Quantifying a relationship between heart rate (fH) and rate of oxygen consumption(O2) allows the estimation of O2from fH recordings in free-ranging birds. It has been proposed that this relationship may vary throughout an animal's annual cycle,due to changes in physiological status. Barnacle geese, Branta leucopsis, provide an ideal model to test this hypothesis, as they exhibit significant intra-annual variability in body mass, body composition and abdominal temperature, even in captivity. Heart rate data loggers were implanted in 14 captive barnacle geese, and at six points in the year the relationship between fH and O2 was determined. The fH/O2relationship was also determined in seven moulting wild barnacle geese to examine whether relationships from captive animals might be applicable to wild animals. In captive barnacle geese, the fH/O2relationship was significantly different only between two out of the six periods when the relationship was determined (late September–early October and November). Accounting for changes in physiological parameters such as body mass, body composition and abdominal temperature did not eliminate this difference. The relationship between fH and O2 obtained from wild geese was significantly different from all of the relationships derived from the captive geese, suggesting that it is not possible to apply calibrations from captive birds to wild geese. However, the similarity of the fH and O2 relationship derived during moult in the captive geese to those during the remainder of the annual cycle implies it is not unreasonable to assume that the relationship between fH/O2during moult in the wild geese is indicative of the relationship throughout the remainder of the annual cycle.

Efficient use of energy is a key determinant of fitness(Brody, 1945; Kleiber, 1961; Butler et al., 2004). Therefore, information about the energy expenditure of free-ranging animals holds the key to understanding major aspects of their behaviour(Nagy et al., 1978; Nagy et al., 1984; Froget et al., 2001; Tolkamp et al., 2002). Using heart rate (fH) as an indicator of the rate of oxygen consumption (O2)has proved to be an increasingly popular tool when investigating the energy expenditure of free-ranging animals(Malhotra et al., 1963; Owen, 1969; Holter et al., 1976; Flynn and Gessaman, 1979; Gessaman, 1980; Butler, 1991; Butler, 1993; Bevan et al., 1995; Green et al., 2001; Butler et al., 2002; Froget et al., 2002). The method is based on the relationship between O2 and fH and relies on accurate calibrations between the two(Butler, 1993). The ideal approach would be to calibrate this relationship in each animal that is to be used in the field. However, the disturbance associated with calibration, the difficulty of obtaining animals and the potential interference with selected individuals usually makes this impractical(Green et al., 2001; Green and Frappell, 2007). In most cases, therefore, this relationship has been established in one group of animals and then applied to a different group of free-ranging animals, taking into account the associated inter- and intra-individual variability (e.g. Froget et al., 2001; Green et al., 2001). Most animals used in calibration experiments have been free-ranging individuals from the same population to minimise potential sources of error(Green et al., 2001; Green et al., 2005). However,studies on wild animals may want to reduce the number of individual animals used, particularly if populations are of conservation concern. It may be possible to use captive animals from zoos or other sources for calibration experiments but differences in metabolism have been identified between wild and captive birds (McKechnie et al.,2006), suggesting that this possibility needs to be explored further.

The usefulness of the fH method depends on the established relationship with O2 being similar under different conditions or at least for any differences to be quantified and taken into account (Butler et al.,2002). For example, Froget et al. found that the relationship between O2 and fH in king penguins (Aptenodytes patagonicus) was influenced by variation in body condition and season(Froget et al., 2001). Similarly, Holter et al. noted that the relationship was influenced by season in white-tailed deer (Odocoileus virginisanus)(Holter et al., 1976). It appears, therefore, that determination of the fH/O2relationship when animals are in different physiological states is essential,and may explain some of the temporal differences observed in a number of studies (e.g. Holter et al.,1976; Gessaman,1980).

The fH/O2relationship has been previously determined in captive barnacle geese Branta leucopsis (Nolet et al.,1992; Ward et al.,2002) (see also Butler and Woakes, 1980). For this species, there was no difference in calibrations between fH and O2 carried out on different individuals, 10 years apart(Nolet et al., 1992; Ward et al., 2002). However,the experiments in both studies were performed at similar times in the annual cycle. Recent work on captive barnacle geese has demonstrated that they undergo changes in body mass, body composition and abdominal temperature similar to those of wild geese (Portugal et al., 2007; Portugal,2008; Portugal et al.,2009). For example, despite ad libitum access to food,captive geese lost approximately 25% of their body mass during their annual wing moult, while simultaneously experiencing an increase in the resting rate of metabolism of approximately 80%(Portugal et al., 2007). These findings are comparable with those found for wild moulting waterfowl (see Hohman et al., 1992). Therefore, captive barnacle geese present an ideal model to investigate how naturally occurring seasonal changes in physiological variables such as body mass, body composition and abdominal temperature may interact and influence the relationship between fH and O2. We hypothesised that any intra-annual variability in the fH/O2relationship could be attributed to quantifiable variation in one or more of these traits.

The primary aim of this work was to construct a model from physiological variables to predict O2 in the geese. More specifically, we aimed to: (1) calibrate the relationship between fH and O2 at points of significance during the annual cycle of a species that shows substantial intra-annual variability in several traits; (2) quantify any variability in the fH/O2relationship between seasonal periods, through the inclusion of physiological variables such as body mass, body composition and abdominal temperature; and(3) calibrate the relationship between fH and O2 in wild barnacle geese, and compare the relationship with that of captive geese.

### Captive birds

A captive population of 44 barnacle geese (Branta leucopsisBechstein) obtained as three-week-old goslings was maintained under natural light in a large outdoor aviary (30 m×20 m) at The University of Birmingham, UK. The goslings were obtained from Bentley Wildfowl Park (Sussex,UK), which has held a self-sustaining population of this species since 1982. The birds were two years old at the time of study. The geese were fed with a 50–50 diet (Lilico, Surrey, UK) of mixed poultry corn (4% fat, 12%protein and 71% carbohydrate) and poultry growers pellets (3% fat, 16% protein and 61% carbohydrate). Food and water were available ad libitum. All regulated procedures were performed by UK Home Office licensed personnel in possession of a Personal License, and working under the auspices of a corresponding Project License, as set out in the Animals (Scientific Procedures) Act 1986.

Fourteen birds out of the flock of 44 were trained to walk on a treadmill for 2–3 weeks prior to the first experiments to build up physical condition and become fully accustomed to the experimental surroundings and the handler. Training consisted of walking the birds at previously determined speeds for 2–3 h a week (see Ward et al., 2002). Sampling periods were selected based on year round body mass data, collected from the same captive flock of barnacle geese the previous year (see Portugal et al.,2007). Points chosen included the maximum and minimum body masses recorded in the year, and periods of body mass increase and decrease, in particular, those associated with wing moult. Sampling periods were,therefore, February, May, late July–early August, mid August–early September, late September–early October and November 2006 (see Table 1 for sample sizes). Sample sizes varied for each respirometry period, as the two periods during moult were very close to each other. At this time, we chose to maximise the number of seasonal periods rather than the number of birds used in all sampling periods, and so not all birds were used in all periods.

### Moult score

A 5-point classification moult score system was used to assign the stage of moult in both the wild and captive geese (e.g. Bridge, 2004; Portugal et al., 2007). Moult score was defined as: (1) pre-wing moult; (2) primaries and secondaries remain, new primary pin visible; (3) all primaries and secondaries missing;(4) new primaries visible well beyond primary coverts and secondaries visible beyond secondary coverts (small flight feathers covering the primaries and secondaries); and (5) post wing-moult.

### Respirometry

In order to minimise the duration of the sampling periods, two respirometry systems (equipment from ADInstruments, Colorado Springs, CO, USA) were used simultaneously to record O2. Full information and particulars on the respirometry set-up, equipment and analysis, is given in detail in Portugal et al.(Portugal et al., 2007) (see also Green et al., 2001; Wilson et al., 2006). The temperature for all respirometry experiments was within the thermoneutral zone for barnacle geese (Ward et al.,2002).

Prior to the walking periods, each goose was rested overnight between the hours of 23:00 h and 07:00 h (GMT) to measure resting O2 (see Portugal et al., 2007). Food(not water) was withheld 8 h prior to placing the birds in the respirometer for overnight experiments. Water was offered to the birds once the overnight sampling period was complete and before walking began. Birds were always walked between 07:30 h and 10:30 h (GMT) for each period. Random number tables were used to determine the order of treadmill speeds so that the geese did not become accustomed to a repeated pattern of speeds. Each bird was exercised at up to seven different speeds (0.5–3.5 km h–1). This speed range included the lowest speed available on the treadmill and the highest speed the birds could maintain for sufficient time to allow for gas equilibration (10–20 min). O2 was calculated as a 5 min average of the steady-state values observed. Birds were monitored constantly, and data were excluded if the birds showed evidence of fatigue and were unable to maintain station within the respirometer for a sufficient length of time to allow for gas equilibration. On some occasions,gas equilibration did not occur at 0.5 km h–1 due to the birds not walking at a constant speed. When this occurred, the data were not included in further analyses. All geese that were involved in respirometry experiments during the late July–early August period were observed to have a moult score between 2 and 4 and were considered to be in peak wing moult.

### fH

fH and abdominal temperature were measured using a miniature heart rate data logger (HRDL), which has been successfully used with this species on previous occasions (Butler et al., 1998; Ward et al.,2002). HRDLs (55 mm×25 mm×7 mm) were programmed to record fH every 5 s. The implantation procedure and further information on the HRDLs are given in detail elsewhere (e.g. Butler et al., 1998; Green et al., 2005).

### Physiological variables

Five other physiological and morphological variables were recorded alongside fH and O2 in each sampling period. Each variable was measured from the same individual geese and taken at the same time as the respirometry experiments. The variables were body mass (BM), lean body mass (LBM), total body fat (TBF), structural size(SS) and abdominal temperature (Tab). Tarsus length was used as an indicator of SS, as previous studies on wild barnacle geese have shown it to be the best indicator of SS, and a principal component analysis showed tarsus length in the captive geese to be significantly correlated with other structural body parts (Larsson et al., 1998; Portugal,2008). TBF and LBM were estimated using the isotope dilution technique (Speakman et al.,2001), which has previously been shown to be a good measure of TBF content in birds (Kerstel et al.,2006; Eichorn and Visser, 2008) (for details, see Portugal, 2008).

### Wild birds

Data were obtained from seven adult barnacle geese of the breeding population at Ny-Ålesund, in the high Arctic (79°55′,11°56′E) on the west coast of the island of Spitsbergen. The birds were captured in corral nets during the post-breeding season moult, during the last two weeks of July and the first week in August when they were flightless. Geese were weighed and housed outdoors as a group, with food and water available ad libitum.

### Respirometry

The same respirometry system used for the captive birds was used to measure O2 in the wild geese, with the following modifications. Air was pulled by a downstream sealed pump at a flow of about 30 l min–1, controlled by a mass flow controller (Aalborg, Orangeburg, NY, USA). The excurrent air was sub-sampled(100 ml min–1), passed through a drying column and analysed by the O2 and CO2 analysers. The recorded traces for O2 and CO2 were averaged every 5 s, corrected to instantaneous values as if the system had been sealed assuming a first-order linear system (see Frappell et al.,1989) and subsequently averaged every 2 min. Calculation of O2 was determined as previously described [see Appendix in Frappell et al.(Frappell et al., 1992)]. As with the captive geese, birds were held in the respirometer overnight to measure resting rates of metabolism(Portugal et al., 2007). At 08:00 h (EST) birds were offered water, before commencement of walking on the treadmill. The protocol for walking, including the range of speeds, was the same as that for the captive geese.

### Heart rate

fH was monitored using a customised heart rate transmitter system (25 g, POLAR a3, Polar Electro Oy, Finland). The heart rate transmitter was attached dorsally to the feathers using lightweight paper tape, and custom-made brass electrodes were inserted under the skin. The transmitter unit had a functional range of approximately 1 m, and the receiver unit was placed on top of the respirometer to ensure reception. The outputs were collected at 200 Hz (Powerlab, ADInstruments).

### Statistical analysis

As stated, the overall aim was to construct a single model of physiological variables (fH, BM, TBF, LBM, SS and Tab) to predict O2 in the geese throughout their annual cycle. Captive geese were examined first, using general linear mixed models (GLM) (Zar,1984) in Minitab to establish the relationship between fH and O2 within each of the six sampling periods.

To compare sampling periods, a novel approach was adopted that accounted for the unbalanced design, which resulted from having different sets of individuals in some periods. The slope of the relationship between fH and O2 for each individual in each sampling period was treated as the dependent variable. We examined the effects of sampling period (fixed effect) with goose ID as a random factor and the physiological variables (fH, BM,LBM, TBF, SS and Tab) as covariates. A model of best-fit was achieved by firstly including all possible main effects and all possible second degree interactions in the model. Each non-significant term was then removed, one iteration at a time, by backwards elimination(Green et al., 2005). Non-significant main effects were kept in the model if the variable in question was part of a statistically significant interaction. Our objective was to create a single model in which both sampling period and the interaction between sampling period and slope were not significant effects.

Five sets' of models were investigated. Model set one used the slopes from the relationships between fH and O2, firstly using whole animal' values of O2 and subsequently O2divided by BM (kg), LBM (kg), TBF (kg) and SS (tarsus length, mm). Model set two repeated this analysis but, in this case, each of these measurements(except SS) was also divided by tarsus length as an indicator of structural size, and re-run with O2 divided by the resultant structurally corrected value. In model sets one and two, a variable was not included as a covariate if it had been used to correct O2. Model set three examined the exponent by which values of O2 were mass corrected. In model sets one and two, values of O2 were corrected to an exponent of 1. However, data from experiments studying resting O2 in the captive geese (Portugal et al.,2007) showed that this exponent may change throughout the season. Therefore, this set re-constructed every analysis from model sets one and two and re-analysed them with mass exponents ranging from 0.5 to 1.5 inclusive in increments of 0.1. Model set four took the best model from sets one to three and investigated cases where differences existed between sampling periods. These post-hoc comparisons were made by comparing each sampling period in a pairwise fashion, with the appropriate sequential Bonferonni corrections applied (Holm,1979; Shaffer,1986). Those sampling period comparisons, which indicated no significant interaction between sampling period and fHwere pooled into groups, until the calibration regression lines from all sampling periods within one group were not significantly different from each other. Finally, model set five compared the calibration regression lines from the captive geese with that of the wild barnacle geese, studied during their annual wing moult in mid July–early August. For all statistical tests used α=0.05, with α adjusted for the number of tests when applying Bonferroni corrections.

GLM revealed fH and O2 to be significantly related in each of the individual sampling periods of the captive geese (Fig. 1, Table 2). Considerable variation could be observed in both slope and intercept and the range of both fH and O2 values between the different sampling periods.

### Model set one

This set of analyses aimed to determine what correction factor might be applied to O2 to improve the relationship with fH and examined differences between sampling periods. Each analysis in this set was run with either whole-animal values of O2 or O2 corrected by BM, LBM, TBF, Tab or SS. Of note was: (1) in four of the models, there was a non-significant interaction between fHand BM but BM remained a significant main effect in all of the models; and (2)none of the GLMs was successful in removing sampling period as a significant effect.

### Model set two

This set of analyses also incorporated whole-animal values of O2 along with values of O2corrected to BM, LBM and TBF. However, in this instance, each value of BM, LBM and TBF had in turn been corrected according to structural size by dividing by tarsal length. Each analysis in model set one was thus re-run with the structurally corrected values of BM, LBM and TBF applied to correct O2. Key results from this set of analyses were: (1) structurally correcting values of BM, LBM and TBF had no significant effect on the results from model set one; and (2)thus, none of the models from this set of analyses was successful in removing sampling period as a significant main effect.

### Model set three

This set examined the exponent by which values of O2 were mass corrected, by re-analysing all of the models from sets one and two with mass exponents ranging from 0.5 to 1.5 (inclusive) in increments of 0.1. Of note was: (1) as with model set two, altering the mass exponent did not change the overall results nor which main effects were significant nor the significant interactions; and (2) in summary, changing the mass exponent did not improve the accuracy (based on the R2) of the model nor did it remove sampling period as a significant interaction.

At this point, none of the analyses had been successful in removing sampling period as a significant influence on O2, and so it was not possible to create one single calibration line that could be constructed from all six sampling periods and applied to year round data sets. For further analysis, O2 corrected for body mass (mass-specific O2) was chosen as it had the greatest effect on reducing variability between sampling periods. The model with mass corrected O2 had the highest R2 and this approach was the most parsimonious, in that it included the least additional variables and hence error terms(Fig. 2; Table 2).

### Model set four

This set compared the slope of the relationship between fH and mass specific O2 between sampling periods in a pairwise fashion. The purpose of this analysis was to derive the fewest sampling periods that were not significantly different from one another. The key results of these analyses were: (1) only late September–early October and November (regression lines 5 and 6 in Fig. 2) were significantly different from each other; and (2) the result of these analyses was two regression lines, late September–early October, and then a second regression line incorporating February, May, late July–early August, mid August–early September and November(Table 2). All regression lines within this group were not significantly different from each other, so the data were combined.

### Model set five

This final stage of analysis compared wild and captive geese. There was a significant relationship between fH and O2 in the wild geese, and a simple GLM model (Table 2) revealed fH to explain 67% and 60% of the variance for whole-animal and mass-specific O2,respectively. The slope of the relationship between fH and O2 for each individual in each period for the captive geese was compared to that for the wild birds. For this, mass-specific O2 was used again (Fig. 3). The wild goose data were then also compared in the same manner with the two regression lines constructed on the captive geese in model set four. Key results from this analysis were: (1) all of the regression lines from the six sampling periods in the captive geese were significantly different from that of the wild birds;(2) the two regression lines constructed in model set four were also significantly different from that of the wild geese; and (3) using whole-animal values of O2 did not change these conclusions.

### Using fH to predict V̇O2 in barnacle geese

Many authors have previously highlighted the need to establish the relationship between fH and O2 at various stages throughout the annual cycle (e.g. Bevan et al., 1995; Green et al., 2001; Butler et al., 2002). Here,for the first time, we provide this relationship from six points in the annual cycle, for a species that exhibits significant seasonal changes in body mass,body composition and abdominal temperature, even when in captivity(Portugal et al., 2007; Portugal, 2008; Portugal et al., 2009).

For each sampling period in the captive barnacle geese, with both whole-animal and mass-specific values of O2, there was a significant relationship between fH and O2, and the amount of variability explained by the GLMs (70%) is consistent with that found by Ward et al. (Ward et al.,2002). Apart from the mid August–early September sampling period (Table 2), including the effect of body mass by calculating mass-specific O2, as opposed to whole-animal, values greatly improved the accuracy of the models. During mid August–early September, the captive geese are completing their annual wing moult, resulting in different individuals being at slightly different stages of moult completion and thus in different physiological states. Therefore, the effect of body mass and body fat on the fH/O2relationship may be more complex then that at other points in the annual cycle.

The suggested effects of body fat (adipose tissue) on rate of energy expenditure during exercise are somewhat contradictory. Fat is metabolically inactive but can elevate overall metabolic rate by increasing the energy demands of other tissues, i.e. energy is required from lean tissues to heat and transport the fat tissue (e.g. Froget et al., 2001; Nudds and Bryant, 2001), and also increase the cost of locomotion (the net cost of transport). However, our results suggest that the changes in the cardiovascular system in response to gains in body fat may be systematic, as the relationship between fH and O2 did not change significantly between five out of the six of the sampling periods,despite significant changes in body mass, body composition and total body fat(Portugal et al., 2007; Portugal, 2008; Portugal et al., 2009).

Rates of O2achieved at high speeds during treadmill exercise in late September–early October are greater than those recorded at any other point in the year in the captive geese. During strenuous exercise, both running and flying, birds will dissipate excess heat from the legs and feet or through panting (Bech and Nomoto,1982). Froget et al. noted that during treadmill walking, king penguins were exposed to an unnatural situation, i.e. performing intense exercise in their thermoneutral zones or possibly even above it(Froget et al., 2002). Therefore, it could be possible that although the captive geese were in their thermoneutral zone, running in the treadmill experiments in late September–early October, when they have the highest relative body percentage of fat during their annual cycle(Portugal, 2008), caused the birds to be heat stressed, resulting in a change in the relationship between fH and O2.

Alternatively, in wild barnacle geese, Bishop et al. suggested that a selective perfusion of tissues vital for support of locomotor activity could possibly occur as a physiological strategy to maximise flight performance(Bishop et al., 2002). Captive geese may undergo the same physiological change so that during late September–early October they are better suited for flight as opposed to leg-powered locomotion. As a result, the potential increase in blood flow to the legs and feet to dissipate heat during treadmill exercise may influence the overall cardiovascular system and, in turn, the relationship between fH and O2.

### Comparing captive and wild geese

The fH/O2relationship in the captive geese that is most similar (although still significantly different) to that of the wild geese is the relationship determined during late September–early October. This was also the only calibration equation in the captive geese that was significantly different from that of any other sampling period (November). Late September–early October is the migratory period of the wild birds. Much evidence gathered from studying captive geese suggests they undergo similar patterns of physiological and behavioural change to those of their wild conspecifics(Butler and Woakes, 2001; Portugal et al., 2007; Portugal, 2008; Portugal et al., 2009). However, the difference in the fH/O2relationships between the captive and wild geese suggests though that there is a limitation to the similarities between them nonetheless, which is likely to be linked to the difference in physical fitness between the two. However, it may be possible that in the autumn, the captive geese are also undergoing similar adjustments to their cardiovascular system in preparation for their`autumn migration' as the wild geese would be undertaking(Butler et al., 1998). Just prior to their autumn migration, wild barnacle geese exhibit hypertrophy of their hearts in preparation for the long migratory flight southwards to their wintering grounds (Bishop et al.,2002). Data on year round body mass, body composition and abdominal temperature suggest these factors at least are responding in the same manner as wild birds in the autumn period(Butler and Woakes, 2001; Portugal et al., 2007; Portugal, 2008; Portugal et al., 2009).

Regardless of what correction factor was applied to O2, the significant difference in the fH/O2relationship between captive and wild barnacle geese persisted. The wild moulting barnacle geese had significantly lower fH for a given O2. This was corroborated by data from overnight resting O2 and fH experiments. Resting O2 during moult was not significantly different between captive and wild geese. However,resting fH was significantly lower in the wild geese (50 beats min–1) in comparison with 75 beats min–1 in the captive geese(Portugal, 2008). This is more than likely related to the greater overall physical fitness of the wild geese in comparison with their captive conspecifics. Generally, the more athletic,or fitter, the individual, the lower its heart rate for a given O2, both during rest and exercise (Henderson et al.,1927; Bock et al.,1928; Butler and Turner,1988; Bjornstad et al.,1993). As fitness improves, the heart enlarges, ventricular stretching is enhanced and blood volume increases, resulting in an increased stroke volume, allowing a reduction in the heart rate(Beswick and Jordan, 1961; Margaria, 1963; Butler, 1993; McPhee et al., 2003). Butler and Turner compared various morphological aspects of the cardiovascular and locomotor system in trained and untrained captive tufted ducks, Aythya fuligula (Butler and Turner,1988). Trained tufted ducks had increased maximal swimming velocities, lower heart rates, higher oxygen pulses, more capillaries in their locomotor muscles, and these capillaries were present at a higher density. Maximal O2(O2,max) was greater in the trained ducks but there was no change in maximum fH. The greater O2,max was the result of larger cardiac stroke volume in the trained birds, giving greater cardiac output, and increased oxygen extraction related to the increase in capillary density. Pelsers et al. also noted that intracellular transportation of fatty acids from the capillary network to the mitochondria in the muscles is better developed in wild barnacle geese(Pelsers et al., 1999). Long-term captive geese had lower concentrations of fatty acid binding proteins and citrate synthase in the pectoral muscle than those of wild pre-migratory geese (Pelsers et al.,1999).

In addition to this difference, wild birds had much higher O2 than captive birds during exercise, despite the range of speeds being identical in the two sets of experiments. We suggest that this may be due to the wild birds being unfamiliar with the experimental setup and unaccustomed to being handled or running on the treadmill. As a result the wild birds expended more energy to move at the same speeds as the captive birds. The captive birds each had several training sessions, and observations of the exercise sessions suggested that the wild birds were less likely to run rhythmically or consistently than the captive birds. This lack of familiarity and inconsistency in gait may also explain the lower R2 values for the fH/O2relationships in wild birds (Table 2) (Butler and Turner,1988; Bishop and Bulter,1995).

### Application to free-ranging animals

As all six of the fH/O2calibration lines from the captive geese were significantly different from the relationship derived for the wild geese, it would not be possible to apply calibrations obtained from captive birds to fH data recorded in wild geese. However, what the calibration lines from captive birds do demonstrate is that the relationship between fH and O2 determined during the flightless phase of wing moult is not significantly different from those derived during most of the year (the exception being late September–early October). This suggests that it may be possible to apply the calibration between fH and O2 established during moult in the wild geese to the rest of the annual cycle. Caution,however, should be used when applying this calibration to fH data recorded in the autumn period. In any case, during late September, wild barnacle geese undertake 4–5 flights as part of their southward migration. The fH/O2relationship of barnacle geese during flight is significantly different from that obtained from walking geese (Butler and Woakes, 1980; Nolet et al., 1992; Bishop and Butler,1995; Bishop et al.,2002). Therefore, the fH/O2calibrations from the captive and wild geese in this study would not be suitable for converting fH values during flight into O2 during late September.

LIST OF ABBREVIATIONS

• BM

body mass

•
• fH

heart rate

•
• GLM

general linear model

•
• HRDL

heart rate data logger

•
• LBM

lean body mass

•
• SS

structural size

•
• Tab

abdominal body temperature

•
• TBF

total body fat

•
• O2

rate of oxygen consumption

All experiments in Svalbard were done with the permission of the Governor of Svalbard, and the Norwegian Animal Welfare Committee. We are grateful to NERC and BBSRCfor funding fieldwork in Svalbard, and to Maarten Loonen and his team who caught wild birds for us. We would like to thank Craig White for his assistance with respirometry, statistical advice and for comments on an earlier draft. Thank you to Graham Martin, Theunis Piersma and Lewis Halsey for useful discussions, and to three anonymous reviewers who gave valuable comments. We are also grateful to Alan Gardner, Phil Archer, Ben Heanue and Pete Jones, for looking after the captive geese. S.J.P. was funded by a BBSRC studentship.

Bech, C. and Nomoto, S. (
1982
). Cardiovascular changes associated with treadmill running in the Pekin duck.
J. Exp. Biol.
97
,
345
-358.
Beswick, F. W. and Jordan, R. C. (
1961
). Cardiological observations at the sixth British Empire and Commonwealth Games.
Br. Heart J.
23
,
113
-130.
Bevan, R. M., Butler, P. J., Woakes, A. J. and Prince, P. A.(
1995
). The energy expenditure of free-ranging black-browed albatrosses.
Philos. Trans. R. Soc. Lond. B Biol. Sci.
350
,
119
-131.
Bishop, C. M. and Butler, P. J. (
1995
). Physiological modeling of oxygen consumption in birds during flight.
J. Exp. Biol.
198
,
2153
-2163.
Bishop, C. M., Ward, S., Woakes, A. J. and Butler, P. J.(
2002
). The energetics of barnacle geese (Branta leucopsis) flying in captive and wild conditions.
Comp. Biochem. Physiol. A
133
,
225
-237.
Bjornstad, H., Storstein, L., Meen, H. D. and Hals, O.(
1993
). Electrocardiographic findings according to level of fitness and sport activity.
Cardiology
83
,
268
-279.
Bock, A., Vancaulaert, C., Dill, D. B., Fo-lling, A. and Hurxthal, I. (
1928
). Studies in muscular activity III:dynamical changes occurring in man at work.
J. Physiol.
66
,
136
-161.
Bridge, E. S. (
2004
). The effects of intense wing moult on diving in alcids and potential influence on the evolution of molt patters.
J. Exp. Biol.
207
,
3003
-3014.
Brody, S. (
1945
).
Bioenergetics and Growth
. New York: Reinhold.
Butler, P. J. (
1991
). Exercise in birds.
J. Exp. Biol.
160
,
233
-262.
Butler, P. J. (
1993
). To what extent can heart rate be used as an indicator of metabolic rate in free-living marine mammals?
Symp. Zool. Soc. Lond.
66
,
317
-322.
Butler, P. J. and Woakes, A. J. (
1980
). Heart rate, respiratory frequency and wing beat frequency of free flying barnacle geese Branta leucopsis.
J. Exp. Biol.
85
,
213
-226.
Butler, P. J. and Turner, D. L. (
1988
). Effect of training on maximal oxygen-uptake and aerobic capacity of locomotory muscles in tufted ducks, Aythya fuligula.
J. Physiol.
401
,
347
-359.
Butler, P. J. and Woakes, A. J. (
2001
). Seasonal hypothermia in a large migrating bird: saving energy for fat deposition.
J. Exp. Biol.
204
,
1361
-1367.
Butler, P. J., Woakes, A. J. and Bishop, C. M.(
1998
). Behaviour and physiology of Svalbard Barnacle Geese Branta leucopsis during their autumn migration.
J. Avian Biol.
29
,
536
-545.
Butler, P. J., Frappell, P. B., Wang, T. and Wikelski, M.(
2002
). The relationship between heart rate and rate of oxygen consumption in Galapagos marine iguanas (Amblyrhynchus cristatus) at two different temperatures.
J. Exp. Biol.
205
,
1917
-1924.
Butler, P. J., Green, J. A., Boyd, I. L. and Speakman, J. R.(
2004
). Measuring metabolic rate in the field: the pros and cons of the doubly labeled water and heart rate methods.
Funct. Ecol.
18
,
168
-183.
Eichhorn, G. and Visser, G. H. (
2008
). Evaluation of the deuterium dilution method to estimate body composition in the barnacle goose: accuracy and minimum equilibration time.
Physiol. Biochem. Zool.
81
,
508
-518.
Flynn, R. K. and Gessaman, J. A. (
1979
). An evaluation of heart rate as a measure of daily metabolism in pigeons(Columbo livia).
Comp. Biochem. Physiol.
63A
,
511
-514.
Frappell, P. B., Blevin, H. A. and Baudinette, R. V.(
1989
). Understanding respirometry chambers: what goes in must come out.
J. Theor. Biol.
138
,
479
-494.
Frappell, P. B., Lanthier. C. and Baudinette, R. V.(
1992
). Metabolism and ventilation in acute hypoxia: a comparative analysis in small mammalian species.
Am. J. Physiol.
262
,
R1040
-R1046.
Froget, G., Butler, P. J., Handrich, Y. and Woakes, A. J.(
2001
). Heart rate as an indicator of oxygen consumption:influence of body condition in the king penguin.
J. Exp. Biol.
204
,
2133
-2144.
Froget, G., Handrich, Y., Le Maho, Y., Rouanet, J. L., Woakes,A. J. and Butler, P. J. (
2002
). The heart rate/oxygen consumption relationship during cold exposure of the king penguin: a comparison with that during exercise.
J. Exp. Biol.
205
,
2511
-2517.
Gessaman, J. A. (
1980
). An evaluation of heart rate as an indirect measure of daily energy metabolism of the American Kestrel.
Comp. Biochem. Physiol.
65
,
273
-289.
Green, J. A. and Frappell, P. B. (
2007
). Improving the precision and accuracy for estimating energy expenditure using the heart rate method.
Physiol. Biochem. Zool.
80
,
551
-555.
Green, J. A., Butler, P. J., Woakes, A. J., Boyd, I. L. and Holder, R. L. (
2001
). Heart rate and rate of oxygen consumption of exercising macaroni penguins.
J. Exp. Biol.
204
,
673
-684.
Green, J. A., Boyd, I. L., Woakes, A. J., Warren, N. L. and Butler, P. J. (
2005
). Behavioural flexibility during year-round foraging in macaroni penguins.
Mar. Ecol. Prog. Ser.
296
,
183
-196.
Henderson, Y., Harrard. H. W. and Dolley, F.(
1927
). The efficiency of the heart and the significance of rapid and slow pulse rates.
Am. J. Physiol.
82
,
512
-524.
Hohman, W. L., Ankney, C. D. and Gordon, D. H.(
1992
). Ecology and management of postbreeding waterfowl. In
Ecology and Management of Breeding Waterfowl
(ed. B. D. J. Batt, A. D. Afton, M. G. Anderson, C. D. Ankney, D. H. Johnson, J. A. Kadlec and G. L. Krapu), pp.
128
-189. Minneapolis,MN: University of Minnesota.
Holm, S. (
1979
). A simple sequentially rejective multiple test procedure.
Scand. J. Stat.
6
,
65
-60.
Holter, J. B., Urban, W. E., Hayes, H. H. and Silver, H.(
1976
). Predicting metabolic rate from telemetered heart rate in white-tailed deer.
J. Wildl. Manage.
40
,
626
-629.
Kerstel, E. R. T., Piersma, T. A., Gessaman, G. J., Dekinga, A.,Meijer, H. A. J. and Visser, G. H. (
2006
). Assessment of the amount of body water in the red knot (Calidris canutus): an evaluation of the principle of isotope dilution with H-2, O-17, and O-18 as measured with laser spectrometry and isotope ratio mass spectreometry.
Isotopes Environ. Health Stud.
42
,
1
-7.
Kleiber, M. (
1961
).
The Fire of Life: An Introduction to Animal Energetics
. New York:Wiley.
Larsson, K., van der Jeugd, H. P., van der veen, I. T. and Forslund, P. (
1998
). Body size declines despite positive directional selection on heritable size traits in a barnacle goose population.
Evolution
52
,
1169
-1184.
Malhotra, M. S., Sen-Gupta, J. and Rai, R. M.(
1963
). Pulse count as a measure of energy expenditure.
J. Appl. Physiol.
18
,
994
-996.
Margaria, R. (
1963
). Energy cost of running.
J. Appl. Physiol.
18
,
367
-368.
McKechnie, A. E., Freckleton, R. P. and Jetz, W.(
2006
). Phenotypic plasticity in the scaling of avian basal metabolic rate.
Proc. Biol. Sci.
273
,
931
-937.
McPhee, J. M., Rosen, D. A. S., Andrews, R. D. and Trites, A. W. (
2003
). Predicting metabolic rate from heart rate in juvenile Steller sea lions, Eumetopias jubatus.
J. Exp. Biol.
206
,
1941
-1951.
Nagy, K. A., Seymour, R. S., Lee, A. K. and Braithwaite, R.(
1978
). Energy and water budgets in free-living antechinus stuartii (Marsupialia dasyuridae).
J. Mammal.
59
,
60
-68.
Nagy, K. A., Siegfried, W. R. and Wilson, R. P.(
1984
). Energy utilization by free-ranging Jackass penguins, Spheniscus demersus.
Ecology
65
,
1648
-1655.
Nolet, B. A., Butler, P. J., Masman, D. and Woakes, A. J.(
1992
). Estimation of daily energy expenditure from heart rate and doubly labelled water in exercising geese.
Physiol. Zool.
65
,
1188
-1216.
Nudds, R. L. and Bryant, D. M. (
2001
). Exercise training lowers the resting metabolic rate of zebra finches, Taeniopygia guttata.
Funct. Ecol.
15
,
458
-464.
Owen, R. B. J. (
1969
). Heart rate, a measure of metabolism in blue winged teal.
Comp. Biochem. Physiol.
31
,
431
-436.
Pelsers, M. A., Butler, P. J., Bishop, C. M. and Glatz, J. F. C. (
1999
). Fatty acid binding protein in heart and skeletal muscles of the migratory barnacle goose throughout development.
Am. J. Physiol.
276
,
R637
-R643.
Portugal, S. J. (
2008
).
Ecophysiological aspects of the annual cycle of barnacle geese, Branta leucopsis
. PhD thesis, University of Birmingham.
Portugal, S. J., Green, J. A. and Butler, P. J.(
2007
). Annual changes in body mass and resting metabolism in captive barnacle geese (Branta leucopsis): the importance of wing moult.
J. Exp. Biol.
210
,
1391
-1397.
Portugal, S. J., Thorpe, S. K. S., Green, J. A., Myatt, J. P. and Butler, P. J. (
2009
). Testing the use/disuse hypothesis:pectoral and leg muscle changes in captive barnacle geese Branta leucopsis during wing moult.
J. Exp. Biol.
212
,
2403
-2410.
Shaffer, J. P. (
1986
). Modified sequentially rejective multiple test procedures.
J. Am. Stat. Assoc.
81
,
826
-831.
Speakman, J. R., Visser, G. H., Ward, S. and Krül, E.(
2001
). The isotope dilution method for the evaluation of body composition. In
Body Composition Analysis of Animals
(ed. J. R. Speakman), pp.
56
-98. Cambridge. UK:Cambridge University Press
Tolkamp, B. J., Emmans, G. C., Yearsley, J. and Kyriazakis,I. (
2002
). Optimization of short-term animal behaviour and the currency of time.
Anim. Behav.
64
,
945
-995.
Ward, S., Bishop, C. M., Woakes, A. J. and Butler, P. J.(
2002
). Heart rate and the rate of oxygen consumption of flying and walking barnacle geese (Branta leucopsis) and bar-headed geese(Anser indicus).
J. Exp. Biol.
205
,
3347
-3356.
Wilson, R. P., White, C. R., Quintana, F., Halsey, L. G.,Liebsch, N., Martin, G. R. and Butler, P. J. (
2006
). Moving towards acceleration for estimates of activity-specific metabolic rate in free-living animals: the case of the cormorant.
J. Anim. Ecol.
75
,
1081
-1090.
Zar, J. H. (
1984
).
Biostatistical Analysis, 2nd edn.
Englewood Cliffs, NJ: Prentice Hall.