The decrease in body temperature (Tb) observed in most vertebrate classes in response to hypoxia has been attributed to a regulated decrease in set-point, protecting organs against tissue death due to oxygen depletion. Hypoxia, however, imparts particular challenges to metabolic function which may, in turn, affect thermoregulation. In ectotherms, where thermoregulation is mainly behavioural, stressors that influence the propensity to move and respond to temperature gradients are expected to have an impact on thermoregulatory control. Using low oxygen as a potent stressor,we evaluated the variability and level of thermoregulation of inland bearded dragons. To examine the source of thermoregulatory variability, we studied their behaviour in an electronically controlled temperature-choice shuttle box, a constant temperature dual-choice shuttle box, and a linear thermal gradient. A significant increase in the size of the Tbrange was observed at the lowest oxygen concentration (4% O2),reflecting a decrease in thermoregulatory precision in the temperature-choice shuttle box. This was also accompanied by a drop of ∼2–4°C in Tb, the drop being greatest in situations where Tb must be actively defended. Situations that force the lizards to continually choose temperatures, rather than passively remain at a given temperature, lead to an increase in the variability in the manifested Tb, which is further exaggerated in hypoxia. This study reveals that a decrease in thermoregulatory precision caused by a diminished propensity to move or effect appropriate thermoregulatory responses may be a contributing component in the lowering of selected body temperatures observed in many hypoxic ectotherms.

Body temperature (Tb) is one of the most important variables in an ectotherm's life history, since it determines the rate at which most physiological functions take place. Optimal performance of these functions occurs within a narrow range, and the capability of an animal to regulate Tb within this range ultimately determines its ability to escape predators, grow and reproduce. Low oxygen, however, causes reptiles to select lower temperatures in the laboratory(Hicks and Wood, 1985) as well as in the field (Rollinson et al.,2008). Indeed, it is common for most animals studied to reduce Tb in hypoxia (Wood,1995; Wood and Gonzales,1996), an inherent regulatory response that protects tissues against oxygen depletion, particularly in life-sustaining organs such as the heart and the brain. This fall in Tb can reduce oxygen demands by up to 50% (Hicks and Wood,1985) through a combination of lowered metabolic rate via Q10 effects [as well as metabolic depression(Bickler and Buck, 2007; Hicks and Wang, 2004)], an accompanying reduction in ventilatory costs, and an increase in the oxygen-loading capacity of the lungs (Wood and Gonzales, 1996).

It is not uncommon for reptiles in nature to encounter situations which can cause hypoxaemia. Infections from blood-borne parasites are known to reduce the oxygen-carrying capacity of the blood(Saint-Girons, 1970; Schall et al., 1982); anaemia and exhaustive exercise can also cause temporary hypoxaemia. These conditions are known to induce decreases in Tb similar to those induced by external hypoxia (Hicks and Wood, 1985; Petersen et al.,2003). A fall in Tb also commonly occurs in response to numerous stressors in addition to hypoxia [e.g. dehydration,hypoglycaemia, toxins (Kozak,1997)], suggesting a common regulatory process that responds to alterations in the neurochemistry of the brain(Bicego et al., 2007). Thus,the relevance of examining hypoxia-induced alterations in Tb also lies in its potential for shedding light on the thermoregulatory defence mechanisms common to a number of stressors.

The natural and laboratory Tb values of reptiles have a skewed distribution (Dewitt and Friedman,1979), dropping off sharply at higher temperatures with a wider distribution at lower temperatures. One explanation for this may be the decrease in locomotor performance observed in reptiles when exposed to low temperatures, prolonging the time animals spend at low Tbvalues. Higher temperatures, on the other hand, are thought to be typically avoided due to lethal effects and potential cardio-respiratory limitations(Wood, 1984). Any factor that decreases the responsiveness to cold could increase the overall variability in Tb, because animals will not respond as readily to lower temperatures. The net result of encountering a wider distribution of low temperatures is an overall fall in selected Tb. Although hypoxia has been shown to lower the preferred Tb along with other thermoeffectors [e.g. panting, skin reflectivity(de Velasco and Tattersall,2008; Hicks and Wood,1985; Petersen et al.,2003; Tattersall and Gerlach,2005)], the precision of thermoregulation has been largely overlooked. If the decrease in Tb observed in conditions like hypoxia is associated with a decrease in thermoregulatory precision, it is possible that changes to the underlying thermosensation may be significant contributors to the change in Tb(McKemy, 2007; Sayeed and Benzer, 1996).

On the other hand, severe hypoxia can limit the capacity for aerobic metabolism (Hicks and Wang,2004; Wood and Glass,1991) and therefore the potential for aerobic activity. Given that ectotherms commonly reduce metabolic expenditure in response to hypoxia(Bickler and Buck, 2007), it is expected that lizards exposed to low oxygen concentrations will be less apt to exhibit movement than those in normoxia. If extensive movement is required for thermoregulation, it is plausible that hypoxia will induce a decrease in thermoregulatory precision. To test this prediction we took advantage of the different metabolic efforts associated with three different techniques(Withers and Campbell, 1985)commonly used to assess thermal preference.

This study aimed to answer whether the hypoxia-induced lowering of Tb consists solely of a regulated decrease in Tb (i.e. decrease in set-point), as has been extensively suggested, or whether the decrease in Tb is accompanied by a decrease in thermoregulatory precision (i.e. an increase in load error). In addition, this study examined the importance of effort' (i.e. amount of locomotion required to maintain a constant Tb) on the variability and magnitude of the hypoxic thermoregulatory response. The recognition of a decrease in thermoregulatory precision during hypoxia would shed light on the potential for a reduction in thermosensitivity and/or a diminished propensity to move as contributing factors to the hypoxic decline in Tb.

### Experimental animals

Twelve bearded dragons (eight male and four female; mass 323.9±63.2 g, age 2–3 years), Pogona vitticeps (Ahl 1926), raised from eggs at Brock University were used in the study. A detailed description of the animals' housing and care conditions is given elsewhere(Cadena and Tattersall, 2009). All procedures involving the use of these animals were approved by the Brock University Animal Care and Use Committee (protocol no. 041001).

### Experimental design

Lizards were fasted for a period of 12 h prior to the experiments. Experiments were run from 08:00 to 20:00 h, with the first 4 h consisting of a habituation period (in normoxia). Previous experiments indicated a considerable decline of putative non-thermoregulatory activity (exploratory shuttling') after an initial 4 h period inside the shuttle box(Cadena and Tattersall, 2009). At the beginning of the day the animals were cold (Tb at 08:00 h, 28.5±0.4°C) and had just emerged from their nocturnal shelter. To avoid further cooling-induced lethargy, individual lizards were placed on the warm side of the experimental apparatus at the start of each experiment. Upon completion of the experiment, lizards were returned to their housing facilities.

#### Series I: influence of hypoxia on thermoregulatory behaviour in a ramping shuttle box

Each individual lizard was exposed to five different oxygen concentrations(4, 5, 7, 10 and 21% O2) in a random order, one oxygen level per day. After the initial habituation period, the oxygen concentration was manipulated using an oxygen controller (Pro-Ox, model 110, BioSpherix,Redfield, NY, USA). This was done by delivering nitrogen into the shuttle box and flushing out the air, until the desired level of oxygen (±0.2%O2) was reached, usually within 30 min. This half hour following the initiation of hypoxia was not included in the data analysis, leaving a total of 7.5 h of analysable data. Normoxic levels (21% O2) were produced by leaving the shuttle box open to room atmosphere. Ta and Tb were recorded at 30 s intervals throughout the experiments. The time and temperature at the moment the lizard exited either compartment of the box were also recorded. The number of times each lizard shuttled was also calculated over the 7.5 h period as an indication of thermoregulatory effort and activity.

#### Series II: influence of methodology on the assessment of thermoregulatory behaviour in hypoxia

We used nine lizards in these experiments and exposed them to 21% and 4%O2 [normoxic data are derived from our previous study(Cadena and Tattersall, 2009)]. As in series I, individual lizards were placed in the experimental apparatus for an initial 4 h habituation period and 30 min were required for oxygen to reach the desired level of 4% O2. Lizards were then allowed to thermoregulate inside the experimental apparatus for an additional 7.5 h.

### Data recording and analysis

Assessment of the thermoregulatory variables, data acquisition, recording and processing followed the methodology described elsewhere(Cadena and Tattersall, 2009). Significant differences between groups detected by a repeated measures analysis of variance (RM ANOVA) (see below for details) were further explored using the Holm–Sidak procedure as a post-hoc method. Differences were considered significant at P≤0.05. Whenever data were not normally distributed, log transformations were applied to comply with the homoscedasticity requirements of the statistical tests. An analysis of the residuals on the log-transformed data indicated these were normally distributed. On occasions where normality could not be achieved, an RM ANOVA on ranks was used instead. All statistical analyses were performed using SigmaStat statistical software (version 3.0.1, Systat Software Inc., San Jose,CA, USA).

#### Series I: influence of hypoxia on thermoregulatory behaviour in a ramping shuttle box

Medians for upper escape ambient temperature (ambient temperature at which a lizard exited the hot side of the shuttle box; UETa), lower escape ambient temperature (ambient temperature at which a lizard exited the cold side of the shuttle box; LETa), Tb and Ta and means for the Tb range (central 68% range of the Tb distribution; RTb), as well as coefficients of variation (c.v.) of UETa, LETa, upper escape Tb (body temperature at which a lizard exited the hot side of the shuttle box; UETb) and lower escape Tb (body temperature at which a lizard exited the cold side of the shuttle box; LETb) across the 7.5 h of experimental conditions were compared between treatments using RM ANOVA. The c.v. was used as an indicator of intra-individual variability since it standardises differences in variability that may result from the level of the values. RTb was described using the central 68% of the data range, as it is analogous to the standard deviation around the median (see Dewitt, 1967). Due to the intrinsic dependence between RTb and the corresponding high and low limits of this range (HTbL and LTbL), statistical analysis was only performed on RTb.

A one-way RM ANOVA was also used to test differences in the number of shuttles between oxygen levels. To account for possible confounds between the multiple test procedures performed on the Ta and Tb variables (i.e. Tb, Ta, UETa, LETa and RTband also between c.v. of escape temperatures) Bonferroni–Holm procedures(Holm, 1979) were performed on the P-values.

#### Series II: influence of methodology on the assessment of thermoregulatory behaviour in hypoxia

The median values for Tb from each individual(N=9) were calculated. Tb and RTb were compared between the three different methodologies (i.e. ramping shuttle box,extreme temperatures shuttle box and thermal gradient) and oxygen levels using a two-way RM ANOVA, with methodology and oxygen level as factors.

The number of shuttles that occurred in the extreme trials was compared between 21% and 4% O2 using Student's paired t-test. Additionally, in the extreme temperatures shuttle box, animals were noted to spend time straddling the cold and warm compartments. The proportion of time spent in the cold, hot and middle' (straddling the two compartments) was calculated for both oxygen concentrations, but to avoid type I error only the proportion of time spent in the middle was compared between 21% and 4%O2 using an RM ANOVA. For the thermal gradient experiments,movement' was approximated by summing up the number of times that animals moved greater than 20 cm in a 1 min period. This value was subsequently compared between the 21% and 4% O2 treatments. P values for movement estimates in the extreme temperatures shuttle box and the thermal gradient and for time spent in the middle between compartments of the shuttle box were Bonferroni–Holm corrected.

### Series I: influence of hypoxia on thermoregulatory behaviour in a ramping shuttle box

By continuously shuttling back and forth between the heating and the cooling compartment of the box, lizards were able to maintain a relatively constant Tb at all oxygen levels(Fig. 1). Five of the twelve lizards (three males, two females) required at least three repetitions to exhibit active thermoregulation throughout the entire 7.5 h period. This was unrelated to oxygen treatment. Data obtained from these five lizards fell within the range of data obtained from the other seven lizards and therefore data from all lizards were pooled for the analysis. Despite several repetitions (>6) four lizards did not thermoregulate at one of the experimental conditions presented to them (one at 5%, one at 7% and two at 21%oxygen). These experiments comprised only 4% of the total of 96 experiments conducted, and were excluded from the analysis.

UETa exhibited no change with oxygen levels, although there was a trend to increase at lower oxygen concentrations(F4,40=2.42, P=0.064). LETa decreased significantly with exposure to 4% and 5% oxygen(F4,40=10.65, P<0.001; Fig. 2; Table 1) with respect to normoxia.

Table 1.

Thermoregulatory variables of bearded dragons during exposure to different oxygen concentrations in the ramping shuttle box trials

O2 (%)UETa (°C)LETa (°C)Ta (°C)HTbL (°C)LTbL (°C)RTb (°C)Tb (°C)
21 43.0±2.4 25.4±2.6 34.4±0.6 35.9±0.5 33.4±1.0 2.3±1.0 34.7±0.7
10 43.0±2.9 24.8±1.8 34.1±0.8 35.9±0.7 33.3±0.9 2.6±0.6 34.7±0.6
44.1±2.0 24.4±2.1 33.6±0.9 35.4±0.8 32.6±1.0 2.8±1.1 34.2±0.7
43.6±1.5 23.0±2.0* 32.9±1.2* 34.8±1.5 31.6±1.6 3.2±1.4 33.5±1.2*
44.7±1.3 20.6±2.7* 32.1±2.0* 34.7±1.5 30.1±2.4 4.6±1.8* 32.5±1.7*
O2 (%)UETa (°C)LETa (°C)Ta (°C)HTbL (°C)LTbL (°C)RTb (°C)Tb (°C)
21 43.0±2.4 25.4±2.6 34.4±0.6 35.9±0.5 33.4±1.0 2.3±1.0 34.7±0.7
10 43.0±2.9 24.8±1.8 34.1±0.8 35.9±0.7 33.3±0.9 2.6±0.6 34.7±0.6
44.1±2.0 24.4±2.1 33.6±0.9 35.4±0.8 32.6±1.0 2.8±1.1 34.2±0.7
43.6±1.5 23.0±2.0* 32.9±1.2* 34.8±1.5 31.6±1.6 3.2±1.4 33.5±1.2*
44.7±1.3 20.6±2.7* 32.1±2.0* 34.7±1.5 30.1±2.4 4.6±1.8* 32.5±1.7*

Values are means of medians (±s.d.) of 12 lizards during a 7.5 h period. *Significant differences relative to 21% oxygen, with the Holm–Sidak post-hoc test (P<0.05). UETa,upper escape ambient temperature; LETa, lower escape ambient temperature; Ta, ambient temperature; HTbL and LTbL, high and low body temperature limits, respectively (limits of the central 68% range of the body temperature distribution); RTb,ambient temperature range (HTbL–LTbL); Tb, body temperature. Statistical analysis was not performed on the HTbL and LTbL to minimize type I error due to the intrinsic dependence between these variables and RTb

There were no significant differences in the c.v. of UETa and UETb (F4,40=1.45, P=0.235 and F4,40=2.22, P=0.167, respectively) between oxygen treatments. The c.v. of LETa and LETb increased significantly at the lowest oxygen concentrations. When compared with normoxia, the c.v. of LETa was significantly higher at 4%, 5% and 7% oxygen (F4,40=8.71, P<0.001) and the c.v. of LETb rose to reach significance at 4% oxygen(F4,40=5.65, P=0.003; Fig. 3).

Fig. 1.

Representative trace of body temperature (Tb) and selected ambient temperature (Ta) of two different lizards(P. vitticeps) allowed to thermoregulate inside an electronic temperature-choice shuttle box. Traces are plotted for the 4.5 h acclimation period and the subsequent 7.5 h of exposure to (A) 21% O2 and (B)5% O2. Arrow denotes the onset of hypoxia at 12:00 h. Arrowhead represents the moment at which 5% oxygen was reached (12:30 h).

Fig. 1.

Representative trace of body temperature (Tb) and selected ambient temperature (Ta) of two different lizards(P. vitticeps) allowed to thermoregulate inside an electronic temperature-choice shuttle box. Traces are plotted for the 4.5 h acclimation period and the subsequent 7.5 h of exposure to (A) 21% O2 and (B)5% O2. Arrow denotes the onset of hypoxia at 12:00 h. Arrowhead represents the moment at which 5% oxygen was reached (12:30 h).

The pronounced decrease in LETa was reflected in the resulting preferred RTb; both the high (HTbL) and low(LTbL) limits of RTb decreased with lower oxygen concentrations (Fig. 2; Table 1). Because LTbL exhibited a more pronounced decrease than HTbL,RTb increased significantly at 4% oxygen(F4,40=5.92, P<0.001; Table 1).

Exposure to 5% and 4% oxygen led to significantly lower Ta values (F4,40=9.91, P<0.001) than exposure to 21% oxygen(Table 1). Preferred Tb followed a similar pattern, presenting slightly higher values than Ta at all levels of oxygen(Fig. 2; Table 1). Tb was 34.7±0.7°C under normoxic conditions. Tb decreased significantly at 4% and 5% O2(F4,40=15.99, P<0.001) compared with normoxia(Fig. 1; Table 1).

Hypoxia elicited a progressive decrease in the number of shuttles between compartments (Fig. 4). This decrease was significant at 5% and 4% O2(F4,40=5.69, P=0.001) where lizards shuttled 27.4±7.4 and 20.7±6.7 times, respectively, during a 7.5 h period compared with 59.5±61.2 times in normoxia.

Fig. 2.

Values for the different parameters of precision of thermoregulation with exposures to different levels of oxygen. Upper and lower escape ambient temperatures (UETa and LETa, respectively), high and low limits of the Tb range (HTbL and LTbL) and preferred Tb are plotted as the mean of the 7.5 h median values of 12 lizards (P. vitticeps)±s.e.m. (for visual clarity) instead of s.d. Values for s.d. are reported in Table 1. Animals were tested for 7.5 h in an electronic shuttle box at each of the experimental conditions. *Significant effect compared with normoxic values with the Holm–Sidak post-hoc test (P<0.05).

Fig. 2.

Values for the different parameters of precision of thermoregulation with exposures to different levels of oxygen. Upper and lower escape ambient temperatures (UETa and LETa, respectively), high and low limits of the Tb range (HTbL and LTbL) and preferred Tb are plotted as the mean of the 7.5 h median values of 12 lizards (P. vitticeps)±s.e.m. (for visual clarity) instead of s.d. Values for s.d. are reported in Table 1. Animals were tested for 7.5 h in an electronic shuttle box at each of the experimental conditions. *Significant effect compared with normoxic values with the Holm–Sidak post-hoc test (P<0.05).

### Series II: influence of methodology on the assessment of thermoregulatory behaviour in hypoxia

Methodology had a significant effect on Tb; lizards exhibited significantly lower Tb values at both 21% and 4%O2 (two-way RM ANOVA, F2,16=24.20, P<0.001) when evaluated in the extreme temperatures shuttle box compared with both the thermal gradient and the ramping shuttle box. Regardless of the methodology used, however, hypoxia (4% O2)induced a significant decrease in Tb (two-way RM ANOVA, F1,8=57.75, P<0.001; Table 2; Fig. 5). LTbL and HTbL were affected by hypoxia (4% O2) in all methodologies used (Table 2);however, a significant effect of hypoxia on RTb was only seen in the ramping shuttle box, but not in either of the other two treatments(two-way RM ANOVA, F1,8=16.09, P=0.003).

Table 2.

Comparison of thermoregulatory variables of bearded dragons during exposure to different oxygen concentrations across the different methodological trials

TreatmentO2 (%)Tb (°C)ΔTb 21%–4%LTbL (°C)HTbL (°C)RTb (°C)
Ramping shuttle box 21 34.7±0.7 2.3±1.5 33.5±1.0 35.9±0.5 2.3±1.0
32.5±1.7*  30.1±2.4 34.7±1.5 4.6±1.8*
Extreme temperatures shuttle box 21 33.2±1.4 4.3±2.0 30.7±2.0 35.0±1.3 4.3±2.1
29.3±2.3*  27.6±2.6 32.2±2.3 4.6±2.4
Thermal gradient 21 34.9±1.6 2.6±2.3 33.4±1.5 36.2±1.2 2.8±1.3
32.4±1.7*  30.6±1.7 34.1±1.5 3.4±1.5
TreatmentO2 (%)Tb (°C)ΔTb 21%–4%LTbL (°C)HTbL (°C)RTb (°C)
Ramping shuttle box 21 34.7±0.7 2.3±1.5 33.5±1.0 35.9±0.5 2.3±1.0
32.5±1.7*  30.1±2.4 34.7±1.5 4.6±1.8*
Extreme temperatures shuttle box 21 33.2±1.4 4.3±2.0 30.7±2.0 35.0±1.3 4.3±2.1
29.3±2.3*  27.6±2.6 32.2±2.3 4.6±2.4
Thermal gradient 21 34.9±1.6 2.6±2.3 33.4±1.5 36.2±1.2 2.8±1.3
32.4±1.7*  30.6±1.7 34.1±1.5 3.4±1.5

LTbL and HTbL were not statistically analysed in order to avoid type I error due to the dependence between these variables and RTb. Significant influence of methodology on the decline in Tb during 4% O2 exposure. *Significant effect of 4% O2 exposure on the parameter of interest

Fig. 3.

Coefficients of variation of upper and lower (A) ambient (UETaand LETa, respectively) and (B) body (UETb and LETb, respectively) escape temperatures of bearded dragons during exposure to different oxygen concentrations. Values are means of 12 lizards during a 7.5 h period. *Significant differences relative to 21%oxygen, with the Holm–Sidak post-hoc test(P<0.05).

Fig. 3.

Coefficients of variation of upper and lower (A) ambient (UETaand LETa, respectively) and (B) body (UETb and LETb, respectively) escape temperatures of bearded dragons during exposure to different oxygen concentrations. Values are means of 12 lizards during a 7.5 h period. *Significant differences relative to 21%oxygen, with the Holm–Sidak post-hoc test(P<0.05).

Movement between the cold and hot side of the extreme temperature shuttle box was significantly affected by hypoxia (t11=2.47, P=0.031). The shuttling frequency over the 7.5 h recording period fell from 77.7±68.9 shuttles in normoxia to 23.7±25.1 shuttles at 4% O2 (Fig. 4). Similarly, in the thermal gradient trials, movement (assessed as the number of times in 7.5 h that the animals moved more than 20 cm min–1)fell significantly from 76.8±66.5 in normoxia to 25.2±25.6 at 4%O2 (t10=3.24, P=0.027). Interestingly,in the extreme trials, lizards spent a considerable amount of time straddling the transition zone between the cold and warm chambers, something not observed in the ramping shuttle box trials. Moreover, at 4% oxygen, lizards spent significantly more time (F1,9=13.10, P=0.018)inactive in the middle section between the two choice chambers(49.4±36.9% of the time compared with 17.2±23.6% in normoxia; Table 3).

Table 3.

Comparison of time spent in the various compartments within the shuttle box in normoxia and hypoxia

TreatmentO2 (%)Cold (%)Warm (%)Middle (%)
Ramping shuttle box 21 48.7±1.7 51.3±1.7
49.1±4.2 50.9±4.2
Extreme temperatures shuttle box 21 42.9±13.8 39.9±13.5 17.2±23.6
29.9±25.4 20.7±16.0 49.4±36.9
TreatmentO2 (%)Cold (%)Warm (%)Middle (%)
Ramping shuttle box 21 48.7±1.7 51.3±1.7
49.1±4.2 50.9±4.2
Extreme temperatures shuttle box 21 42.9±13.8 39.9±13.5 17.2±23.6
29.9±25.4 20.7±16.0 49.4±36.9

Note, during the ramping trials lizards were never observed to sit in the middle between the two chambers. Significant effect of hypoxia (within a treatment) on the time spent in the middle between the two compartments (RM ANOVA, F1,9=3.6, P=0.006)

### Time course and Tb distributions: comparison of methodology

Low oxygen induced changes in selected Ta, and therefore Tb, within the first hour of exposure in all methodologies (Fig. 6). The decline in Tb in the ramping shuttle box protocol and thermal gradient was gradual, taking 2–3 h, after which it became more variable, whereas the extreme shuttle box induced a rapid (within 30 min)decline in Tb that was nearly sustained throughout the 7.5 h of measurement. The resulting Tb distribution patterns(Fig. 6) reveal the widening of the Tb distribution that occurred in both shuttle box experiments compared with the thermal gradient trials.

Fig. 4.

Number of shuttles at different oxygen concentrations. Plotted values are mean values of 12 lizards ±s.d. during a 7.5 h period inside an electronically controlled temperature-choice shuttle box. *Significant difference vs 21% oxygen conditions. For comparison, the filled circles show the average number of times a lizard shuttled during the 7.5 h in the absence of a thermal stimulus (constant 34.5°C at 21% O2) (Cadena and Tattersall, 2009). The open square denotes the average number of shuttles (error bars not shown; see text for details) in the extreme temperatures shuttle box trials at 21% and 4% O2 (offset slightly for clarity).

Fig. 4.

Number of shuttles at different oxygen concentrations. Plotted values are mean values of 12 lizards ±s.d. during a 7.5 h period inside an electronically controlled temperature-choice shuttle box. *Significant difference vs 21% oxygen conditions. For comparison, the filled circles show the average number of times a lizard shuttled during the 7.5 h in the absence of a thermal stimulus (constant 34.5°C at 21% O2) (Cadena and Tattersall, 2009). The open square denotes the average number of shuttles (error bars not shown; see text for details) in the extreme temperatures shuttle box trials at 21% and 4% O2 (offset slightly for clarity).

### Behavioural thermoregulation in hypoxia

This study shows a proportional effect of hypoxia on the level and, in situations where frequent locomotion is required, the precision of behavioural thermoregulation in the bearded dragon (P. vitticeps). The cost–benefit model for thermoregulation in lizards proposed by Huey and Slatkin (Huey and Slatkin,1976) predicts that lizards will actively thermoregulate if the associated costs are low. Although the model was intended to account for costs of an ecological nature (i.e. food availability, homogeneity of the thermal environment, accessibility of basking sites, etc.), it should also be relevant to situations where metabolic or physiological costs are associated with thermoregulation. In severe hypoxia, oxygen becomes a limited resource and locomotion becomes impossible to conduct aerobically. In addition, if frequent movements are necessary to maintain a narrow Tb range,lizards are expected to minimise locomotion at the expense of precise behavioural thermoregulation.

Fig. 5.

Effect of methodology on Tb of lizards when exposed to hypoxia. The preferred' Tb of nine bearded dragons was determined at 4% and 21% oxygen in an extreme temperatures shuttle box, a thermal gradient and an electronically operated temperature shuttle box(ramping shuttle box). Values are means of nine lizards' median Tb values ±s.d.*Significant difference between the extreme temperature trials and the other trials. Significant effect of 4% O2 on Tb within a trial.

Fig. 5.

Effect of methodology on Tb of lizards when exposed to hypoxia. The preferred' Tb of nine bearded dragons was determined at 4% and 21% oxygen in an extreme temperatures shuttle box, a thermal gradient and an electronically operated temperature shuttle box(ramping shuttle box). Values are means of nine lizards' median Tb values ±s.d.*Significant difference between the extreme temperature trials and the other trials. Significant effect of 4% O2 on Tb within a trial.

In the present study, bearded dragons did not abandon temperature regulation at low oxygen conditions, as might be expected if the response was an unregulated (i.e. hypothermic) decrease in Tb. Instead,they reduced locomotory oxygen expenditure by reducing the frequency of shuttles between compartments in the shuttle box(Fig. 4) and overall movement in the thermal gradient. In the case of the ramping shuttle box experiments,this led to a decrease in the precision of thermoregulation, manifested in the widening of the Tb range. The fact that the frequency of shuttles at the most extreme level of hypoxia (∼21 shuttles) is still well above the predicted number of shuttles (∼8) if the behaviour was purely exploratory (Cadena and Tattersall,2009) provides further evidence that hypoxic lizards are indeed still actively thermoregulating. Surprisingly, we did not observe a decrease in thermoregulatory precision in the extreme temperatures shuttle box, where locomotory requirements are expected to be largest. However, animals in this protocol were able to compensate by remaining for longer periods of time straddling the cold and the warm compartments, thus decreasing the amount of locomotion required to maintain a constant Tb (albeit lower, compared with the other methodologies; Fig. 5; Table 2), and avoiding theextreme temperatures' that would have otherwise resulted in a decrease in thermoregulatory precision.

Fig. 6.

Influence of methodology on the normoxic (black lines) and hypoxic (grey lines) thermoregulatory responses of bearded dragons. A, B and C show the average change in Tb over the 7.5h of measurements(starting at 0h) in the ramping shuttle box, extreme shuttle box and thermal gradient trials, respectively. D, E and F show the corresponding average frequency histograms (expressed as per cent of observations) for lizards in the ramping shuttle box, extreme shuttle box and thermal gradient trials,respectively.

Fig. 6.

Influence of methodology on the normoxic (black lines) and hypoxic (grey lines) thermoregulatory responses of bearded dragons. A, B and C show the average change in Tb over the 7.5h of measurements(starting at 0h) in the ramping shuttle box, extreme shuttle box and thermal gradient trials, respectively. D, E and F show the corresponding average frequency histograms (expressed as per cent of observations) for lizards in the ramping shuttle box, extreme shuttle box and thermal gradient trials,respectively.

The effect of hypoxia on lizard thermoregulation was previously described by Hicks and Wood who showed significant decreases in the preferred Tb of four different species of lizard exposed to 7%oxygen (Hicks and Wood, 1985). Similar decreases in Tb have been observed in situations such as anaemia and exhaustive exercise, both conditions in which the oxygen content of the blood is low (Hicks and Wood, 1985; Petersen et al.,2003), suggesting similar underlying mechanisms to the thermoregulatory response. The response observed by Hicks and Wood, however,was much more pronounced (Hicks and Wood,1985) than the one observed here (a decrease of 5–7°C vs the ∼2–4°C decrease observed by us with exposure to 4% O2). Although this difference may be attributed to differences between species, it might also be due to methodological differences between the studies. Hicks and Wood used a thermal gradient and an extreme temperatures shuttle box to estimate the preferred Tb of lizards (Hicks and Wood,1985). Even though thermal gradients require less locomotoryeffort' from the animal, it is hard to discriminate between thermoregulating lizards and lizards that are not actively thermoregulating but remaining stationary, due to lack of knowledge of an animal's motivational state. As we and others (Schurmann and Steffensen,1994) have observed, ectotherms may not be motivated to actively thermoregulate in every experiment, and the inclusion of these data may lead to a lower (usually) estimation of the preferred Tb(Cadena and Tattersall, 2009). An extreme temperatures shuttle box, on the other hand, is an energetically costly condition since it requires frequent and constant shuttling between compartments to maintain relatively high and constant Tblevels and can therefore cause decreases in normal Tbitself (Fig. 5)(Cadena and Tattersall, 2009). Our results indicate that the use of a dual-choice shuttle box can inflate the effect of hypoxia on Tb by twofold compared with other techniques.

The decrease in Tb of hypoxic lizards observed by Hicks and Wood was accompanied by a decrease in both UETb and LETb [when thermoregulating inside an extreme temperatures shuttle box (Hicks and Wood, 1985)],suggesting a decline in the upper and lower temperature set-points(Barber and Crawford, 1977). Escape Tb and Tb set-points, however,are not necessarily analogous, since the thermal inertia of even small animals is usually large enough to cause changes in Tb to lag behind those of Ta, which is why we report different variables in our study (see Table 1). Interestingly, however, the upper escape Ta of bearded dragons in our study rose slightly (but not significantly), while the lower escape Ta decreased at lower oxygen concentrations. In other words, it was the drastic decline of the lower escape Ta and not a decrease in both the upper and lower temperature set-points that caused the general decrease in Tb. These findings suggest that in conditions in which continuous locomotion is required, hypoxia induces changes in behavioural thermoregulatory precision, a fact hitherto unappreciated.

### Implications for the neurophysiological control of Tb

Similar thermoregulatory responses to those reported here have been observed in experiments in which the medial preoptic area of the brain of the lizard Dipsosaurus dorsalis was lesioned(Berk and Heath, 1975b). When allowed to thermoregulate inside a thermal shuttle box, lesioned lizards decreased their lower escape temperature and increased their upper escape temperature while significantly reducing their shuttling frequency compared with unoperated control groups. Thus, interference with the thermoregulatory integrative centres of the brain will decrease the precision of thermoregulation at both upper and lower ranges. Interestingly, hypoxia has also been shown to significantly decrease the cutaneous sensitivity of humans to cold but not to warm temperatures(Golja et al., 2004). This is consistent with the observation that rhesus monkeys exposed briefly to anoxia drastically diminished the firing rate of cutaneous cold-sensitive units(Iggo and Paintal, 1977). The firing rate of thermosensitive neurons in the preoptic area of rats has also been shown to change in hypoxia (Tamaki and Nakayama, 1987), although, generally, the main effect is to demonstrate an increase in firing rate at low oxygen levels. It is possible,therefore, that hypoxia causes a decrease in the sensitivity of skin temperature receptors, thereby diminishing peripheral feedback, as well as altering the preoptic thermosensitive neurons responsible for thermoregulatory control in lizards. The tendency for bearded dragons to select lower Ta and the greater variability in their lower escape temperatures in low oxygen is consistent with a greater influence of hypoxia on the cold-sensitive side of thermoregulation. These changes in sensitivity would be manifested in a decrease in locomotory behaviour, as lizards would tolerate higher than and lower than normal temperatures and remain stationary for longer periods of time at Ta values outside their normally preferred range. This is consistent with the observations in this study. Although we do not exclude the possibility that the decrease in Tb observed in hypoxia is a regulated response (i.e. decrease in set-point), the present study reveals another component and potential mechanism of the thermoregulatory response to hypoxia; a decrease in thermoregulatory precision, possibly due to a hypoxia-induced decrease in thermosensitivity.

### Conclusions

The evidence presented here provides further information that, in addition to the effect on the level of thermoregulation, the temporal patterns and variance in selected Tb are influenced by the methodology of assessing thermal preference (Fig. 6). These differences are further increased by hypoxia. The integrated behavioural thermoregulatory response to hypoxia could be viewed as resulting from a number of factors. Firstly, severe hypoxia has an impact on the capacity for aerobic activity. This decreases movement and thereby causes decreased precision in behavioural thermoregulation requiring locomotory efforts. Indeed, we observed a decline in the locomotory behaviour of lizards in hypoxia under all experimental conditions, suggesting a reduction in activity. Secondly, this behavioural response could be manifested through differential effects on the cold- and warm-sensing pathways that influence thermal sensation and effect the changes that accompany behavioural thermoregulation. This is consistent with data from studies of thermoregulation in mammals which also show that the precision of Tb control in hypoxia may be lower than that observed in normoxia (Barros et al., 2001; Dupré et al., 1988; Gordon and Fogelson, 1991; Tattersall and Milsom,2002).

Other stress stimuli (that in one way or another increase the costs of thermoregulation), such as low environmental thermal quality, risk of predation, territorial defence, or water or food availability have been shown to evoke similar decreases in Tb and/or thermoregulatory precision (Cabanac, 1985; Dewitt, 1967; Huey and Slatkin, 1976; Ladyman and Bradshaw, 2003; Lorenzon et al., 1999; Mathies and Andrews, 1997),suggesting conserved thermoregulatory mechanisms in response to costly conditions. Discovering how these stressors alter the sensation of temperature and the underlying neurophysiological control of Tbremains a challenge for future studies.

We would like to thank Joshua Shaw and Dimitri Skandalis who assisted in data collection, and acknowledge the high quality of animal care provided by Tom Eles. We would also like to thank two anonymous referees for their valuable comments on an earlier version of this article. This project was funded by the Canadian Foundation for Innovation, Premier's Research Excellence Award, and an NSERC operating grant to G.J.T. (grant no. 262-087-2003).

Barber, B. J. and Crawford, E. C. (
1977
). A stochastic dual-limit hypothesis for behavioral thermoregulation in lizards.
Physiol. Zool.
50
,
53
-60.
Barros, R. C., Zimmer, M. E., Branco, L. G. and Milsom, W. K. (
2001
). Hypoxic metabolic response of the golden-mantled ground squirrel.
J. Appl. Physiol.
91
,
603
-612.
Berk, M. L. and Heath, J. E. (
1975a
). An analysis of behavioral thermoregulation in the lizard, Dipsosaurus dorsalis.
J. Therm. Biol.
1
,
15
-22.
Berk, M. L. and Heath, J. E. (
1975b
). Effects of preoptic, hypothalamic, and telencephalic lesions on thermoregulation in lizard, Dipsosaurus dorsalis.
J. Therm. Biol.
1
,
65
-78.
Bicego, K. C., Barros, R. C. H. and Branco, L. G. S.(
2007
). Physiology of temperature regulation: comparative aspects.
Comp. Biochem. Physiol. A, Mol. Integr. Physiol.
147
,
616
-639.
Bickler, P. E. and Buck, L. T. (
2007
). Hypoxia tolerance in reptiles, amphibians, and fishes: life with variable oxygen availability.
Annu. Rev. Physiol.
69
,
145
-170.
Blumberg, M. S., Lewis, S. J. and Sokoloff, G.(
2002
). Incubation temperature modulates post-hatching thermoregulatory behavior in the Madagascar ground gecko, Paroedura pictus.
J. Exp. Biol.
205
,
2777
-2784.
Cabanac, M. (
1985
). Strategies adopted by juvenile lizards foraging in a cold environment.
Physiol. Zool.
58
,
262
-271.
Cadena, V. and Tattersall, G. J. (
2009
). The effect of thermal quality on the thermoregulatory behaviour of the bearded dragon, Pogona vitticeps: influences of methodological assessment.
Physiol. Biochem. Zool.
82
(in press).
de Velasco, J. B. and Tattersall, G. J. (
2008
). The influence of hypoxia on the thermal sensitivity of skin colouration in the bearded dragon, Pogona vitticeps.
J. Comp. Physiol. B, Biochem. Syst. Environ. Physiol.
178
,
867
Dewitt, C. B. (
1967
). Precision of thermoregulation and its relation to environmental factors in desert iguana Dipsosaurus Dorsalis.
Physiol. Zool.
40
,
49
-66.
Dewitt, C. B. and Friedman, R. M. (
1979
). Significance of skewness in ectotherm thermoregulation.
Am. Zool.
19
,
195
-209.
Dupré, R. K., Romero, A. M. and Wood, S. C.(
1988
). Thermoregulation and metabolism in hypoxic animals. In
Oxygen Transfer from Atmosphere to Tissues
(ed. N. C. Gonzalez and M. R. Fedde), pp.
347
-351. New York:Plenum Press.
Golja, P., Kacin, A., Tipton, M. J., Eiken, O. and Mekjavic, I. B. (
2004
). Hypoxia increases the cutaneous threshold for the sensation of cold.
Eur. J. Appl. Physiol.
92
,
62
-68.
Gordon, C. J. and Fogelson, L. (
1991
). Comparative effects of hypoxia on behavioral thermoregulation in rats,hamsters, and mice.
Am. J. Physiol.
260
,
R120
-R125.
Hicks, J. W. and Wang, T. (
2004
). Hypometabolism in reptiles: behavioural and physiological mechanisms that reduce aerobic demands.
Respir. Physiol. Neurobiol.
141
,
261
-271.
Hicks, J. W. and Wood, S. C. (
1985
). Temperature regulation in lizards: effects of hypoxia.
Am. J. Physiol.
248
,
R595
-R600.
Holm, S. (
1979
). A simple sequentially rejective multiple test procedure.
Scand. Stat. Theory Appl.
6
,
65
-70.
Huey, R. B. and Slatkin, M. (
1976
). Cost and benefits of lizard thermoregulation.
Q. Rev. Biol.
51
,
363
-384.
Iggo, A. and Paintal, A. S. (
1977
). Metabolic dependence of primate cutaneous cold receptors.
J. Physiol.(Lond.)
272
,
P40
-P41.
Kozak, W. (
1997
). Regulated decreases in body temperature. In
Fever: Basic Mechanisms and Management
, vol.
36
(ed. P. A. Mackowiak),pp.
467
2003
). The influence of dehydration on the thermal preferences of the Western tiger snake, Notechis scutatus.
J. Comp. Physiol. B, Biochem. Syst. Environ. Physiol.
173
,
239
-246.
Lorenzon, P., Clobert, J., Oppliger, A. and John-Alder, H.(
1999
). Effect of water constraint on growth rate, activity and body temperature of yearling common lizard (Lacerta vivipara).
Oecologia
118
,
423
-430.
Mathies, T. and Andrews, R. M. (
1997
). Influence of pregnancy on the thermal biology of the lizard, Sceloporus jarrovi: why do pregnant females exhibit low body temperatures?
Funct. Ecol.
11
,
498
-507.
McKemy, D. D. (
2007
). Temperature sensing across species.
Pflugers Arch.
454
,
777
-791.
Myhre, K. and Hammel, H. T. (
1969
). Behavioral regulation of internal temperature in the lizard Tiliqua scincoides.
Am. J. Physiol.
217
,
1490
-1495.
Petersen, A. M., Gleeson, T. T. and Scholnick, D. A.(
2003
). The effect of oxygen and adenosine on lizard thermoregulation.
Physiol. Biochem. Zool.
76
,
339
-347.
Rollinson, N., Tattersall, G. J. and Brooks, R. J.(
2008
). Overwintering habitats of a northern population of Painted Turtles (Chrysemys picta): winter temperature selection and dissolved oxygen concentrations.
J. Herpetol.
42
,
312
-321.
Saint-Girons, M. C. (
1970
). Morphology of the circulating blood cells. In
Biology of the Reptilia
,vol.
3
(ed. C. Gans and T. Parson), pp.
73
Sayeed, O. and Benzer, S. (
1996
). Behavioral genetics of thermosensation and hygrosensation in Drosophila.
93
,
6079
-6084.
Schall, J. J., Bennett, A. F. and Putnam, R. W.(
1982
). Lizards infected with malaria-physiological and behavioral consequences.
Science
217
,
1057
-1059.
Schurmann, H. and Steffensen, J. (
1994
). Spontaneous swimming activity of Atlantic cod Gadus morhua exposed to graded hypoxia at three temperatures.
J. Exp. Biol.
197
,
129
-142.
Tamaki, Y. and Nakayama, T. (
1987
). Effects of air constituents on thermosensitivities of preoptic neurons-hypoxia versus hypercapnia.
Pflugers Arch.
409
,
1
-6.
Tattersall, G. J. and Gerlach, R. M. (
2005
). Hypoxia progressively lowers thermal gaping thresholds in bearded dragons, Pogona vitticeps.
J. Exp. Biol.
208
,
3321
-3330.
Tattersall, G. J. and Milsom, W. K. (
2002
). Central thermoregulation and the hypoxic metabolic response.
FASEB J.
16
,
A45
-A46.
Withers, P. C. and Campbell, J. D. (
1985
). Effects of environmental cost on thermoregulation in the Desert Iguana.
Physiol. Zool.
58
,
329
-339.
Wood, S. C. (
1984
). Cardiovascular shunts and oxygen transport in lower vertebrates.
Am. J. Physiol.
247
,
R3
-R14.
Wood, S. C. (
1995
). Interrelationships between hypoxia and thermoregulation in vertebrates. In
Advances in Comparative and Environmental Physiology
, vol.
22
, pp.
209
-231. Berlin:Springer-Verlag.
Wood, S. C. and Glass, M. L. (
1991
). Respiration and thermoregulation of amphibians and reptiles. In
Physiological Strategies for Gas Exchange and Metabolism
(ed. A. J. Woakes, M. K. Grieshaber and C. R. Bridges), pp.
107
-124. Cambridge: Cambridge University Press.
Wood, S. C. and Gonzales, R. (
1996
). Hypothermia in hypoxic animals: mechanisms, mediators, and functional significance.
Comp. Biochem. Physiol. B, Biochem. Mol. Biol.
113
,
37
-43.