full gular pouch, were used for the study. Birds were captured close to their nests with noose carpets, weighed to the nearest gram (Pesola spring balance) and individually marked by a unique combination of three colour plastic bands and a numbered steel band. We then injected each bird intraperitoneally with 0·26 mL of DLW containing 62·1 and 56·8 atom percent excess (APE) oxygen-18 (18O), and 33·9 and 41·1 APE Deuterium (2H) in 2006 and 2007, respectively.

Following the single-sample DLW technique (Webster & Weathers 1989), birds were released immediately after injection. The single-sample method avoids additional handling and bleeding of the study animals and therefore minimizes potential adverse effects (Amat et al. 2000; Weathers, Hodum & Blakesley 2001). A pilot study carried out at a little auk colony at Kap Höegh on the east coast of Greenland (Harding et al. 2009a) revealed that application of the commonly used double-sample DLW method (Speakman 1997), in which birds are held captive for c. 1 h and an initial blood sample is taken before release, resulted in very low recapture rates in this species. We therefore decided to use the single-sample technique to maximize our recapture probability and final sample size. In the course of this present study, 121 birds (2006: n = 61, 2007: n = 60) were injected of which we were able to recapture 43 individuals (36%) in time for 18O levels to be sufficiently elevated above background to allow for accurate estimation of DEE. Opportunistic observations of injected birds throughout the study period suggested that the treatment had no adverse effect on the behaviour of the birds, which continued to feed their chicks normally.

Between 24 and 72 h post injection, we opportunistically attempted to recapture injected birds by use of mist-nests and noose carpets. Little auks at the study site do not exhibit a diurnal activity pattern (Welcker et al. in press) and we, therefore, did not attempt to recapture birds precisely 24 h or a multiple thereof after injection. In theory, this can introduce a small increase in variance in the DLW estimates (Speakman & Racey 1988). Mean time elapsed between injection and recapture was similar in both study years (42·43 h ± 3·11 (SE) and 42·63 h ± 3·36 (SE) in 2006 and 2007, respectively). Collecting samples across periods exceeding 24 h reduces the error introduced because of high day-to-day variance in DEE (Speakman et al. 1994; Berteaux et al. 1996). Upon recapture, birds were weighed a second time and a blood sample taken from the brachial vein. Blood was collected into 2–4 70 μL glass capillaries which were immediately flame-sealed. An additional droplet of blood was collected for subsequent molecular sexing following standard techniques as detailed in Fridolfsson & Ellegren (1999). Furthermore, 6 individuals were captured and bled each year to determine mean background levels of isotopes (Speakman & Racey 1987: method C).

Isotopic enrichments of blood samples were determined by isotope ratio mass spectrometric analysis as described in Speakman & Krol (2005). Briefly, water from blood samples was vacuum distilled into glass Pasteur pipettes (Nagy 1983). 2H enrichment was determined from H2 gas derived from the distilled water from blood samples and produced by online chromium reduction of water. 18O analysis was performed by equilibration of water from blood with CO2 gas of known 18O isotopic enrichment using the small-sample equilibration technique (Speakman et al. 1990). 2H : 1H and 18O : 16O ratios were determined by a gas source isotope ratio mass spectrometer (IRMS) with isotopically characterized gases of H2 and CO2 in the reference channels. Four sub-samples of each blood sample were analyzed and mean values of enrichments were used for all subsequent analyses. Enrichment of the injectate was established by a dilution series with tap water and mass spectrometric analysis of 5 sub-samples of each solution (Speakman 1997).

Isotope enrichments were converted to values of daily energy expenditure using a single pool model, as recommended for this size of animal by Speakman (1993). There are several alternative approaches for the treatment of evaporative water loss in the calculation (Visser & Schekkerman 1999). We chose the assumption of a fixed evaporation of 25% of the water flux (equation 7·17: Speakman 1997) which has been established to minimize error in a range of conditions in studies of birds (Visser & Schekkerman 1999; Van Trigt et al. 2002). As we applied the single-sample method and did not obtain initial blood samples, we estimated initial isotope enrichments based on the relationship of body mass to body water established in an earlier study on little auks, by desiccation of carcasses [Body water (mL) = 7·70 + 0·589 body mass (g), R2 = 0·89, = 27; Gabrielsen et al. 1991]. We found a very similar relationship based on data from our pilot study in 2004 (Harding et al. 2009a). Standard errors of parameter estimates of the regression equation derived from these data [Body water (mL) = 18·72 + 0·504 body mass (g), R2 = 0·69, = 23] encompassed the parameter estimates of the equation reported in Gabrielsen et al. (1991), indicating that body water content varies only to a small extent in this species when sampled at a similar reproductive stage.

To convert estimates of CO2 production to DEE, we used a caloric equivalent of 27·97 J mL CO2−1. The caloric equivalent was calculated according to Gessaman & Nagy (1988) assuming a nutrient catabolism of 2/3 lipids and 1/3 protein, which corresponds to the nutrient composition of the main prey species found in diet samples obtained from little auks during this study (Scott et al. 2000; see below).


To quantify baseline plasma levels of corticosterone of little auks during the period of the DLW study, an additional 104 parent birds that were not used for DLW measurements were caught and bled using the methods described earlier. Blood samples (∼ 100 μL) were collected within 3 min of capture, the time usually taken for CORT levels to increase in response to a stressor (Romero & Reed 2005). In 10 cases, handling time exceeded 3 min and these birds were excluded from the data set. For the remaining samples (2006: = 38, 2007: = 56), we did not find a significant relationship between time after capture and plasma levels of CORT (F1,92 = 2·06, = 0·15, R2 = 0·02; mean capture-bleed time = 2·10 min ± 0·44 (SD), = 94). Thus our samples provided a measure of baseline circulating CORT and did not reflect the stress induced by the capture procedure.

To estimate maximum induced CORT levels, a second blood sample was obtained from 10 individuals each year at 30 min after capture, the time when plasma concentration of CORT normally peaks (Kitaysky et al. 2001; Kitaysky, Piatt & Wingfield 2007). Birds were held in cloth bags between first and second blood sampling and the handling procedure was standardized among birds. Blood samples were kept cool and centrifuged within 14 h after collection. Due to technical reasons, plasma was kept frozen at −20 °C in 2006 and preserved in 70% ethanol in 2007 until assayed for CORT. To verify the comparability of CORT measurements derived from both preservation methods, a validation study was conducted (see Harding et al. 2009b). In that study, we found a strong positive relationship between samples being frozen and preserved in ethanol [CORTfrozen (ng mL−1) = 0·0297 + 1·1329 CORTethanol (ng mL−1); R2 = 0·99, = 45·64, df = 23, < 0·001]. In addition, serial dilution of little auk plasma samples preserved in ethanol yielded displacement curves (between sample concentration and % bound to steroid antiserum) parallel to standard hormone and similar to frozen samples. These validations confirmed the effectiveness of ethanol as a field preservation method for little auk plasma samples. Prior to data analysis, we corrected values derived from samples preserved in ethanol using the regression equation stated earlier to allow for direct comparisons between years.

Corticosterone concentrations (ng mL−1) were measured using radioimmunoassays (for details, see Kitaysky, Piatt & Wingfield 2007). Each sample was equilibrated with 2000 cpm of tritiated CORT prior to extraction with 4·5 mL distilled dichloromethane. After extraction, percent tritiated hormone recovered from each individual sample (average hormone recovery was 92·7%, SD = 4·41) was used to correct final CORT concentrations. Samples (in duplicates) were reconstituted in PBSG buffer and combined with antibody and radiolabel in a radioimmunoassay. Dextran-coated charcoal was used to separate antibody-bound hormone from unbound hormone. All samples were analyzed in three different assays; sensitivity of the assays was 7·8 pg per tube, and inter- and intra-assay variations were less than 3·5% and 2%, respectively.

Adult return rate

To estimate the probability of parent little auks to return to the colony the following year, we conducted intensive re-sightings of all individuals that were injected and colour-marked in the course of the DLW study. Re-sighting was conducted for a total of 10 days in 2007 and 8 days in 2008. Re-sighting effort in both years was scheduled during egg-laying and close to hatching of chicks, with an additional re-sighting session before chicks fledged in 2007. Re-sighting curves reached a plateau in both years, indicating that within-year re-sighting probability was close to one.

Chick provisioning and fledging success

At the study site, little auks rearing chicks do not return to the colony without food for the chick (Welcker et al. in press). We therefore regarded the chick feeding rate to be reflected by the number of foraging trips individuals completed per day. To compare feeding rates between the two study years, we used data collected from birds additionally caught at the study site and equipped with VHF radio-transmitters (Holohil Systems Ltd., Carp, Canada; and Biotrack Ctd., Dorset, UK), the presence and absence of which was continuously and automatically recorded at the colony throughout the time period of the DLW measurements. During that time, 500 and 771 foraging trips were recorded in 2006 and 2007, respectively (2006: = 22, 2007: = 23 individuals).

Meal size (number of prey items and dry mass) was determined from whole food samples (2006: = 44, 2007: = 46 samples) obtained from little auks randomly captured at the study site at a mean chick age of 10·0 ± 1·1 (SE) days and 8·3 ± 0·9 (SE) days in 2006 and 2007, respectively. The largely undamaged prey items stored in the gular pouch were gently scooped out immediately after capture and preserved in 5% formalin. In the laboratory, prey items were counted, their length measured to the nearest 0·01 mm and identified to species and copepodid stage level. Dry mass of samples was calculated based on length-dry mass relationships of prey species from published literature (Falk-Petersen 1981; Scott et al. 2000; Auel et al. 2002; Werner & Auel 2005; Dale et al. 2006; Timofeev 2006).

Fledging success of chicks was estimated by monitoring the survival of chicks from hatching to fledging (2006: = 45, 2007: = 42 nests). We marked nests during incubation and determined hatch dates by regular nest controls every 2–3 days. We used light-scopes (Moritex Europe ltd., Cambridge, UK) to verify the presence of an egg or chick in the nest chamber. No sign of chick predation was observed in both years, and chicks that disappeared from the nest after 20 days old were therefore considered fledged (Harding et al. 2004).

Data analysis

We evaluated inter-annual differences in DEE, CORT, body mass and meal size by fitting linear least-square models. To control for differences between genders and changes across the chick-rearing period, ‘sex’ and ‘chick age’ were included as additional predictor variables in all models. As we did not know the age of the chick of each sampled individual, median age of chicks monitored for fledging success was used instead. Furthermore, we tested for and did not find an effect of time to recapture on DEE (F1,41 = 1·31, = 0·259) and therefore, did not include this factor in our models. Model fits were assessed by examining diagnostic plots; CORT data were log-transformed to meet condition of normality.

To examine the effect of body mass on our results, we re-ran models on mass-independent DEE. Mass-independent DEE was calculated as the residuals of the regression of loge DEE on loge body mass (R2 = 0·11, F1,41 = 5·17, = 0·028). To assess the effect of ambient temperature and wind speed on DEE in little auks, we obtained data (daily means) from a weather station at Ny-Ålesund (78°55′N, 11°56′E), c. 12 km from the little auk colony. Mean values of both parameters were calculated for the DEE measurement period for each bird and their effect on energy expenditure was evaluated by simple linear regression.

We assessed inter-annual differences in chick feeding rate by fitting generalized linear mixed-effects models with Poisson error distribution and logarithmic link function. ‘Year’, ‘sex’ and ‘chick age’ were applied as fixed effects, and ‘individual’ was included as a random component to avoid problems of pseudo-replication due to repeated measurements of the same individuals. We estimated the effect of energy expenditure on the probability of adults to return to the colony the following year using a generalized linear model with binomial error distribution and logit link function. A larger data set including all birds that were included in the DLW study (i.e. also individuals that were injected but not recaptured) was used to compare adult return rate between years. ‘Sex’ and ‘body mass at capture’ were included as additional predictors in this model.

In all cases, we started by fitting maximum models containing all predictor variables and interaction terms. Model simplification was done by likelihood ratio tests (LRT) eliminating variables and interactions, if their removal from the model did not result in a significant increase in deviance. Significance of terms in the most parsimonious model was assessed by F-tests and Wald-statistics, respectively. All statistical analyses were performed in R 2.7 (R Development Core Team 2008). Mean values are given ± 1 SE unless stated otherwise.


Energy expenditure

Energy expenditure differed significantly between the two study years (Fig. 2; F1,40 = 16·33, < 0·001). Mean DEE of chick-rearing little auks was 26·7% higher in 2007 (757·2 kJ day−1 ± 24·2, = 27) than that in 2006 (597·7 kJ day−1 ± 34·7, = 16). This effect was independent of body mass as mass-independent DEE was also significantly higher in 2007 compared with that in 2006 (F1,40 = 14·11, < 0·001). Furthermore, there was a positive relationship between DEE and chick age (F1,40 = 4·84, = 0·034). Energy expenditure increased similarly with chick age in both seasons (‘year’בchick age’: F1,38 = 0·05, = 0·827) at a rate of 8·52 kJ day−1 ± 3·87. DEE did not differ between sexes (F1,39 = 0·57, = 0·453).

Figure 2.

 Inter-annual differences in mean (± 1 SE) daily energy expenditure (DEE) and baseline plasma concentrations of the stress hormone corticosterone (CORT) in little auks. DEE was significantly lower in 2006 (light bars) compared to 2007 (dark bars) while CORT was significantly higher in 2006 than in 2007. Sample sizes are given in parentheses.

Inter-annual difference in DEE was not attributable to differences in wind speed or temperature. There was no difference in average wind speed (3·22 m s−1 ± 0·36 vs. 2·99 m s−1 ± 0·27; = 0·52, = 0·611) and only a small but statistically significant difference in ambient temperature between the two study years (5·80 °C ± 0·23 vs. 6·69 °C ± 0·24 in 2006 and 2007, respectively; = −2·65, = 0·013). Furthermore, we did not find a direct relationship between DEE and wind speed (F1,41 = 0·53, = 0·471, R2 = 0·01) or DEE and ambient temperature (F1,41 = 1·18, = 0·284, R2 = 0·02) at the time of the DLW measurements.

Mean body mass of injected birds was 2·3% higher in 2007 (163·0 g ± 1·47, = 60) compared with that in 2006 (159·4 g ± 1·28, = 61; F1, 117 = 3·77, = 0·054). Similarly, individuals that were recaptured and DEE estimated were 2·2% heavier in 2007 (161·7 g ± 1·90, = 27) than in 2006 (158·2 g ± 2·89, = 16), although due to lower sample size this difference was not statistically significant (F1,40 = 1·32, = 0·257). In both years, females weighed significantly less than males (F1,118 = 9·02, = 0·003).


Circulating baseline levels of CORT differed significantly between study years (Fig. 2; F1,92 = 10·25, = 0·001). Baseline CORT was substantially lower in 2007 with high energy expenditure (2007: 1·75 ng mL−1 ± 0·10, = 83) and elevated in 2006 with low DEE (2006: 2·38 ng mL−1 ± 0·19, = 51). Similarly, maximum induced CORT levels were significantly higher in 2006 (22·30 ng mL−1 ± 2·60, = 9) than in 2007 (8·61 ng mL−1 ± 1·00, = 10; F1,17 = 26·23, < 0·001). We did not find a significant relationship between baseline CORT and chick age or baseline CORT and sex (‘chick age’: F1,91 = 3·09, = 0·082; ‘sex’: F1,90 = 0·81, = 0·370).

Chick provisioning and fledging success

Chick feeding rate was higher and meal size was larger in 2007, the year with high energy expenditure. On average, chicks were fed 15·8% more meals per day in 2007 compared with that in 2006 (Table 1; Wald statistic = 2·96, = 0·003). Similarly, chicks received on average larger meals in 2007 than in 2006, both in number of prey items (Table 1; F1,87 = 6·82, = 0·011) and in dry mass (Table 1; F1,87 = 13·87, < 0·001). Although chick feeding rate did not vary in relation to chick age, meal size increased as chicks grew older (items: F1,87 = 31·89, < 0·001; dry mass: F1,87 = 10·80, = 0·001). This increase in meal size was similar in both study years (‘year’בchick age’: F1,87 < 1·1, > 0·3). Higher year-specific provisioning rate was associated with increased chick survival: fledging success was significantly higher in 2007 than in 2006 (Table 1; Fisher’s Exact test: = 0·015).

Table 1.  Inter-annual differences in mean (± 1 SE) food provisioning rate (foraging trips per day), chick meal size (items and dry mass), chick fledging success (i.e. the probability of chicks to survive from hatching to fledging) and adult return rate (i.e. the probability of parent birds to return to the colony the year following measurements) of little auks at Kongsfjorden, Norway
  1. Sample sizes are given in parentheses. All parameters differed significantly between study years (see text for details).

Feeding rateTrips day−12·81 ± 0·14 (210)3·25 ± 0·12 (245) 
Meal sizeItems1228 ± 92 (46)1462 ± 117 (44) 
Dry mass (g)0·76 ± 0·05 (44)0·99 ± 0·06 (46) 
Fledging success(%)80·0 (45)97·6 (42) 
Return rate(%) 78·7 (61)93·3 (60)

Adult return rate

There was no effect of energy expenditure on the probability to return to the colony the following year (Wald statistic = −0·60, = 0·552, = 43). However, adult return rate was significantly lower following the year with low energy expenditure (2007: 78·7%) compared to the year with high energy expenditure (2008: 93·3%, Table 1; Wald statistic = 1·98, = 0·047, = 121). Moreover, return rate of little auks was significantly positively related to body mass at capture (Fig. 3, Wald statistic = 2·46, = 0·014). This relationship did not differ between years (LRT, χ2 = 0·02, = 0·892). Little auks that did not return to the colony were on average 5·0% lighter during the preceding season than birds that were re-sighted (154·6 g ± 2·2 vs. 162·3 g ± 1·0).

Figure 3.

 Relationship between body mass of little auks during the chick rearing period and the predicted probability of their return to the colony the following season. Filled symbols represent body mass of birds that returned, open symbols birds that did not return to the colony. Predicted probabilities are based on a generalized linear model with binomial error structure (see text for further details).


In this study, we examined whether little auks rearing chicks responded to changes in food availability by increasing or reducing their rate of energy expenditure, i.e. whether the regulation of DEE reflects enabling or forcing effects of the environment. We used plasma concentrations of the stress hormone corticosterone as an indirect proxy for food availability. In addition, we examined fitness consequences of changes in DEE in terms of reproductive success and adult return rate. We found that rates of energy expenditure were elevated in the year where birds exhibited low levels of CORT. This suggests that high levels of food supply from the environment resulted in increased metabolic rates, supporting the hypothesis that favourable food conditions enable high DEE in this species. Furthermore, consistent with predictions from the enabling hypothesis, the elevated level of DEE coincided with increased parental effort, as indicated by a higher chick feeding rate and larger chick meals, and was associated with higher current reproductive success. Contrary to expectations from several previous studies (Wolf & Schmid-Hempel 1989; Bryant 1991; Daan, Deerenberg & Dijkstra 1996), high levels of energy expenditure were not related to a reduced probability of adults returning to the colony the following year. Our results therefore indicate that high DEE in little auks resulted in fitness benefits during the current reproductive episode, and was not associated with fitness costs in terms of a reduced opportunity to reproduce during the subsequent breeding season.

Measuring food availability in marine systems is notoriously difficult and in a species like the little auk, with a potentially large foraging range and unknown feeding areas (Welcker et al. in press), prone to large error. Plasma levels of the stress hormone corticosterone co-vary with food availability as has been demonstrated empirically (Kitaysky, Wingfield & Piatt 1999b; Kitaysky, Piatt & Wingfield 2007) and experimentally (Kitaysky et al. 1999a; Pravosudov et al. 2001; Lynn, Breuner & Wingfield 2003; Clinchy et al. 2004; Schoech, Bowman & Reynolds 2004) in many bird species including the little auk (Harding et al. 2009b; N. Karnovsky, unpublished data). The study of both baseline concentrations and maximum stress-induced levels of CORT offer the opportunity to obtain information on short-term and persisting changes in food stress (Kitaysky et al. 2005; Kitaysky, Piatt & Wingfield 2007). Our data strongly suggest that the amount of food available to little auks in the study area differed largely between the two study years. This disparity was visible in both baseline and maximum levels of CORT, indicating that the reproductive period 2006 was characterized by persistently inferior food availability compared with that in 2007. This view is supported by the fact that the rate of chick provisioning was reduced in the year with high CORT levels, a response to poor food availability typical for seabird species (Hamer et al. 1993; Uttley et al. 1994).

We applied the single-sample DLW method to estimate DEE. This method depends on accurate estimation of the total body water (TBW) of sampled individuals to lead to reliable estimates of DEE (Speakman 1997). To rule out the possibility that the inter-annual difference in DEE we found in this study was driven by differences in TBW of individuals between years rather than by differences in the rate of energy expenditure, we recalculated DEE assuming a 5% difference in TBW between years. Mean estimated DEE in 2007, assuming a 5% lower body water content than in 2006, was still significantly higher than DEE in 2006 (698·13 kJ day−1 ± 22·51 vs. 597·67 kJ day−1 ± 34·71; F1,40 = 7·077, = 0·011), confirming that inter-annual difference in DEE was not attributable to a possible difference in TBW.

In addition, inter-annual variation in DEE was not explained by other potentially confounding factors, such as ambient temperature and wind speed, which have been shown to affect energy expenditure in free-ranging birds (Gabrielsen et al. 1991; Furness & Bryant 1996; Bryant 1997). We observed a positive relationship between energy expenditure and chick age. Similar results have been reported for kittiwakes (Fyhn et al. 2001) and may indicate an increase in parental effort in response to increasing energetic needs of the offspring. This relationship was similar in both study years and did not affect inter-annual differences in DEE.

We therefore suggest that little auks regulate their rate of energy expenditure in response to food availability, with high levels of food availability enabling an increase in DEE. The observed shift in DEE is thereby likely to be linked to alterations in the foraging and chick provisioning behaviour. Meal delivery rate was on average 15·8% higher during the year with elevated DEE. Little auks use a bimodal foraging strategy during the chick rearing period (Steen et al. 2007), in which they alternate several short trips to provide food for their chick with a single long trip used primarily for self-feeding (Chaurand & Weimerskirch 1994; Weimerskirch 1998). It has been shown that energetic costs of the two foraging trip types differ considerably. Despite shorter travel distances, energy demands per time of short trips exceed those of long trips by more than 100% (Weimerskirch et al. 2003). Little auks increase their chick provisioning rate mainly by reducing the duration of long trips and increasing the frequency of short trips (Welcker et al. in press), and hence changes in time allocation to the different trip types may proximately explain the observed between-year differences in DEE in little auks.

Our results indicate that little auks used the excess energy available in 2007 to increase both allocation of resources to self-maintenance, as suggested by slightly higher mean body mass, and parental effort. Based on estimates of food delivery rate and chick meal size, offspring received c. 38% more food per day in 2007 than in 2006, thus enhancing the probability of the chicks to fledge successfully. Therefore, elevated DEE was ultimately associated with short-term fitness benefits in terms of increased success in the current reproductive episode. However, in contrast to earlier studies, we did not find evidence for a direct cost of elevated DEE in the little auk. Adult return rate was not directly related to energy expenditure during the preceding reproductive period, and a higher between-year probability to return coincided with high rates of energy expenditure. Although failure to return to the colony may not only imply mortality, it is likely to be indicative of fitness costs as (temporary) emigration from the colony (Brown, Brown & Brazeal 2008) or a skipped breeding opportunity also results in reduced reproductive output.

Several studies have suggested a long-term fitness penalty due to increased mortality as a trade-off to increased DEE (Wolf & Schmid-Hempel 1989; Bryant 1991; Daan, Deerenberg & Dijkstra 1996). However, the ‘free radical damage hypothesis’ (Harman 1956) proposed as a proximate explanation has in recent years been called into question (reviewed in Speakman 2005). First, our increasing knowledge of the stoichiometry of free-radical production in mitochondria strongly suggests that elevated free-radical production should not be a direct consequence of increased energy expenditure (Brand 2000). Indeed, individual mice that have higher levels of DEE and greater levels of mitochondrial uncoupling live longer (Speakman et al. 2004). Moreover, experimental elevations of DEE in field voles (Microtus agrestis) in a protected environment resulted in no increase in oxidative damage and no decrement in survival (Selman et al. 2008). Taken together, these studies suggest that the main impact of elevated DEE on future survival is not due to a physiologically driven process such as elevated free-radical damage but rather depends on the ecological context of the increased DEE (see also Speakman 2008). Our data are in line with this hypothesis; an increase in DEE did not inevitably lead to reduced future survival. However, it needs to be noted that our data were restricted to two study years and additive costs of increased energy expenditure over longer time periods may have remained undetected. The probability to return to the colony was positively related to body mass. This suggests that the ability of little auks to maintain energetic balance and a sufficiently high body mass during reproduction was impaired when food was limited. This may have led to a higher risk of starvation during the post-breeding period and to carry-over effects resulting in a reduced probability to initiate a breeding attempt the following season. These results indicate that DEE may not be intrinsically linked to future survival, but its implications for energy balance and other ecological factors may be.


Many thanks to L. Borg, S. Christensen-Dalsgaard, J. Delingat, S. Natterer, J. Schultner and N. Seifert for their invaluable help in the field, and the staff of the Sverdrup Station (Norwegian Polar Institute) in Ny-Ålesund, especially W. Moskal, for logistic support. We are grateful to P. Thomson and P. Redman for technical assistance with isotope analysis, and to B. Moe and E. Noreen for their help with molecular sexing. JW and GWG were supported by the Research Council of Norway (MariClim, 165112/S30) and Svalbard Science Forum. AMAH was funded by the National Science Foundation (grant 0612504) and supported by USGS-Alaska Science Center. We thank R. Ims and two anonymous reviewers for their helpful comments on an earlier draft of the manuscript. All fieldworks were conducted with the permission of the Governor of Svalbard.