Factors influencing the prediction of metabolic rate in a reptile


†Author to whom correspondence should be addressed. E-mail: timothy.clark@adelaide.edu.au


  • 1Measurements of the rate of oxygen consumption (o2) in the field are usually impractical, so several studies of endotherms have utilized heart rate (fH) as a correlate of o2 because of the tight relationship that often exists between the two variables. There have been several reports, however, where the relationship between fH and o2 changes or disassociates under different physiological or psychological circumstances. This may be further confounded in ectothermic vertebrates, which experience relatively large fluctuations in body temperature (Tb).
  • 2The aim of the present study was to characterize in Rosenberg's Goanna (Varanus rosenbergi) the relationship that exists between Tb, fH and o2 at rest and at different levels of exercise, during periods of heating and cooling, and following ingestion of a meal.
  • 3The combinations of Tb and fH were accurate at predicting o2 of animals at different levels of exercise and recovery, and during the postprandial period.
  • 4Predictions of o2 became less reliable during periods of relatively rapid heating when fH and blood flow increase for thermoregulatory purposes with no associated increase in o2. To counter this, fH was excluded from the prediction equation when the rate of heating exceeded 20% of the predicted mass-dependent maximum attainable rate, and o2 was predicted using Tb alone.
  • 5The resultant o2 prediction equation was used to estimate o2 of seven animals that were allowed to thermoregulate behaviourally, and the mean predicted o2 (o2pred) was not significantly different from the mean measured o2 (o2meas) for fasting or postprandial lizards.


Energy turnover and, in particular, how it is allocated to specific activities, is of central importance to the understanding of the physiological, behavioural and evolutionary ecology of organisms (McNamara & Houston 1996). Measurements of energy turnover through the determination of oxygen consumption rates (o2) are usually impractical in the natural environment and, as a consequence, two main approaches have been developed to estimate the energetics of animals in the field (see Butler et al. 2004 for review).

One of these approaches is to utilize the association that exists between o2 and heart rate (fH). Heart rate may be readily telemetered or logged in wild animals, and shares a relationship with o2 as described by the Fick equation for convection of oxygen through the cardiovascular system:

o2 = fH × VS × (CaO2 − CO2),(eqn 1)

where VS is cardiac stroke volume (the product of fH and VS is cardiac output) and (CaO2 − CO2) is the difference in oxygen content between arterial (CaO2) and mixed venous (CO2) blood. If oxygen pulse (VS × (CaO2 − CO2)) remains constant, or varies in a systematic manner with o2, then it should be possible to predict o2 based on measurements of fH.

For this to work, it is necessary to calibrate the relationship between fH and o2, preferably using animals undergoing activities of interest and in similar environmental conditions to those in the wild. For example, Barnacle Geese and Bar-Headed Geese display a markedly different fH/o2 relationship walking on a treadmill compared with flying in a wind tunnel (Ward et al. 2002). Furthermore, it has been shown in King Penguins that the relationship between fH and o2 resulting from treadmill exercise is different from that when thermoregulating at low environmental temperatures (Froget et al. 2002). The cardio-metabolic consequences of digestion confuse the relationship between fH and o2 in Steller Sea Lions, where postprandial increases in metabolism (i.e. specific dynamic action, SDA) are not associated with concomitant increases in fH (McPhee et al. 2003); implying that increases in VS and/or (CaO2 − CO2) are more important than an increase in fH for circulatory oxygen transport during the digestive period. Despite possible changes occurring in the relationship between fH and o2 with physiological or psychological state, several recent studies of endotherms have demonstrated that fH is an accurate correlate of the rate of energy expenditure (e.g. Bevan et al. 1995; Green et al. 2001; Ward et al. 2002).

Predictions of o2 based on fH may be confounded in ectothermic vertebrates because of large daily and seasonal fluctuations in body temperature (Tb). In this context, it has been reported for some species of fish that the effect of an increase in temperature is a right shift in the linear regression that describes the relationship between fH and o2 during exercise (e.g. Claireaux et al. 1995; see Clark et al. 2005a and references within).

Few studies exist that determine the combined effects of activity and temperature on fH and o2 in reptiles (Bennett 1972; Wilson 1974; Butler et al. 2002; Clark et al. 2005b). A study of marine iguanas exercising on a treadmill indicated that the fH/o2 regression line at a Tb of 36 °C was significantly right-shifted in comparison with that at a Tb of 27 °C and, consequently, the subsequent o2 prediction equation incorporated a Q10 function (Butler et al. 2002). This earlier study was limited to only two temperatures, and it is not known how accurately the equation predicts o2 at temperatures other than those studied. A further complication in predicting o2 from fH in reptiles could result from the thermally related hysteresis in fH that is known to exist in some species during heating and cooling (Bartholomew & Tucker 1963; Smith, Robertson & Davies 1978; Grigg & Seebacher 1999), where fH and blood flow are modified for thermoregulatory purposes without an accompanying change in o2 (see Clark, Butler & Frappell 2005c).

Consequently, using the Fick equation to predict field o2, particularly of ectotherms, requires a model that incorporates other variables in addition to fH. The present study characterizes in a reptile, Rosenberg's Goanna (Varanus rosenbergi), the relationship that exists between Tb, fH and o2 at rest and at different levels of exercise, during periods of heating and cooling, and following ingestion of a meal. Thus, it was our aim to test the hypothesis that a combination of these biological factors could be utilized accurately to predict o2 of animals under a controlled situation that simulated natural conditions. Rosenberg's Goanna was chosen as it lives at a relatively high latitude and is therefore exposed to a broad range of Tb values in the natural environment (10–38 °C; see Christian & Weavers 1994; Rismiller & McKelvey 2000).

Materials and methods


Lizards were obtained at the beginning of March from Kangaroo Island, South Australia, and kept in a temperature-controlled holding facility at La Trobe University at approximately 28 °C for a maximum of 25 days before use in experiments. They were maintained on a 12:12 h light : dark photoperiod with access to a heat lamp during light hours. Animals had unlimited access to water and were fed twice weekly, though, where appropriate, they were fasted prior to experimentation (see below). All animals were returned to the holding facility following experimentation.

rate of oxygen consumption

The rate of oxygen consumption was determined by placing a light-weight (approximately 4·5 g), transparent and loose-fitting mask over the head of the animal. The mask was fitted with an outlet tube through which air was drawn at a rate appropriate for the activity (1–8 l min−1), monitored by a calibrated mass flow meter (Sierra, model 810C Monterey, CA). A subsample of the air leaving the pump was passed through columns containing a drying agent (Drierite, Hammond, Xenia, Ohio, USA) and a carbon dioxide absorbent (Dragersorb, Lübeck, Germany) and analysed for the fractional content of oxygen by a gas analyser (model S-3A/I, Applied Electrochemistry, Pittsburgh, PA, USA). The rate of oxygen consumption was calculated from airflow through the mask and the difference between incurrent and excurrent fractional concentrations of dry, carbon dioxide-free air (see appendix in Frappell et al. 1992). All values of o2 are at standard temperature, pressure and dry (STPD) and expressed per kg.

heart rate and body temperature

Heart rate was obtained by attaching self-adhesive Ag/AgCl electrocardiogram (ECG) electrode pads and ECG leads to the dorsal surface of the animal, positioned such that they triangulated the heart. The ECG leads were connected to an amplifier (BIO amp, ADInstruments, Sydney, Australia). Body temperature was obtained by inserting a thermocouple 5–6 cm into the cloaca. All outputs were collected at 100 Hz (Powerlab 800, ADInstruments, Sydney, Australia) and displayed on a computer using Chart software (ADInstruments, Sydney, Australia).


The relationship between Tb, fH and V̇o2

Six adult animals (three males, three females; determined by morphological examination) of mean body mass (Mb) 1·55 ± 0·15 kg were fasted for at least 3 days then individually placed in a constant temperature room at the desired experimental temperature (range 14–36 °C) for at least 6 h prior to experimentation. Although Tb was determined immediately before and after experimentation, Tb values stated in the present report are an average of the two. Animals were then instrumented as described above and left to rest for at least 40 min on a variable-speed treadmill in the controlled temperature room until low, stable values of fH and o2 were observed. They were then run on the treadmill at the maximum speed they could maintain for 5–10 min. The treadmill was ramped to maximum speed typically within 1 min; maximum speed varied with individual goannas and Tb (range 0·3–2·9 km h−1). When a lizard would no longer run while being enticed with gentle tapping on the hind legs, the treadmill was stopped and recordings continued for approximately 60 min during the recovery phase. All animals were studied on seperate occasions at six or more Tb values between 14 and 36 °C. Values (∼1 min average of data points) were obtained from each animal 3–4 min prior to running, during the final 2 min of exercise, and at three stages during the recovery period while on the treadmill. Data from these experiments were used to establish an equation from which o2 could be predicted from fH and Tb.

Validating the prediction equation

Raw data collected from V. rosenbergi used in a previous study (see Clark et al. 2005c) were used to validate the equation established for predicting o2. In brief, seven animals were used both when fasted (>3 days fasted; fasted Mb 1·35 ± 0·09 kg), and then 25 h after ingestion of a meal (meal size 10 ± 1% of fasted Mb; causes a significant increase in fH and o2 approximately 25 h after ingestion, see Clark et al. 2005c). They were individually instrumented in a manner similar to that described above, and placed in a constant temperature room (floor area 1·5 × 2·0 m) at 14 °C until Tb (monitored continuously) fell below 19 °C. At this point, a heat lamp positioned in one corner of the room was switched on. The lizards were free to move (all leads and tubes were suspended from a central point above the animal) and, on the majority of occasions, lizards positioned themselves under the lamp within an hour, otherwise they were moved by the experimenter. After heating, lizards were free to explore the room for 1–5 h, after which time the heat lamp was switched off and the lizards were allowed to cool again until Tb fell below 19 °C. All data from resting and active animals were used to validate the equation established for predicting o2.

data analysis and statistics

Repeated measures multiple linear regression was used to determine the relationship between Tb, fH and o2. For the validation studies, data were averaged into 5-s blocks. Least-squares regressions were used where appropriate (see Results, Figs 4 and 5) to determine the relationships between variables. Repeated measures anova was used to compare between measured and predicted values for animals freely allowed to thermoregulate (Table 1). Significance was considered at P < 0·05. Data are presented as mean ± SEM unless otherwise indicated. N = number of animals, n = number of data points.

Figure 4.

Semi-log plot of the rate of oxygen consumption (o2) as a function of body temperature (Tb) during heating and cooling for fasting (open circles, N = 7) and postprandial (closed circles, N = 7) V. rosenbergi (modified from Clark et al. 2005c). There was no difference for either group between heating and cooling. Dashed lines describe the regression for fasting [log(o2) = 0·023Tb − 0·267; r2 = 0·94] and postprandial [log(o2) = 0·021Tb + 0·002; r2 = 0·93] animals. ancova revealed no difference in slope between the groups (P > 0·3). The solid line is the common regression (given as equation 4 in main text).

Figure 5.

Predicted rate of oxygen consumption (o2pred; calculated using equation 5) as a function of the measured rate of oxygen consumption (o2meas; N = 7 both for fasting and postprandial animals). Each point is the average value for 1 h. The solid line is the regression that describes the relationship both for fasting and postprandial animals (o2pred = 1·11o2meas − 0·16; r2 = 0·88, n = 79), with 95% confidence limits (dotted lines). ancova revealed that the regression was not different from the line of equality (dashed line; P > 0·2).

Table 1.  Measured and predicted values of the mean and total rate of oxygen consumption (o2) for V. rosenbergi when thermoregulating behaviourally
AnimalTotal duration (min)Measured o2Predicted o2Δ (%)
Mean (ml min−1 kg−1)Total (ml kg−1)Mean (ml min−1 kg−1)Total (ml kg−1)MeanTotal
  • Predicted values were calculated using equation 5. Δ is the difference between the predicted and measured o2 expressed as a percentage.

  • *

    No significant difference between mean values of measured o2 and predicted o2 within a group (P > 0·1, repeated measures anova).

  • Data for Animal 1 in the fasted state are illustrated in Fig. 3.

13703·8814213·931437   1·44   1·13
23913·2212493·241257   0·61   0·61
33572·61 7782·49 777−4·87−0·12
44432·6311252·781180   5·52   4·70
53752·9510943·141162   6·16   5·82
63952·07 8142·18 856   4·89   4·89
73802·03 7662·14 807   5·11   5·11
Mean3872·7710352·84*1068   2·69   3·16
Absolute Δ        4·09   3·20
11004·89 4444·84 440−1·00−1·00
23722·69 9942·751016   2·17   2·17
36241·46 9311·37 870−7·05−7·01
46351·7310761·771121   2·57   4·01
52343·05 7142·87 672−6·32−6·32
61532·19 3352·32 354   5·42   5·42
73831·68 6241·54 584−9·11−6·84
Mean3572·53 7312·49* 723−1·90−1·37
Absolute Δ        4·81   4·68


the relationship between tb, fh and o2

The relationship at different Tb values between fH and o2 changed in a predictable fashion during periods of treadmill exercise and recovery (Fig. 1). The rate of oxygen consumption, as a function of fH and Tb, was best described by the following equation:

Figure 1.

The relationship between body temperature (Tb), heart rate (fH) and the rate of oxygen consumption (o2) for V. rosenbergi at different levels of activity. Included are the relationships between fH and o2 that occur during exercise and recovery at 15 °C, 22 °C, 29 °C and 36 °C (dashed lines), calculated using equation 2. N = 6, n = 290.

log(o2) = 1·738 log(fH) − 0·028Tb − 1·472,(eqn 2)

where r2 = 0·80, P < 0·001, N = 6, n = 290.

the effect of feeding

The increase in o2 that occurs following consumption of a meal (∼10% of fasted Mb) has previously been characterized for V. rosenbergi (see Clark et al. 2005c) and, as it is primarily governed by fH and Tb (Tb increases owing to the heat increment of feeding), equation 2 reliably predicts changes in o2 during the postprandial period (Fig. 2).

Figure 2.

The measured change in the rate of oxygen consumption (o2) of V. rosenbergi at an ambient temperature of 30 °C following consumption of a meal (∼10% of fasted body mass; closed circles, N = 7, modified from Clark et al. 2005c). Displayed in open squares is the trend in postprandial o2 predicted using equation 2. There was no significant difference between measured and predicted o2 at any given time (P > 0·05).

the effect of heating and cooling

When given the opportunity to thermoregulate behaviourally, the majority of the lizards basked under the heat lamp to obtain a Tb of 32–35 °C (e.g. Fig. 3a). Once the desired Tb was reached, the lizards either moved away and cooled or shuttled in and out from underneath the heat lamp to maintain their preferred Tb until the lamp was switched off.

Figure 3.

Traces for a fasted individual animal of (a) body temperature (Tb) and heart rate (fH), and (b) and (c) the measured rate of oxygen consumption (o2meas) while thermoregulating behaviourally in a 14 °C room with access to a heat lamp (available for the duration indicated by the black bar at top of figure). Data have been binned into 5-min blocks. Also illustrated are the predicted values of o2 (o2pred; dashed lines) calculated using (b) equation 2 and (c) equation 5.

During these periods, equation 2 tended to overestimate o2 for all individuals, particularly during periods of relatively rapid radiant heating when the increase in fH was not matched by an increase in o2 (Fig. 3b). The maximum rate at which a lizard can heat (ΔTbt; °C min−1) is negatively related to body mass for fasting and postprandial animals:

log(ΔTbt) = −0·823 log(Mb) + 0·012,(eqn 3)

where r2 = 0·28, P < 0·05, N = 7, n = 14. It was iteratively determined that equation 2 became inaccurate when the rate of heating exceeded 20% of the maximum mass-dependent attainable rate (this typically occurred during 70% of the initial heating period, and for approximately 15% of the entire experiment). As such, when an animal heated faster than 20% of the predicted maximum attainable rate for its body mass (calculated from equation 3), it was assumed that fH and o2 had uncoupled for the purpose of thermoregulation. During these periods, fH was excluded from the prediction equation and o2 was recalculated using only Tb for resting animals when undergoing rapid heat exchange (see equation 4; Fig. 4). This relationship between Tb and o2 is the same during heating and cooling within fasting and within postprandial lizards, but postprandial lizards have a higher o2 than fasting lizards as a result of SDA (Fig. 4). A common regression was adopted for the purpose of the o2 prediction equation:

log(o2) = 0·022Tb − 0·132,(eqn 4)

where r2 = 0·43, P < 0·001, N = 7, n = 68.

predicting o2

To summate, equation 2 reliably predicts o2 except during periods of relatively rapid heating when an uncoupling of fH and o2 results in an overestimate of o2 (see Fig. 3b). This overestimation is corrected by incorporating equation 4 when Tb increases faster than 20% of the predicted maximum attainable rate (see Fig. 3c). The resultant equation can therefore be formalized as:

log(o2) = C1 × log(fH) − C2 × Tb − C3,(eqn 5)

where C1 = 1·738, C2 = 0·028 and C3 = 1·472 when the mass-dependent rate of heating is less than 20% of the predicted maximum attainable value for the given body mass, and C1 = 0·000, C2 = −0·022 and C3 = 0·132 when the rate of heating exceeds 20% of the maximum attainable value.

Subsequently, equation 5 was used to predict o2 of the seven individuals that were used in the validation study. Mean measured o2 (o2meas) of these animals was not significantly different from mean predicted o2 (o2pred) either for fasting or postprandial animals (Table 1). The data were binned into 1-h blocks to be analysed at a greater resolution and, in this case, equation 5 could still reliably be used to predict o2 (Fig. 5).



As far as the circulatory system of varanids is concerned, the increase in o2 during maximum exercise is typically achieved by increases in all of fH, VS and (CaO2 − CO2) (Gleeson, Mitchell & Bennett 1980; Frappell, Schultz & Christian 2002a,b; Clark et al. 2005b), where the increase in fH usually accounts for the majority of the increase in cardiac output (Bennett 1972; Gleeson et al. 1980; Frappell et al. 2002b). In the present study of V. rosenbergi, a tight relationship between fH and o2 was maintained at all levels of treadmill exercise over the entire range of temperatures studied, and it is likely that the observed range in these variables encompasses the majority of what would be experienced in the field.

For a given activity over a range of intensities, the relationship between fH and o2 at a given Tb is linear or curvilinear for most animals studied, including birds (e.g. Grubb, Jorgensen & Conner 1983; Bevan et al. 1994; Green et al. 2001; Ward et al. 2002), mammals (e.g. Fedak, Pullen & Kanwisher 1988; Williams, Kooyman & Croll 1991; Williams, Friedl & Haun 1993), reptiles (e.g. Butler et al. 2002; present study) and some fish (e.g. Armstrong 1986; Lucas 1994; Webber, Boutilier & Kerr 1998; Clark et al. 2005a). With the possible exception of some species of fish (see Clark et al. 2005a and references within), it appears for most vertebrates that VS and (CaO2 − CO2) change systematically during exercise and proportionally across temperature such that a strong relationship between fH and o2 is maintained (Wilson 1974; Gleeson et al. 1980; Grubb et al. 1983; Jones et al. 1989; Butler et al. 1992; Clark et al. 2005a,b).


The act of processing, digesting and absorbing food is typically associated with an increase in the rate of oxygen consumption (i.e. SDA) which, in most vertebrates, is largely accompanied by an increase in fH (Kelbaek et al. 1987; Dumsday 1990; Wang, Burggren & Nobrega 1995; Hicks, Wang & Bennett 2000; Secor, Hicks & Bennett 2000; Wang et al. 2000, 2001; Clark et al. 2005c). However, in accordance with that occurring during exercise, the overall increase in systemic blood flow during digestion may not match the factorial increase in o2 (e.g. Hicks et al. 2000; Secor et al. 2000), thus an associated increase in (CaO2 − CO2) may also occur. It appears for V. rosenbergi that the pattern of change in circulatory variables is such that the relationship between fH and o2 determined for exercising animals (equation 2) is maintained during digestion (Fig. 2).

The combined effects of digestion and exercise have been shown in some animals to elicit a response in o2 that is higher than either process could elicit on its own, yet the response is typically less than the sum of both processes combined (i.e. a partial ‘additivity’ of the responses; Segal & Gutin 1983; Secor et al. 2000; Bennett & Hicks 2001). From the data currently available, it appears that fH may be at least partly responsible for the cumulative effect in o2 that occurs during simultaneous digestion and exercise (e.g. Secor et al. 2000). Given that equation 5 did not significantly underestimate o2 of active postprandial V. rosenbergi in the present study (Table 1 and Fig. 5), this ‘additivity’ of the responses should not be problematic when predicting energy expenditure of free-ranging, ectothermic vertebrates.


Of all vertebrates, reptiles may undergo the largest and most rapid daily fluctuations in Tb, thus contributing to the difficulty of predicting field metabolic rate of such animals. To account for the disassociation between fH and o2 that occurs in V. rosenbergi during rapid heat exchange (see Introduction), the prediction of o2 in the present study required that fH be excluded from the prediction equation during periods of relatively rapid heating (>20% of maximum attainable rate). Clearly, operative temperatures in the natural environment will vary from those to which the animals were exposed in the present study, thus heat exchange will occur at different rates. It may be necessary in the wild therefore to monitor thermal conditions at ground level (e.g. radiant and ground temperatures), in combination with maximum attainable rates of heat exchange (determined from Tb trace on data loggers or transmitters; see below), to calculate at which point fH is to be excluded from the o2 prediction equation. Correcting for this should not be difficult given that variations in operative temperature should influence only the intercept, but not the slope, of the relationship between Mb and the rate of heating (see equation 3).

As it is unlikely that the digestive state of an animal in the natural environment would be known, the incorporation of equation 4 into the prediction equation during relatively rapid heating assumes that all animals have a small postprandial increment in o2 during this period. Because such periods of rapid heating typically occur rather infrequently during the day for reptiles in the natural environment (see Christian & Weavers 1994; Rismiller & McKelvey 2000; Seebacher & Grigg 2001), it is unlikely that this assumption would produce a substantial error in a prediction of daily energy expenditure irrespective of the digestive state of the animal. In this context, if equation 4 is replaced either with the linear regression equation describing the relationship between o2 and Tb during the fasting period, or with that describing the relationship during the postprandial period (see Fig. 4), the resultant mean o2pred values are not significantly different from the mean o2meas values (P > 0·1 in both cases).

predicting field metabolic rate

With the advent and continued miniaturization of electronic data loggers/transmitters of fH and Tb (e.g. Woakes, Butler & Bevan 1995), it is becoming increasingly possible to predict and monitor the energetics of animals in the natural environment (see Butler et al. 2004 and references within). The principal advantage of using this approach is that it can provide estimates of o2 at a far greater temporal resolution than many other methods (e.g. doubly labelled water, DLW); the resolution being limited only by the calibration procedure and the method of recording fH. The present study provides the most comprehensive evidence to date that it should be possible, based primarily on fH and Tb, to predict the energy expenditure of free-ranging ectotherms that undergo daily fluctuations in Tb.

The suggested approach (i.e. equation 5) is accurate at predicting the mean o2 of a group of animals, though the mean values hide the individual algebraic errors which ranged from −4·87% to +6·16% in fasting animals, and from −9·11% to +5·42% in postprandial animals (Table 1). Similar interindividual error ranges are typical when using a group prediction equation to estimate o2 of individual animals (e.g. Nolet et al. 1992; Bevan et al. 1994, 1995; Boyd et al. 1995; Hawkins et al. 2000; Froget et al. 2001; McPhee et al. 2003). It is generally accepted therefore that, although this method is capable of predicting o2 for specific activities in the field, it should be used only to estimate mean o2 of a group of animals rather than estimating it for an individual animal (see Butler et al. 2004), unless specific animals are individually calibrated such that the prediction of energy expenditure of each animal is based on its own calibration equation. In either case, the error associated with the prediction of o2 must be taken into account (see Green et al. 2001; Butler et al. 2004).

Having established that there is a significant relationship between o2 and the variables of the prediction equation (e.g. fH and Tb), it is important to remember that the relationship may not remain constant for individuals of a different physiological state. For example, it has been noted for King Penguins that the fH/o2 relationship changes during the fasting period when animals spend long periods ashore (Froget et al. 2001; Fahlman et al. 2004). There are data for lizards that suggest seasonal changes in o2 (Christian & Conley 1994; Christian, Bedford & Schultz 1999; de Souza et al. 2004), though it has not yet been ascertained whether fH follows a similar seasonal pattern. Nevertheless, if our approach is to be used to predict field o2 of V. rosenbergi throughout the entire year, it will be necessary to clarify the relationship between Tb, fH and o2 on wild-caught animals at different times of the year, and potentially incorporate a seasonal component into the resultant equation.

The present study details the principles involved when predicting metabolic rate of a thermoregulating ectotherm, and the findings suggest that it is possible to use primarily fH and Tb to predict the energetics of free-ranging ectothermic vertebrates in the natural environment. The interest in using fH to estimate field o2 of birds and mammals has increased in recent years, and the principles outlined in the present study should encourage similar studies on energetics of free-ranging ectothermic vertebrates.


Animal collection and experiments were performed with the approval of the Animal Ethics Committee of La Trobe University (code: 99/42 L). Brian Green is thanked for the collection of animals. Eva Suric and Tobie Cousipetcos are commended for animal husbandry. Craig White is thanked for statistical advice. TDC was the recipient of an Australian Postgraduate Award scholarship. PJB is a La Trobe University Visiting Distinguished Professor to the Adaptational and Evolutionary Respiratory Physiology Laboratory.